One of the major challenges facing our society is to cope with the increase in natural hazards caused by climate change, human activity and population growth. The frequency of heavy rainfall and changes in vegetation cover are intensifying in most regions, greatly increasing the risk of landslides and the tsunamis they generate.

Exact prediction of the time, location and precise characteristics of a landslide is generally out of reach. However, it is possible to anticipate hazards by numerically simulating a series of probable scenarios using granular flow models on realistic topographies. However, there are two major obstacles to the use of these models. Firstly, the frictional behavior of these natural flows remains highly enigmatic, and most models fail to describe processes that play an important role at field scale, such as the interaction between grains and a fluid phase. On the other hand, very little data is available on the dynamics of these flows. In this context, the analysis of landslide-generated seismic waves coupled with the development of mathematical, physical and numerical models of granular flows on complex topography opens a unique opportunity to address this challenge.

I will present recent studies we have carried out with mathematicians and physicists to quantify landslide dynamics in order to assess the associated hazards. A key aspect of this work is to successfully couple state-of-the-art numerical simulation of these complex rheological flows with seismic wave analysis, moving back and forth between laboratory and field scales. I will describe the challenges posed in terms of modeling, such as dilatancy effects in a grain/fluid mixture, wave-flow interaction, and the relevance of multilayer Saint-Venant-type approaches.