Landslide dynamics from seismology: new results

Abstract:

We have recently shown how large landslides can be detected, located, and their dynamics measured in near-real-time through inversion of long-period surface waves. Here I will talk about our discovery, over several years, of multi-km-scale landslides in remote regions across the world, several of which remain unreported, and how we have developed insights into landslide physics using a large inventory of such observations.

Slope failures accelerating masses exceeding 10 megatons at 1-3m/s^2 generate forces large enough to trigger teleseismic waves detectable at broadband stations located 100s of kms distant. Inversion of such signals using a single-force model locates the epicenter to around +/-20km, and generates an approximate time series of the forces exerted by the evolving mass motion at that fixed point. This 3D vector sequence indicates the direction and magnitude of accelerations and thus momentum vector changes of the bulk landslide mass. Assuming constancy of this unknown mass during the main phase of motion allows us to infer a mass-scaled trajectory of motion of the mass center in 3D. Calibration against an observed runout distance leads to an estimate of then bulk mass. Conversely, use of an otherwise-estimated landslide mass (e.g., from differential topo mapping) leads to a calibration of the runout path without the need to observe it directly. In either case, comparison can be made against field and remote-sensing observations to check the validity of the method.

Here we present multiple such tests against landslides discovered by our group in Alaska, Tibet, the Andes, and more. I will summarize analysis of the dynamical properties of these slope failures and discuss their implications. Simple scaling is seen among a host of dynamical properties, including peak force, mass, momentum, peak kinetic energy, etc.

Roughly 5-15 events of this magnitude are seen globally each year. The majority strike in the periglacial environments of Alaska and Himalaya-Tibet-Karakoram. Most interesting are those that collapse onto glaciers and runout of many kms. Since momentum changes are weak during sliding over such low-angle, low-friction surfaces, we cannot invert for this phase of motion using long-period signals. The question remains whether the short-period seismicity has useful information to impart.