Why aftershock duration matters for probabilistic seismic hazard assessment

Abstract

Most hazard assessments assume that high background seismicity rates indicate a higher probability of large shocks and, therefore, of strong shaking. However, in slowly deforming regions, such as eastern North America, Australia, and inner Honshu, this assumption breaks down if the seismicity clusters are instead aftershocks of historic and prehistoric mainshocks. Here, therefore we probe the circumstances under which aftershocks can last for 100–1000 years. Basham and Adams (1983) and Ebel et al. (2000) proposed that intraplate seismicity in eastern North America could be aftershocks of mainshocks that struck hundreds of years beforehand, a view consonant with rate–state friction (Dieterich, 1994), in which aftershock duration varies inversely with fault‐stressing rate. To test these hypotheses, we estimate aftershock durations of the 2011 Mw 9 Tohoku‐Oki rupture at 12 sites up to 250 km from the source, as well as for the near‐fault aftershocks of eight large Japanese mainshocks, sampling faults slipping 0.01 to 80 mm/yr . Whereas aftershock productivity increases with mainshock magnitude, we find that aftershock duration, the time until the aftershock rate decays to the premainshock rate, does not. Instead, aftershock sequences lasted a month on the fastest‐slipping faults and are projected to persist for more than 2000 years on the slowest. Thus, long aftershock sequences can misguide and inflate hazard assessments in intraplate regions if misinterpreted as background seismicity, whereas areas between seismicity clusters may instead harbor a higher chance of large mainshocks, the opposite of what is being assumed today.