Temporal and spatial correlations in earthquake dynamics: physical modeling and data analysis
In seismology, aftershocks are smaller earthquakes originating from the destabilised crust in the regions close to a larger earthquake, the mainshock. In this thesis we study this phenomenon from three distinct perspectives. In the first one we unveil the connections between simple models of aftershocks occurrence and record statistics of stochastic processes. The second approach builds up on the similarities between earthquake sequences and mainshock-aftershocks groups with the creep dynamics of a domain wall expanding in a thin magnetic film. Finally, we use machine learning tools to construct a forecasting method for aftershock occurrence by exploiting measurements of surface deformation coming from Global Positioning System (GPS) satellites. As an addendum, in the last part of the thesis, we discuss the physics of yield-stress fluids in porous media and its links to the physics of disorder systems and directed polymers in random media.
Jury : Aurélien Decelle, François Landes (co-directeur de thèse), Cécile Monthus, Alberto Rosso (directeur de thèse), Herbert Spohn (rapporteur), Jérôme Weiss (rapporteur)