The edges of large earthquakes and the epicenters of future earthquakes: Stress-induced correlations in elastodynamic fault models

Fault models can generate complex sequences of events from frictional instabilities, even when the material properties are completely uniform along the fault. These complex sequences arise from the heterogeneous stress and strain fields which are produced through the dynamics of repeated ruptures on the fault. Visual inspection of the patterns of events produced in these models shows a striking and ubiquitous feature: future events tend to occur near the edges of where large events died out. In this paper, we explore this feature more deeply. First, using long catalogues generated by the model, we quantify the effect. We show, interestingly, that it is an even larger effect for future small events than it is for future large events. Then, using our ability to directly measure all aspects of the model, we find a physical explanation for our observations by examining the stress fields associated with large events. Looking at the average stress field we see a large stress concentration left at the edge of the large events, out of which the future events emerge. Further, we see the smearing out of the stress concentration as small events occur. This indicates why the epicenters of future small events are more correlated with the edges of large events than are the epicenters of future large events. Finally, we discuss how results from our simple model may be relevant to the more complicated case of the earth.