Automated detection and location of seismic events by continuous waveform migration, and application to Piton de la Fournaise volcano

Because of the ever increasing amount of seismological data available, automating the processing of such data has become crucial. For that purpose, we developed an algorithm, called Waveloc, whose principle relies on the migration of continuous waveforms. The kurtosis traces of the data are computed in order to enhance the first arriving phases, then migrated using a P-wave velocity model and stacked. Thus the coherent information is highlighted and we avoid the delicate association step common to many location algorithms. The method has been applied successfully to 12 seismic swarms recorded at Piton de la Fournaise volcano, La Réunion Island, France, between 2009 and 2011, and has shown consistent locations compared to manual ones. A more detailed analysis was carried out by clustering and double-difference relocation, as well as local magnitude computation. Final results highlight a size-dependent distribution of the events in a main seismic area just above sea-level.