Fault Response to Variations of Loading Rate and/or Dynamic Stress Transients, from the Analysis of 1811 Repeating Micro-Earthquakes at Parkfield (CA)

Using an approach developed for quantifying ground-motion, we analyze the spectral characteristics of 1811 repeating micro-earthquakes from 163 clusters near Parkfield (CA). Corner frequencies of a selected subset of events are also computed (Brune spectral model). Our calculations provide unprecedented accuracy on the calculation of the seismic moments of each individual event. A strong transient (3-10+ years) affects the spectral amplitudes of most repeating micro-earthquakes after the Parkfield Mw 6.0 main shock of 2004, showing a significant increase of amplitudes at the onset of the SAF postseismic response. The largest relative moment variation ((ΔM0)max/M0regular) with respect to unperturbed events, the maximum creep velocity (VMax), and the postseismic relaxation time (τ), all systematically increase, along the creeping section of the SAF, toward the area that slipped coseismically during the 2004 Parkfield main earthquake. Postseismic perturbations of the repeats' seismic moments are up to an order of magnitude larger than the typical seismic moments of the undisturbed events; recurrence times also show a systematic behavior. Dynamic stress perturbations (increased loading rate of the specific asperities, and/or changes of normal stress across the SAF) are likely to be responsible for the observed variations of the repeats' seismic moments and recurrence times. In addition to the postseismic response to a large event on the (locked) adjacent portion of the SAF, anomalous responses may be induced by: i) giant events at teleseismic distances (i.e., Sumatra, 2004); ii) strong regional events, within ~100 km (i.e., Loma Prieta, 1989, and Landers, 1992); iii) small earthquakes (M~2) in close vicinity to specific clusters. After the 2004 Parkfield main shock, the (shallow) clusters of small-magnitude repeats showed the largest relative perturbations, and the largest absolute cumulative slips. We point out an apparent inconsistency between the self-similar source scaling observed in this study, and the slip-moment relationship proposed by Nadeau and Johnson (1998). The model described by Sammis et al. (1999), in which a fractal distribution of asperities describes the seismogenic patches, may be used to explain the mentioned differences in scaling.