Does elastic rebound theory apply to seismic faults?

We examine the evolution of and the exchange between two forms of elastic energies stored in the quasidynamic fault model of Ziv and Cochard [1]. The first, E_tect, is due to the integrated slip deficit accumulated between the plate boundaries and the fault surface, and the second, E _fault, is the elastic energy stored as a result of differential slip along the fault surface. Time series of these energies show that large earthquakes occur during the descending portions of the E_tect-curve, and close to the maxima of the E_fault-curve. Interestingly, these results are not in agreement with the classical view of the elastic rebound theory (Reid, [2]). While the elastic rebound theory predicts that large earthquakes occur at the maxima of the E_tect-curve, in our model they occur at the maxima of the E_fault-curve. The latter, arising from the slip heterogeneity along the fault, is not at all accounted for in the elastic rebound theory, on which present earthquake prediction models rely.