The former Provence coal basin, closed since 2003, has been monitored by a microseismic network since 2008. The objective is to detect the precursor signs of a brittle subsidence that would be caused by the collapse of the old underground mining works. Since the start of monitoring, no subsidence has affected the risk areas, and nearly 4000 events with a local magnitude between −3 < ML < 3 have been recorded. One sector in particular, called the Fuveau swarm, located outside the risk zones and therefore outside the brittle subsidence hazard zones, has attracted attention since 2012 because it was the subject of several seismic episodes felt in 2010, 2012, 2014, late 2016–early 2017, and August 2017. Since 2017, it has been established that the observed seismicity cannot be explained only by instability phenomena in the old mining works. The most likely hypothesis is that of the remobilization of faults hydraulically connected to the mining works, with seismic activity that is closely linked to variations in the groundwater level, which are themselves influenced by pumping and effective rainfall. This paper shows, through multiplet analysis method of the seismic data recorded by the monitoring network stations, that part of the seismicity in the monitoring areas is also due to the reactivation of tectonic faults. This conclusion is based on the concordance between the location of the multiplets and the orientation of the main faults mapped in the studied areas, as well as on the fact that the strongest events belong to these multiplets. This finding underscores the need to integrate fault reactivation into seismic monitoring strategies, beyond the current focus on mining-induced instabilities. This conclusion leads us to recommend revising the list of post-mining hazards, as post-mining seismic risk is often overlooked in many European regulations.
Recent advances in tomographic methods, such as Helmholtz/Eikonal tomography, leverage seismic wavefields recorded by dense regional seismic networks to produce finely resolved images of the subsurface. This approach directly uses phase and amplitude measurements of coherent wavefronts generated by large earthquakes, bypassing the need to solve large tomographic inverse problem or compute synthetic seismograms. However, since these earthquakes are mainly located at plate boundaries, the azimuthal distribution of the sources is often limited. In addition, obtaining robust phase velocity maps typically requires years of continuous seismic data. In contrast, oceanic microseisms - responsible for ambient seismic noise - are more evenly distributed and generate coherent wavefronts continuously. In this study, we apply a matched filtering technique to iteratively extract these coherent surface wavefronts. Within a 4-hour time window, we typically extract 10 to 20 coherent wavefronts, from which we can measure arrival time and amplitude at each station. After removing outliers, phase measurements are interpolated using smoothing splines on either a Cartesian or a spherical grid, depending on the size of the domain under study. The gradient of the interpolated phase velocity surfaces is then used in the eikonal equation to generate phase velocity maps. These maps are stacked to produce average isotropic phase velocity maps for periods ranging from 6 to 25 s. We will present applications of this method in California and Western Europe.