Broadband analysis of landslides seismic signal : example of the Oso-Steelhead landslide and other recent events

Abstract:

Landslide failures on the scale of mountains are spectacular, dangerous, and spontaneous, making direct observations hard to obtain. Measurement of their dynamic properties during runout is a high research priority, but a logistical and technical challenge. Seismology has begun to help in several important ways. Taking advantage of broadband seismic stations, recent advances now allow: (i) the seismic detection and location of large landslides in near-real-time, even for events in very remote areas that may have remain undetected, such as the 2014 Mt La Perouse supraglacial failure in Alaska; (ii) inversion of long-period waves generated by large landslides to yield an estimate of the forces imparted by the bulk accelerating mass; (iii) inference of the landslide mass, its center-of-mass velocity over time, and its trajectory.

Key questions persist, such as: What can the short-period seismic data tell us about the high-frequency impacts taking place within the granular flow and along its boundaries with the underlying bedrock? And how does this seismicity relate to the bulk acceleration of the landslide and the long-period seismicity generated by it?

Our recent work on the joint analysis of short- and long-period seismic signals generated by past and recent events, such as the Bingham Canyon Mine and the Oso-Steelhead landslides, provides new insights to tackle these issues. Qualitative comparison between short-period signal features and kinematic parameters inferred from long-period surface wave inversion helps to refine interpretation of the source dynamics and to understand the different mechanisms for the origin of the short-period wave radiation. Our new results also suggest that quantitative relationships can be derived from this joint analysis, in particular between the short-period seismic signal envelope and the inferred momentum of the center-of-mass. In the future, these quantitative relationships may help to constrain and calibrate parameters used in inversion or simulation of long-period waves generated by landslides. Relating the center-of-mass dynamics to the short-period seismic signal might also yield a new means to estimate kinematic parameters for the smaller events that generate too weak long-period seismic waves to allow inversion or simulation of the seismic source.