Constraining earthquake sources by seismic time reversal

Earthquake rupture is a complex phenomenon of which we understand comparatively little. In mapping the rupture of a seismic event in both space and time, different techniques and datasets have been used, which often led to incoherent results for the same event. Most of these methods rely on seismic recordings neglecting the information carried by the surface waves, and focus on the arrivals of P- and S-waves. The central idea of our project is to implement surface-wave time reversal, to study the focusing of the time-reversed field at the source location and to better constrain the details of rupture processes at the source of seismic events. Our method combines the seismic time reversal approach with a ray-tracing algorithm, relying on the generalized harmonic parameterization to trace surface-wave ray paths in the presence of laterally varying azimuthal anisotropy. We validated our time-reversal method, and quantified its limitations, through a number of synthetic tests at the global scale. In our experiments, a prominent maximum of the time-reversed wave filed is systematically obtained at or very close to the original location and time of the source. The uncertainties in the original source location and time are governed by the distribution of stations, and velocity model used. We next applied our method to narrow-frequency-band-filtered surface-wave data from the great 26 December 2004 Sumatra-Andaman earthquake. We reproduce the results of earlier studies, including the reconstruction of the source location, direction of rupture propagation, its spatial extent, its duration, and identify the region where most seismic energy is released. Further, we applied our method to a volcanic setting, i.e., to recordings of very long period events that occurred in Mayotte, Comoro Islands. Our results are found to be in good agreement with the centroid locations obtained by moment tensor inversion. A precise location of this type of events helps in constraining the depth, size, and the geometry of the seismogenic volume, and hence to shed light on deep processes associated with volcanism.