Automated segmentation of hillslope profiles across ridges and valleys using a digital elevation model

We developed a three-step algorithm to automate the recognition and segmentation of hillslopes across ridges and valleys using a gridded digital elevation model (DEM). The first step involves the delineation of a side-slope profile by locating the lower and upper boundaries. In the second step, the cross-profile pattern of a ridge or a valley floor, which is defined and bounded by a pair of side slopes, is examined in order to distinguish upland flats from ridges and lowlands from valley floors. The width and gradient of these landform units are also measured. In the third step, a side-slope profile is segmented into lower, mid-, and upper slopes, with a line-simplification algorithm to identify changes and breaks in the slope. In this step, the footslope and crestslope can also be distinguished within the lower and upper slopes, respectively. Using a 15-m DEM, this method was applied to five sites in low- to high-relief fluvial landscapes in Japan. The results showed good agreement (86.9 ± 2.7%) with manual slope segmentation, indicating the effectiveness of our algorithm.

► We developed an algorithm to automate hillslope-profile segmentation using a DEM. ► The algorithm was examined in fluvial landscapes of Japan. ► The segmentation results agreed well with the manual interpretation of contours.