Static Coulomb stress load on a three‐dimensional rate‐and‐state fault: Possible explanation of the anomalous delay of the 2004 Parkfield earthquake

We perform quasi‐dynamic modeling of earthquake cycle using laboratory derived rate‐and‐state laws of friction on a homogeneous three‐dimensional fault model. We study effects of the static Coulomb stress loading on clock advance and clock delay of the subsequent event. We carefully investigate dependences of the clock advance on the onset time of the stress load, its amplitude, areal extent, and place of application of the load. We find that these dependences are complex, being controlled by the actual ongoing slip velocity on the fault, especially at the domain of the stress load. In particular, the stress (un)load can initiate the occurrence of quasiperiodic creep‐like episodes, which could be associated with episodic increases of microseismicity on real faults, such as observed on the locked Parkfield segment of the San Andreas Fault. Depending on the load parameters including its timing within the earthquake cycle, one of such creep‐like events may trigger the next (clock advanced) system‐size earthquake. In some cases, the nucleation of the main shock can fail, and the fault experiences one or several seismic events of smaller magnitudes instead. In such a case the next main shock becomes significantly delayed. We speculate that such mechanism could have contributed to the extreme delay of the M 6 2004 Parkfield earthquake. Indeed, the Parkfield segment was subject to Coulomb stress unload due to the 1983 Coalinga‐Nuñez earthquakes and then experienced M 4.9 events in 1993–1994, when the system‐size event was expected. Instead, these failed main shock nucleations delayed the Parkfield earthquake by another ~10 years.