Probability of earthquake recurrence with nonuniform stress rates and time-dependent failure

Current methods for calculation of long-term probabilities for the recurrence of large earthquakes on specific fault segments are based upon models of the faulting process that implicitly assume constant stress rates during the interval separating earthquakes and instantaneous failure at a critical stress threshold. However, observations indicate that the process of stress recovery following an earthquake involves rate variations at all time scales in addition to stress steps caused by nearby earthquakes. Additionally, the existence of foreshocks, aftershocks and possible precursory processes suggest that there may be significant time dependence of the earthquake nucleation process. A method for determining the conditional probabilities for earthquake occurrence under conditions of irregular stressing is developed that could be useful at all time scales including those pertinent to short-and intermediate-term prediction. Used with models for earthquake occurrence at a stress threshold, the addition of variable stressing introduces a simple scaling of the conditional probabilities by stress level and stress rate. A model for the time-dependent nucleation of earthquake slip has been proposed recently that is based upon laboratory observations of fault strength. This failure criterion results in large but relatively short duration changes in the probability of earthquake recurrence particularly following stress steps. Applied to populations of earthquakes the models predicts a 1/t decay of seismicity following stress steps as observed for aftershocks and for frequency of foreshock-mainshock pairs. The model suggests that variations of seismicity rates of small earthquakes in the nucleation zone of the expected earthquake directly indicate variations in probability of recurrence of the large earthquake.