INFERRING CONTROLS ON DEEP-SEATED LANDSLIDE MOTION WITH HIGH-RESOLUTION, THREE-DIMENSIONAL IMAGE CORRELATION OF REPEAT AIRBORNE LIDAR

Many deep-seated landslides move slowly or intermittently for centuries or longer, accumulating large amounts of strain without failing catastrophically. Several mechanistic theories have been proposed to regulate this slow landslide motion, but it remains challenging to make precise, high-resolution measurements of landslide deformation to test these models. Here, we use novel, 3D image correlation techniques on sequential airborne lidar data to monitor two large, slow-moving landslides in the western United States at 1 m spatial resolution, enabling us to infer controls on their spatiotemporal deformation. To derive the 3D displacement fields, we use the phase correlation technique to measure the relative offset of each cell in precisely aligned, lidar-derived digital elevation models with sub-pixel precision. At the Silt Creek landslide in the western Oregon Cascades, a debris flow in 2014 rapidly deposited 10 m of debris at the top of the landslide’s transport zone. This rapid loading initiated slip of the underlying deep-seated landslide, which propagated 1.7 km downslope to the landslide’s toe. Phase correlation of before and after airborne lidar spanning 15 months documented 13 m of total slip at the top of the transport zone, which decayed exponentially with distance to 0.5 m of slip at the toe. This observed displacement field required a transient increase in the shear strength at the failure plane to resist the additional driving stress caused by loading of the debris flow deposit. We infer that dilatancy-pore pressure feedbacks, enhanced by large scale basal roughness of the failure plane, contributed to stabilization. At the Mill Gulch landslide on the northern California coast, correlation of four lidar data sets spanning 12 years documented a spatially heterogeneous velocity field that changed dramatically over yearly time scales. The active shear margins and the inferred basal slip surface changed location from year to year, and we suggest that this abandonment followed by reestablishment of new shear zones in consolidated, saturated soil provided an additional mechanism for transient dilatant strengthening. Both sites illustrate that precise, 3D, high-resolution observations of landslide displacement fields provide a novel technique to test mechanistic models for controls on landslide motion.