Numerical modeling of the links between earthquakes and landsliding during the seismic cycle : triggering processes, size distribution, and geological implications.
Interactions between tectonic processes and erosion have been poorly investigated at short time-scales (<1000 years). However, earthquakes can largely contribute to the erosion of mountain belts by triggering widespread landsliding. Moreover, recent studies have shown that such large erosional events could induce stress changes in the fault environment efficient enough to influence regional seismicity. In this thesis, this problematic is tackled through a numerical approach. Firstly, the development of a simple mechanical model accounting for the complexity and variability of natural hillslopes allowed to demonstrate the role of mechanical parameters (cohesion and friction), and of hillslope shape in the probability density function of landslide sizes. This model has been validated using natural cases of co-seismic landsliding. Secondly, the role of unstable hillslope height on large landslide probability has been demonstrated based on natural data, and the exponential distribution of this unstable height has ben shown. Finally, the potential effect of a large erosional event on seismicity has been explored with a numerical model of seismic cycle, in which has been implemented temporal normal stress variations. The results emphasize the role of eroded sediment volume, but also of the export time of sediments away from the mountain belt. In landscape with high unstable hillslopes, large landslides are favored and in turn, could induce fast an important enough erosion to modify regional seismicity.