Evidence for coupled seismic and aseismic fault slip during water injection in the geothermal site of Soultz (France), and implications for seismogenic transients

We analysed the triggered seismicity recorded near 3 km in depth within granite during the 1993 water injection experiment at Soultz (France). We selected all large multiplets associated with the largest dislocated fault identified on the borehole logging (4.3 cm slip), and showed that each one consisted of the repeated rupture of a single asperity. These asperities were forced to rupture due to fault creep around them, so that their growing, cumulative slip history reveals the creep history on the fault, with a mean slip rate and a seismicity rate both decaying approximately as 1/t. This is consistent with a rate-strengthening friction law on the creeping faults, and differs from the interpretation of Omori type 1/t seismicity rates in terms of a weakening friction control of the unstable fault surfaces. This model of asperity rupture forced by relaxation creep is easily generalized in seismogenic regions, even for single ruptures of asperities on small creeping faults, showing that a significant part of aftershock rates can be controlled by the strengthening friction properties and the creep of faults at all scales. This lead us to introduce the concept of a critical asperity density, above which dynamic interaction between neighbouring asperities can initiate large seismic ruptures. At Soultz, the asperity density is subcritical, but the neighbouring asperities nonetheless interact, at distances up to a few source dimensions, as revealed by their delayed cross-triggering. Considering the seismic cloud in Soultz at a global scale, the strain produced by the injection experiment was mostly aseismic, related to creep on major faults, and causing large permeability changes - influencing in turn the pore pressure diffusion and the creeping process. Our observations demonstrate that the identification and analysis of multiplets can provide accurate images of the geometry and kinematics of transient slip patches, and indirectly detect pore pressure changes, which should contribute to a better understanding of the coupling between transient fluid flow, creep and microseismic activity in seismogenic regions.