GSA Annual Meeting in Phoenix, Arizona, USA - 2019

Paper No. 162-7
Presentation Time: 9:50 AM

RECORDING RAPID LANDSLIDE MOTION IN THREE DIMENSIONS USING LOW-COST, MULTIVIEW VIDEO


RAPSTINE, Thomas1, RENGERS, Francis K.2, ALLSTADT, Kate E.2, SMITH, Joel B.2, OBRYK, Maciej K.2, LOGAN, Matthew2, IVERSON, Richard M.3 and OLSEN, Michael J.4, (1)Geophysics, Colorado School of Mines, Golden, CO 80401, (2)U.S. Geological Survey, Box 25046, MS 966, Denver Federal Center, Denver, CO 80225, (3)United States Geological Survey, Cascades Volcano Observatory, Vancouver, WA 98683, (4)School of Civil and Construction Engineering, Oregon State University, 220 Owen Hall, Corvallis, OR 97331

Landslides are destructive hazards that result in scores of fatalities and cause nearly $3 billion in damage in the U.S. annually. To facilitate efforts to mitigate landslides, we require a better understanding of the mechanics of these phenomena during failure; however, the destructive nature of landslides makes conducting measurements difficult. Landslide mechanics can be constrained by high-resolution topography (HRT) analyses using spatially distributed non-contact measurements of ground surface deformation. Existing technologies, such as InSAR and airborne lidar, provide such measurements of change detection for slowly moving landslides, but lack the temporal resolution to measure rapid landslide motion. Here we present and test a new method for HRT by recording high-speed changes in topography using sequences of pseudo-stereo images from inexpensive, stationary video cameras. We apply the method to videos obtained during an experiment at the USGS debris-flow flume during which a high-speed, liquefying landslide was induced by gradually adding water to a 6 m3 prism of loosely packed sediment on a 31 degree slope. We constructed 30 digital elevation models per second from videos during this experimental landslide. This resulted in 300 DEMs with a resolution of ~3 cm. Concurrent lidar measurements of deformation agree with our low-cost video-derived measurements. The mean elevation difference between the lidar and video topography data is ~0.5 cm during creeping movement prior to failure and ~3.5 cm during rapid movement. Rapid topographic change detection from video-derived digital elevation models is therefore a straightforward and inexpensive method for characterizing fundamental landslide initiation processes. These techniques may prove important for mitigating future hazards from rapidly-moving landslides.