Self-organization of spatio-temporal earthquake clusters

Cellular automaton versions of the Burridge-Knopoff model have been shown t o reproduce the power law distribution of event sizes; that is, the Gutenbe rg-Richter law. However, they have failed to reproduce the occurrence of fo reshock and aftershock sequences correlated with large earthquakes. We show that in the case of partial stress recovery due to transient creep occurri ng subsequently to earthquakes in the crust, such spring-block systems self -organize into a statistically stationary state characterized by a power la w distribution of fracture sizes as well as by foreshocks and aftershocks a ccompanying large events. In particular, the increase of foreshock and the decrease of aftershock activity can be described by, aside from a prefactor , the same Omori law. The exponent of the Omori law depends on the relaxati on time and on the spatial scale of transient creep. Further investigations concerning the number of aftershocks, the temporal variation of aftershock magnitudes, and the waiting time distribution support the conclusion that this model, even "more realistic" physics is missed, captures in some ways the origin of the size distribution as well as spatio-temporal clustering o f earthquakes.