Precursors and fluid flows in the case of the 1996, ML = 5.2 Saint-Paul-de-Fenouillet earthquake (Pyrenees, France): A complete pre-, co- and post-seismic scenario

Earthquake precursors are now regularly described but often detected only after a major or moderate seismic event. Presence and influence of fluids in the seismogenic processes are often observed at the time of earthquake studies. Even today, the understanding of the physical processes involved in the source region is a real challenge for seismic hazard assessment. Here, the aftershock sequence of the ML = 5.2, 1996 Saint-Paul-de-Fenouillet (Eastern Pyrenees, France) earthquake is first re-examined with P-wave cross-correlations, resulting in extracting three multiplets and in determining new locations. Multiplets and spatio-temporal distribution analysis of the aftershocks allow for quantifying the hydraulic diffusivity D at a maximum value of 5 m2/s and the permeability K at 10− 15 m2 in the upper Pyrenean crust. Second, a model is established in order to explain the hydrogeochemical transient anomalies, which occurred during the 15 day-period preceding the 1996 earthquake. These anomalies consist on a temporal and spatial sequence of gas emissions in the epicentral area and on chloride and lead concentration variations in a bottled mineral water 25 km north to the main shock epicenter. The proposed model processed in a standard elastic half-space, consists of creep on a low-angle crustal normal-fault, generating volumetric strain field changes over a distance of 25 km from the epicentral area. This model is able to constrain not only the mechanisms and the locations of the geochemical anomalies, but also their timing and probable casual links to the triggering of the impending major event. Also, the active extension proposed here is compatible with seismological observations in the Pyrenees. Thus, the possibility of such creep, which can be considered as a slow-slip event, is discussed in the Pyrenean tectonic and geological context. The model is discussed and compared to previous proposed models on precursor processes of earthquakes, especially concerning the preparation zone concept. Finally, a complete seismic scenario over the period beginning 15 days before the quake and ending 5 days after is proposed and discussed.