Earthquake Stress Transfer within Continents: Migrating Earthquakes and Long Aftershock Sequences

Stress transfer after large earthquakes has different effects within continental plate interiors than at plate boundaries. These differences explain why the spatio-temporal patterns of earthquakes in the two environments are so different given that the physics of fault rupture is essentially the same. At plate boundaries, steady plate motion reloads faults quickly after large earthquakes. This tectonic reloading soon dominates all other stress effects including those due to earthquakes on other faults. As a result, fault segments produce quasi-periodic earthquakes. In contrast, after a large mid-continental earthquake, stress reaccumulates very slowly - if at all - on the fault that broke. Instead, the resulting stress changes can give rise to earthquakes on other faults that have been quiescent for a long time. As a result, continental seismicity is often episodic, temporally clustered, and migrate between faults over thousands of years. A striking example of this variability is in North China, where large (M>7) earthquakes have been frequent, but not a single one repeated in the same place since 1300 A.D. For similar reasons, aftershock sequences within continents can last hundreds of years. Aftershocks result from changes of stress and fault properties induced by the main shock. At a plate boundary, steady plate motion quickly reloads the fault after a large earthquake and overwhelms the effects of the main shock. Within continents, however, the faults are reloaded much more slowly, allowing aftershocks to continue much longer. The observed aftershock sequence lengths are consistent with a simple model based on rate-and-state fault friction that predicts that the length of aftershock sequences varies inversely with the fault loading rate. Recognizing the migration of large earthquakes and long aftershock sequences is crucial for seismic hazard assessment within continents. Because many small earthquakes are probably aftershocks, their locations are unlikely to indicate the timing or the locations of future large earthquakes. Moreover, because in the short term much of the seismic hazard results from aftershocks, which can be damaging, these aftershocks should not be removed in attempts to infer earthquake recurrence. In the longer term, relying unduly on recent seismicity to predict the locations of future large earthquakes will overestimate the hazard in some places and lead to surprises elsewhere.