Expansion of aftershock areas caused by propagating post-seismic sliding

Abstract

SUMMARY It is known that the aftershock areas of large earthquakes often expand over time. This expansion indicates that the stress-increase associated with a main shock gradually propagates outward from the main shock rupture area. It is likely that this propagating stress-increase is caused by the propagation of post-seismic sliding, which has been detected for many large earthquakes from geodetic observations, mainly global positioning system (GPS). In the present study, I perform a numerical simulation of the expansion of the aftershock area caused by the propagation of post-seismic sliding. The model fault exists within an infinite elastic medium and is loaded at a constant displacement rate. The frictional stress on the fault obeys a laboratory-derived rate- and state-dependent friction law. Non-uniformity of the frictional constitutive parameters is introduced to the model fault plane to represent a large asperity for a large earthquake (main shock) and many possible nucleus sites for aftershocks around the main shock asperity. Negative values of A − B are assigned to areas of seismic slip for the main shock and aftershocks, while positive values of A − B are assigned to regions where aseismic sliding occurs. Here, A − B is defined by dτ ss/d ln V that describes the velocity (V ) dependence of steady-state friction stress τ ss. Seismic slip may be nucleated for negative A − B, and aseismic slip occurs for positive A − B. Slip histories on the model fault plane are simulated through numerical integration of the friction law under the condition of uniform shear loading of a constant velocity. The simulation can reproduce aftershocks triggered by stress-increases related to postseismic sliding and expansion of the aftershock area. The rate of expansion of the aftershock area decreases with increasing distance from the main shock asperity or increasing value of A − B in the velocity-strengthening region. This finding suggests that the A − B value for a plate interface can be estimated from aftershock data. The simulation indicates that the 7-day and 30-day aftershock areas are 14‐22 and 24‐39 per cent larger, respectively, than the 1-day aftershock area. These numbers are approximately consistent with observations from large earthquakes at subduction zones.