Effective friction coefficient and pore pressure along the San Andreas Fault from seismicity rate change after the 2004 Parkfield earthquake

There is growing evidence that the stress changes due to a co-seismic slip can inhibit or promote failure nearby fault. However in many case, including the Landers, Northridge or Chi-Chi earthquakes, the coherence between the observed seismicity rate change and the calculated Coulomb stress change is low (<65%). This low coherence may be due to the use of an inappropriate static stress approach or to the low resolution of the slip distribution. Here we propose an alternative explanation by assessing the spatial variation of failure planes orientation, friction (μ) and Skempton's coefficients (B), which give the greatest correspondence between stress changes and aftershocks locations. Assuming a well-resolved co-seismic slip distribution, we use a 3-D simulated annealing method to obtain the spatial distribution of these parameters from seismicity rate change. Synthetic tests suggest that non-uniform fault plane orientation, μ and B improve by ~10-20 % the coherence between seismicity rate change and Coulomb stress change. Our study demonstrates that fault plane parameters and coefficients (μ and B) can be obtained from seismicity rate change with and accuracy of 20° and 0.2, respectively and it reveals for strike -slip faults a clear trade-off between azimuth and friction coefficient. In the Parkfield earthquake area, calculated failure plane orientations are in good agreement with both mapped fault and focal mechanism. Our study suggests that a low effective friction coefficient (<0.4) is required to explain the observed seismicity rate change. Our results give lateral variations of the pore pressure along the San Andreas Fault, which may be associated with fluid circulation after the Parkfield rupture.