[S-CG64_30PM2] スロー地震

★ [SCG64-19] 速度・状態依存摩擦を用いた階層アスペリティの動的地震サイクルシミュレーションで現れる大地震前の挙動

*野田 博之¹、中谷 正生²、堀 高峰¹ (1.海洋研究開発機構、2.東京大学)

キーワード：Earthquake sequence, Preseismic phenomena, Hierarchical asperity, Numerical simulation

Understanding preseismic phenomena before large earthquakes is of critical importance in assessing possibility of disaster mitigation by detecting and recognizing them. The 2011 Tohoku-Oki earthquake has long recorded geophysical data for tens of years prior to it. Since the earthquake, multiple studies have reported potentially important phenomena involving slow slip which may be particular to ripe asperities. It is our mission for modelers to see if they are consistent with, or appear naturally without fine tuning of numerical models of earthquake sequences accounting for interseismic processes, as well as earthquake ruptures.The off-Miyagi to off-Fukushima region was locked at least from Apr. 1995 to Mar. 2002 [Nishimura et al. 2004], with the shallower region not being able to be constrained by on-land GPS stations [Loveless and Mead, 2011]. The region started creeping from 2005 [Ozawa et al., 2012]. Recently, Katsumata [2013, JpGU] pointed out that seismic quiescence [Katsumata, 2011] correlates with the locked period, and inferred that this region may have been creeping at least from 1980 to 1988. In the shallower region near the hypocenter of the Tohoku-Oki earthquake, a couple of slow slip events were reported by Ito et al. [2013], one in Nov. 2008 and the other in Feb. 2011. This interval is much shorter than that for the larger scale events inferred by Katsumata [2013].Suito et al. [2011] reported that M7-class earthquakes along the Japan Trench after 2005, including the Mw 7.3 preshock 2 days before the Tohoku-Oki earthquake, had unusually large amount of afterslips. The postseismic moment releases are comparable to or even larger than the coseismic ones, with the centroid being located close to the epicenters, not deeper than them.In the present talk, we present that qualitatively similar behaviors to those observations are recognized in numerical models reported by Noda et al. [2013, JGR]. They presented rate-state earthquake sequence simulations accounting for a hierarchical asperity concept [Ide and Aochi, 2005]; a large tough patch has a small fragile patch in it. Importantly, those simulations were not meant to mimic the Tohoku-Oki earthquake through fine tuning of the model, and are representing general behaviors characteristic to the rate-state (aging law) earthquake sequence with a certain kind of heterogeneity in the parameter distribution.In those simulations, interseismic penetration of a creep front into a locked velocity-weakening patch often becomes non-steady and accompanied by aseismic transients before nucleation. This is because the critical length scale for impossibility of coherent steady-state slip [Rice et al., 2001] can be smaller than the nucleation size [Rubin and Ampuero, 2005]. In the simulation, the transients take place both in the large tough patch and in the small fragile patch when a creep front penetrates inwards to a certain extent. A transient does not necessarily, but may lead to nucleation. In addition, such an elevated aseismic slip rate in the large patch seems to be a necessary condition for cascade-up rupture growth from the small patch if it is smaller than the nucleation size of the large patch.A small event which only ruptures the small patch is sometimes followed by a large event before the afterslip smearing out. Such small events are classified as precursory events, since clear causality is recognized between them and the following large ones; the large ones are initiated either by delayed cascade-up or by large nucleation hosted by the afterslip. The precursory small events tend to have larger afterslip than non-precursory ones.In the rate- and state-dependent friction law, logarithmic slip rate is, by definition, proportional to stress minus strength which correlates with fracture energy. Therefore temporal changes in the aseismic slip rate in a so-called asperity, if detectable, could be used to infer the ripeness of it.