Mechanically understand foreshock-afterslip-mainshock sequence of 2011 Great Tohoku-oki earthquake

The 2011 M 9.0 Tohoku-oki, Japan, earthquake is the best recorded event and this gives us a unique opportunity to detailedly investigate the processes of the initiation and propagation of faulting. It was observed that the mainshock of the Tohoku-oki earthquake was triggered by the M7.3 largest foreshock with time delay of two days associated with clear migration of seismicity approaching the mainshock hypocenter; this suggests the existence of underlying post-seismic slip or afterslip (Ando and Imanishi, 2010, EPS). To test this hypothesis, we first concern the kinematically inverted slip profiles of M7.3 foreshock, its afterslip and the mainshock. For the foreshock sequence, the coseismic- and after-slip profiles were geodetically inverted by Ohta et al 2012, GRL, which considered the records from the ocean bottom pressure sensors installed just above the focal area. For the mainshock, Uchide 2013, GRL, conducted the multi-scale slip inversion enabling to resolve the first few ten seconds of the mainshock. Given these slip profiles, we then applied the 3-D BIEM (Ando and Okuyama, 2010, GRL) to calculate the stress drop distribution on the fault plane. The obtained shear traction changes exhibit the increase due to the foreshock and afterslip around the foreshock focal area, and considerable dynamic stress drop due to the mainshock rupture near the foreshock area. Next, we conduct the quasi-static and dynamic rupture simulations (or the forward modeling) to investigate the mechanical compatibility of the kinematic inversion results. The quasi-static simulation is used for the foreshock-afterslip sequence, and the dynamic simulation is used for the mainshock. We assume the mechanical fault model that concerns the rheological heterogeneity on a fault, consisting of the velocity strengthening (ductile) background and otherwise the velocity weakening (brittle) patches on the fault (Ando et al., 2012, JGR). We obtained the preliminary results showing a certain class of simple models can reproduce the observationally constrained traction change associated with the foreshock-afterslip-mainshock sequence. The delayed triggering due to the propagating afterslip is also reproduced, attributed to the velocity strengthening rheology of the considered fault model.