A new method for determination of most likely landslide initiation points and the evaluation of digital terrain model scale in terrain stability mapping
- 1Department of Land and Agroforest Environments, AGRIPOLIS, University of Padova, 35020 Legnaro, Padova, Italy
- 2Department of Civil and Environmental Engineering, Utah State University, Logan, Utah, USA
Abstract. This paper introduces a new approach for determining the most likely initiation points for landslides from potential instability mapped using a terrain stability model. This approach identifies the location with critical stability index from a terrain stability model on each downslope path from ridge to valley. Any measure of terrain stability may be used with this approach, which here is illustrated using results from SINMAP, and from simply taking slope as an index of potential instability. The relative density of most likely landslide initiation points within and outside mapped landslide scars provides a way to evaluate the effectiveness of a terrain stability measure, even when mapped landslide scars include run out zones, rather than just initiation locations. This relative density was used to evaluate the utility of high resolution terrain data derived from airborne laser altimetry (LIDAR) for a small basin located in the Northeastern Region of Italy. Digital Terrain Models were derived from the LIDAR data for a range of grid cell sizes (from 2 to 50 m). We found appreciable differences between the density of most likely landslide initiation points within and outside mapped landslides with ratios as large as three or more with the highest ratios for a digital terrain model grid cell size of 10 m. This leads to two conclusions: (1) The relative density from a most likely landslide initiation point approach is useful for quantifying the effectiveness of a terrain stability map when mapped landslides do not or can not differentiate between initiation, runout, and depositional areas; and (2) in this study area, where landslides occurred in complexes that were sometimes more than 100 m wide, a digital terrain model scale of 10 m is optimal. Digital terrain model scales larger than 10 m result in loss of resolution that degrades the results, while for digital terrain model scales smaller than 10 m the physical processes responsible for triggering landslides are obscured by smaller scale terrain variability.