Regional Back-Analysis of Earthquake Triggered Landslide Inventories: A 2D Method for Estimating Rock Strength From Remote Sensing Data

Abstract

Landslides occur where the stresses below the surface exceed the shear strength of the material. Landslide inventories thus offer opportunities to investigate patterns in subsurface strength provided that the stress conditions at failure can be estimated. Clues to the failure stresses are encoded in the inclination of the slope that failed and the thickness of the sliding mass. We use this insight to develop a two-dimensional (2D) landslide back-analysis model that estimates bedrock strength over the broad scales relevant to earthquake-triggered landslide hazard and landscape evolution. A unique aspect of our model is the incorporation of independent landslide thickness measurements for each landslide, which are provided by differencing pre- and post-failure elevation data or estimated from a volume-area scaling relationship. This approach represents an innovation compared to previous regional-scale models that have assumed constant thickness or have used projections to estimate the depth to the failure plane, and it provides rock strength estimates as a function of depth below the surface. We evaluate our modeling approach in applications to two landslide inventories and compare the results against geotechnical field data. The back-calculated strength estimates are low for rock, which we hypothesize to reflect the contribution of weathering and fracturing, as well as the fact that landslides represent a small part of the entire study area and are likely associated with particularly weak material that is susceptible to failure. Finally, the two applications of our model indicate systematic variations in strength parameters below the surface and along an elevation profile, which we attribute to gradients in chemical and physical weathering.