5:15 PM - 6:30 PM
[SSS11-P18] Ground-motion simulation for the 2016 Kumamoto earthquake (Mj7.3) based on the strong-motion prediction recipe
Keywords:2016 Kumamoto earthquake, strong motion, ground-motion simulation, active faults
The 2016 Kumamoto earthquake occurred on April 16, 01:46 JST (MJMA7.3) revealed several issues on the strong-motion prediction in terms of fault size, seismic moment, etc. In this study, we attempt to “predict” the ground motion for the Kumamoto earthquake based on different fault models, aiming to clarify the key parameters and uncertainties to be considered for improving strong-motion prediction for active faults.
First, we set a reference fault model by following the procedure by the “recipe” (Headquarters for Earthquake Research Promotion of Japan, 2017) assuming the fault length was set to be a realistic number, 34 km a priori. The seismic moment calculated from the reference model via the S - M0 (fault area – seismic moment) relation of the recipe is less than 50 % of the observation. We set two types of alternative models: (1) fault models with the same fault size as the reference model but with doubled seismic moment and (2) fault models with larger fault length but with the same S - M0 relation.
Using the various fault models and a detailed 3-D velocity structure model (Senna et al. 2017), broadband ground motion was computed by a hybrid technique combining 3-D finite-difference method at long periods (> 1 s) and the stochastic Green’s function method at short periods (< 1 s). We show the effect of considering the uncertainties in S - M0 relation, fault length, and the depth of fault top on the performance of the ground motion prediction.
First, we set a reference fault model by following the procedure by the “recipe” (Headquarters for Earthquake Research Promotion of Japan, 2017) assuming the fault length was set to be a realistic number, 34 km a priori. The seismic moment calculated from the reference model via the S - M0 (fault area – seismic moment) relation of the recipe is less than 50 % of the observation. We set two types of alternative models: (1) fault models with the same fault size as the reference model but with doubled seismic moment and (2) fault models with larger fault length but with the same S - M0 relation.
Using the various fault models and a detailed 3-D velocity structure model (Senna et al. 2017), broadband ground motion was computed by a hybrid technique combining 3-D finite-difference method at long periods (> 1 s) and the stochastic Green’s function method at short periods (< 1 s). We show the effect of considering the uncertainties in S - M0 relation, fault length, and the depth of fault top on the performance of the ground motion prediction.