Broadband seismic analysis and modeling of the 2015 Icy Bay landslide (Alaska).

In remote regions, many mass-wasting events would likely go unnoticed without networks of seismometers and satellite imagery. Mass-wasting events, such as landslides, generate seismic signals as the mobilized mass accelerates and decelerates, sliding along the ground. The combination of seismic monitoring and numerical modeling provides a way to detect, locate and infer the dynamics of such events, also in the absence of direct observations. We carry out a broadband analysis of the seismic signals generated by a massive landslide that occurred along the flanks of Icy Bay (Alaska) on October 17th 2015. This is the largest landslide in North America since the collapse of Mount St. Helens in 1980. The recorded broadband seismic signals are characterized by two distinct frequency bands, having slightly different onsets and an unusual clockwise particle motion. We determine the source time function by inverting the data (periods between 10 and 300 s) recorded at five close-in stations and we simulate the propagation of seismic waves between the source and the receivers using Instaseis, a recently developed tool for rapid synthetic-seismogram generation (Van Driel et al., 2015). From the force history, we deduce the runout trajectory and key parameters of the landslide dynamics associated with the center of mass, such as acceleration, velocity, displacement, kinetic, and potential energy.