Geomorphological signatures: classification of aggregated slope unit objects from digital elevation and remote sensing data

The concept of a geomorphological signature is developed for classifying and mapping slope units with an automated procedure for analysing digital elevation and remote sensing data. Slope units are extracted from a digital elevation model (DEM) using a break of slope rule on downslope profiles. Each slope unit is an aggregated object of contiguous pixels and is summarized with five suites of variables: shape, topography, topographic variability, spectral characteristics, and variability in spectral characteristics. The variables are derived from the DEM and a corresponding SPOT satellite image. A ten-class scheme is used to classify slope units for a study area in southwest Yukon, Canada. Discriminant analysis results show the power of various combinations of variables to distinguish the classes, with a maximum classification accuracy of 90 per cent. Training signatures are employed for classifying the entire study area to produce a map with 88·5 per cent accuracy. The study shows that generating extensive geomorphological signatures for aggregated slope unit objects is a valuable exercise for discrimination and mapping. © 1998 John Wiley & Sons, Ltd.