Abstract
We investigate the dynamics of a modified Burridge-Knopoff model by introducing a dissipative term to mimic the bursts of acoustic emission (AE) from rock samples. The model explains many features of the statistics of AE signals observed in experiments such as the crossover in the exponent value from relatively small-amplitude AE signals to a larger regime, and their dependence on the pulling speed. Significantly, we find that the cumulative energy dissipated identified with acoustic emission can be used to predict a major slip event. We also find a data collapse of the acoustic activity for several major slip events describable by a universal stretched exponential with corrections in terms of time to failure.