Objective, automated approximation of landslide source zones from landslide point inventories
Abstract
Characterization of landslide source zones after a widespread event poses a daunting task. Although field work can provide measurements of source-zone width, length, slope, and qualitative estimates of curvature, the effort is typically limited by time constraints, ease of access, vegetation regrowth, and post-event disturbance. Aerial and satellite imagery allows mapping of landslide polygons (including transport and deposition zones) or headscarp points, but source zones are difficult to map because they often look identical to downslope travel paths. Neither field work or mapping from imagery directly provides 1) the statistical distribution of curvature and slope from pre-event topography or 2) source-zone polygons for comparison with source-zone predictions from DEM-based susceptibility models.
We developed an objective, automated, rapid method to approximate source zone as the area downslope of the headscarp constrained by source-zone length. Our method utilizes tools available in TauDEM (Tarboton, 2015) designed for avalanche runout. The mandatory inputs are headscarp point locations, an average source-zone length and headscarp width. When available, landslide polygons can provide supplemental spatial constraints. We applied our method to Puerto Rico, where Hurricane Maria induced more than 40,000 landslides. Using mapped polygons of landslide-affected areas, associated point locations at the upslope center of headscarps, and field estimates of source-zone length, we extracted slope and curvature statistics from a DEM in the area directly downslope from each headscarp. The resulting approximations of source zone areas provided the statistical distribution of slope and curvature values for all mapped landslides and allowed for more accurate ROC analysis of source zone susceptibility models, when compared to analysis using landslide polygons which included run-out zones. Our method is applicable to any landslide inventory where headscarp point locations have been mapped and is the foundation for a new landslide analysis tool that can save time in initial investigation of landslide characteristics including slope, curvature, geology, or other raster data sets related to landslide susceptibility.- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2019
- Bibcode:
- 2019AGUFMNH33D0936B
- Keywords:
-
- 1810 Debris flow and landslides;
- HYDROLOGY;
- 1826 Geomorphology: hillslope;
- HYDROLOGY;
- 4306 Multihazards;
- NATURAL HAZARDS;
- 7212 Earthquake ground motions and engineering seismology;
- SEISMOLOGY