15:30 〜 16:30
[S09-P-03] Time to instability of the seismic event triggered by SSE
At the subduction zone along the Nankai Trough, SW Japan, large earthquakes around M8 had repeatedly occurred. The existence of short-term large aseismic processes at the deeper extension of the seismogenic region before the 1944/46 events has been inferred (e.g. Linde and Sacks, 2002).
We showed that such preseismic process could occur in 2D earthquake cycle models controlled by the friction with intrinsic cut-off time for healing (Ohtani et al., 2016). In the simulated cycles, large slow slip events (SSEs) occurred at the deeper extension of the seismogenic zone, and sometimes triggered the earthquake. We systematically investigated over a range of model parameters and found that the time between an earthquake and the most recent SSE, tf, fell into two separate classes; within 38.4 days or over 7.7 years. The former represented SSE-triggered earthquakes, and tf takes a few days on average. Given that the timing of SSEs is probably not affected by the failure condition of the seismogenic part, the observed concentration of tf to small values seems to be unexpectedly strong.
In order to understand the result, we here investigate the slip response to the stress perturbation in a 1D spring-slider model with RSF friction. To model SSE-triggering in the 1D system, we move the load point at the velocity (1–r)Vpl, and give the step Δu=rVplTr_load every Tr_load=34 years. The slider produces periodic unstable slip with a recurrence interval of Tr=102 years, that is, an earthquake occurs after third SSE. Then, we impose the third SSE at a random timing T0 to synthesize the statistics of tf.
For larger SSE, tf concentrates more strongly to smaller values. We calculated the cumulative probability that earthquake occured by tf. Even in a case of large SSE (r=0.9), 50% of earthquakes required tf>0.36 years. It seems that the strong concentration of tf observed in the 2D model is not simply explained by the delayed (RSF) failure mechanism represented in 1D model.
We showed that such preseismic process could occur in 2D earthquake cycle models controlled by the friction with intrinsic cut-off time for healing (Ohtani et al., 2016). In the simulated cycles, large slow slip events (SSEs) occurred at the deeper extension of the seismogenic zone, and sometimes triggered the earthquake. We systematically investigated over a range of model parameters and found that the time between an earthquake and the most recent SSE, tf, fell into two separate classes; within 38.4 days or over 7.7 years. The former represented SSE-triggered earthquakes, and tf takes a few days on average. Given that the timing of SSEs is probably not affected by the failure condition of the seismogenic part, the observed concentration of tf to small values seems to be unexpectedly strong.
In order to understand the result, we here investigate the slip response to the stress perturbation in a 1D spring-slider model with RSF friction. To model SSE-triggering in the 1D system, we move the load point at the velocity (1–r)Vpl, and give the step Δu=rVplTr_load every Tr_load=34 years. The slider produces periodic unstable slip with a recurrence interval of Tr=102 years, that is, an earthquake occurs after third SSE. Then, we impose the third SSE at a random timing T0 to synthesize the statistics of tf.
For larger SSE, tf concentrates more strongly to smaller values. We calculated the cumulative probability that earthquake occured by tf. Even in a case of large SSE (r=0.9), 50% of earthquakes required tf>0.36 years. It seems that the strong concentration of tf observed in the 2D model is not simply explained by the delayed (RSF) failure mechanism represented in 1D model.