Forecasting the next Parkfield mainshock on the san An- dreas fault

A physical model was recently proposed to describe the phenomenon of coupling erosion that took place in the Japan Trough between 1998 and 2009, and the subsequent dynamic rupture occurred during the 2011 M9.1 Tohoku-Oki earthquake. Although 75% of the coupled area of the Japanese subduction was eroded away before nucleation, coseismic slip displaced both the locked (velocity weakening) and the eroded (velocity-strengthening) parts of the asperity. Here we show that a similar phenomenon of erosion repeatedly takes place at Parkfield on a NW patch of the SAF close to the asperity responsible for the repeating M6 earthquakes. We consider the variance of the spatial center of daily seismic activity along the SAF fault calculated on a moving time window. Initially the variance linearly grows due to increasing frictional engagement up to a maximum value. Then a process of erosion of the coupled area of the fault linearly reduces the variance until the stress is transferred onto the adjacent asperity, leading to failure. When halted due to a stress perturbation from the 1983 Coalinga earthquake, the process promptly resumes a virtually unchanged increasing trend. The stable and regular decrease of the variance started in early 1988 allows a very accurate retrospective prediction of the time of occurrence of the 2004 main shock. The process is repeating itself during the current seismic cycle, which, if undisturbed, will produce another mainshock in mid-2024.