Landslides pose a major natural hazard, and heterogeneous conditions and limited data availability in the field make it difficult to connect mapped landslide inventories to the underlying mass-failure mechanics. To test and build predictive links between landslide observations and mechanics, we monitored 67.89 h of physical experiments in which an incising and laterally migrating river generated landslides by undercutting banks of moist sand. Using overhead photos (every 20 s) and 1-mm-resolution laser topographic scans (every 15-30 min), we quantified the area, width, length, depth, volume, and time of every visible landslide, as well as the scarp angles for those within 3 min prior to a topographic scan. Both the landslide area-frequency distribution and area-volume relationship are consistent with those from field data. Cohesive strength controlled the peak in landslide area-frequency distribution. These results provide experimental support for inverting landslide inventories to recover the mechanical properties of hillslopes, which can then be used to improve hazard predictions.
Bibliographical noteFunding Information:
Stefanie Tofelde designed the experimental basin with intellectual input from Chris Paola, Jean-Louis Grimaud, and Sara Savi, and constructed it with the support of Ben Erickson, Richard Christopher, Chris Ellis, Jim Mullin, Eric Steen, and Sara Savi. Eric Steen also provided valuable assistance with the laser scanner. We heartily thank two anonymous reviewers for their constructive comments; these significantly improved the form and rigor of the final article. Start-up funds awarded to Wickert by the University of Minnesota supported Beaulieu and the construction of the experimental basin. Beaulieu also received support from the Richard C. Dennis Fellowship awarded by the Department of Earth Sciences at the University of Minnesota. This material is based upon work supported by the National Science Foundation under grant EAR-1944782.
© 2021 The Authors. Gold Open Access. All Rights Reserved.