Final Report Summary - SLIDEQUAKE (Detection and understanding of landslides by observing and modelling gravitational flows and generated earthquakes) The goal of the ERC project SLIDEQUAKES is to take a major step in improving the detection and understanding of landslides and their modelling at the field scale through the analysis of generated seismic waves. The seismic signal generated by landslides (i. e. landquakes) provides a unique tool to estimate the properties of the flow and its dynamics and mechanical behavior. Indeed, the force applied by the landslide to the ground, whose fluctuations generate seismic waves, is highly sensitive to the flow history and therefore to the physical properties during mass emplacement. The strategy is to combine a very accurate description of the landslide source, and the simulation and measurements of landquakes from the laboratory to the natural scale (e. g. La Réunion, Montserrat, Iceland, Dolomites), by leading an ambitious interdisciplinary project involving numerical modelling, laboratory experiments and observation. This is done in collaboration with the specialists of these domains.During this project, we have developed a series of new mathematical and numerical models of dry granular flows and two-phase flows (grains and fluid) of increasing complexity and associated increasing computational cost. We have in particular investigated the static/flowing interface in granular flows related in particular to erosion/deposition processes in non-averaged models in order to extract insight into its dynamics. Our result suggest that developing hydrostatic multilayer model is a better direction to include erosion/deposition processes with reasonable computational cost than using ad-hoc erosion laws in depth-averaged models. We have successfully compared the results of these models to a series of laboratory experiments. As an example, we have been able to simulate the strong effect of the initial volume fraction in two-phase flows related to the dilatancy effects that we now take into account in our models. These models are used to simulate the force applied by the landslide on the topography that generates seismic waves. By combining seismic inversion at low frequency (10-100s) and energy calculation at higher frequencies (> 1 Hz) together with numerical modelling we are now able to constrain the flow dynamics and recover key features of essentially dry landslides from the generated seismic waves (e. g. volume, friction coefficient, velocity, etc.). The application of our method to a large range of landslides show its wide applicability. However this work also showed that a major step forward in such direction would be achieved if more complex models including two-phase flow, erosion/deposition processes together with complex topography description would be used. Furthermore, extracting precise information from high frequency waves would require to use discrete element models or continuum models with granular temperature as well as to account for topography effects on seismic wave propagation. To get more insight into the energy partition during landslides we performed laboratory experiments on acoustic emissions of bead impacts, uniform granular flows and granular collapses, quantifying in particular the strong attenuation due to the presence of an erodible bed on the beads trajectory. Confrontation of these experimental results with field data and models of acoustic wave generation by granular flows open new doors to interpret the generated seismic waves. Finally, application of these work on more than 10-year time series of rockfalls on volcanoes showed that rockfall activity could be a proxy for the stability state of the volcanic slope and that rockfalls can be triggered by accumulation of seismicity with time delays of several days, for seismic wave amplitudes much smaller than what is generally observed.Our interdisciplinary group made it possible to get key results in very different domains published in journals of geophysics, physics, acoustics, mathematics, and numerical methods.
