Constraints on Friction, Dilatancy, Diffusivity, and Effective Stress From Low-Frequency Earthquake Rates on the Deep San Andreas Fault

Families of recurring low-frequency earthquakes (LFEs) within nonvolcanic tremor on the San Andreas Fault in central California are sensitive to tidal stresses. LFEs occur at all levels of the tides, are strongly correlated and in phase with the 200Pa shear stresses, and weakly and not systematically correlated with the 2kPa tidal normal stresses. We assume that LFEs are small sources that repeatedly fail during shear within a much larger scale, aseismically slipping fault zone and consider two different models of the fault slip: (1) modulation of the fault slip rate by the tidal stresses or (2) episodic slip, triggered by the tides. LFEs are strongly clustered with duration much shorter than the semidiurnal tide; they cannot be significantly modulated on that time scale. The recurrence times of clusters, however, are many times longer than the semidiurnal, leading to an appearance of tidal triggering. In this context we examine the predictions of laboratory-observed triggered frictional (dilatant) fault slip. The undrained end-member model produces no sensitivity to the tidal normal stress, and slip onsets are in phase with the tidal shear stress. The tidal correlation constrains the diffusivity to be less than 1x10(-6)/s and the product of the friction and dilatancy coefficients to be at most 5x10(-7), orders of magnitude smaller than observed at room temperature. In the absence of dilatancy the effective normal stress at failure would be about 55kPa. For this model the observations require intrinsic weakness, low dilatancy, and lithostatic pore fluid.