We introduce a crust relaxation process in a continuous cellular automaton
version of the Burridge-Knopoff model. Analogously to the original model, o
ur model displays a robust power law distribution of event sizes (Gutenberg
-Richter law). The principal new result obtained with our model is the spat
iotemporal clustering of events exhibiting several characteristics of earth
quakes in nature. Large events are accompanied by a precursory quiescence a
nd by localized fore- and aftershocks. The increase of foreshock activity a
s well as the decrease of aftershock activity follows a power law (Omori la
w) with similar exponents p and q. All empirically observed power law expon
ents, the Richter B-value, p and q and their variability can be reproduced
simultaneously by our model, which depends mainly on the level of conservat
ion and the relaxation time.