INTERPRETATION OF LARGE-AMPLITUDE VELOCITY PULSES DURING THE 2016 KUMAMOTO EARTHQUAKE AND THE EFFECTS OF THE SHALLOW AND DEEP PARTS OF THE FAULT RUPTURES ON THE NEAR-FAULT GROUND MOTIONS

Abstract

 In this study, we constructed a characterized fault model for the main shock of the 2016 Kumamoto earthquake based on the fault models used in previous studies, and the model was found to simulate the strong motion records around the seismic fault well. After estimating the contribution of the SMGA and LMGA at the observation points adjacent to the seismic faults, the effect of the difference in the top depth of the SMGA on the ground motion adjacent to the seismic faults was investigated. Finally, the effects of shallow and deep parts of fault ruptures on the near-fault ground motions were investigated using a simple seismic fault model. The conclusions of this study can be summarized as follows:

 

 1) By referring to Ozawa et al. (2016) and Yoshida et al. (2017), a characterized fault model that was consistent with surface ruptures was constructed. The characterized fault model consisted of three regions, the SMGA, LMGA, and background region, and its arrangement was estimated by forward modeling with reference to the fault models used in the earlier studies. As a result, one or two SMGAs were determined for each segment. LMGAs were set up to be continuous with SMGAs in the fault dip direction except for LMGA2 located between the Mashiki Town and Nishihara Village.

 2) The pulse-like ground motion with a period of around 1 s in KMMH16 made a large contribution owing to the upward rupture directivity effect. The accuracy of the simulation of KMM006 was improved by estimating multiple SMGAs lying directly under the Mashiki Town and dividing the contribution to the velocity pulses. The long-period velocity pulse at the Site 93048 was also affected by the Idenokuchi fault and the shallow part of the Futagawa fault, as has already been presented in Kido et al. (2019). However, the contribution of the SMGA to the synthetized results was found to greater than that of the LMGA not only for short-period components but also for long-period components.

 3) A sensitivity analysis was performed for the Site 93048 based on varying the width of the LMGA (W_LMGA). As W_LMGA became larger, the long-period EW component of theoretical ground motion at the Site 93048 became smaller and was underestimated compared with the observation record. Differences in the upper-edge depth of the SMGA affected evaluation of both the short-period and long-period components of the ground motion.

 4) The effects of the shallow and deep parts of the fault ruptures on the ground motion adjacent to the seismic fault were investigated using a vertical right strike-slip fault model with surface rupture. As W_LMGA became larger, the velocity response value adjacent to the seismic faults by SMGA became smaller in the broadband frequency range. The calculated ground motions in the vicinity of the fault also strongly depended on the rise time of the shallow slip velocity time function. When the rise time was short and the peak slip velocity was large, significant, large ground motions were generated in the LMGA alone.

 5) When the seismic fault approached near the ground surface, the ground motions evaluated using the full method, i.e., considering both the shallow and deep parts of soil the structure were sometimes different from those obtained using the substructure method, in which the site amplification of the shallow part of the soil structure was separately evaluated.