Distinctive pulse motions were recorded near the seismic faults in the source region of the main shock of the 2016 Kumamoto earthquake. Large-scale crustal deformation and surface ruptures occurred. The peak ground velocity in Nishihara Village exceeded 250 cm/s, this velocity was generated by a long-period pulse of approximately 3s. Mashiki town was situated within a severely damaged zone with a local range approximately 1 km wide. In order to clarify the relationship between the building damage and the pulse-type ground motions, it is important to clarify the generation mechanism of large-amplitude pulse-type ground motions, which greatly influence the response of various types of structures including super-high-rise and isolated buildings.
In this study, we constructed a seismic source model of the main shock of the 2016 Kumamoto earthquake to simulate the corresponding strong ground motions, including those recorded in the vicinity of the seismic faults and in the building damage zone in Mashiki town.
The conclusions of this study are summarized as follows:
1) First, we examined the heterogeneous source model proposed by Hikima (2016). The correction spectrum was applied to compensate for the short-period range amplitude reduction and to incorporate the sub-fault rupture propagation into the calculation. Therefore, velocity waveforms at distant observation stations were simulated over the whole period. In the waveform inversion analysis, the effective frequency range is limited and Nishihara Village are not included, therefore, the large-amplitude pulses that have a period of approximately 1s recorded in Mashiki town and those recorded in Nishihara Village could not be reproduced.
2) Second, we constructed a characterized fault model (Model-C1), comprising multiple SMGAs and LMGAs by referring to models proposed by SATOH (2017), Irikura et al. (2017) and Tanaka et al. (2017), respectively. The source process, including rupture from the bottom towards Mashiki town, was indispensable for generating large pulse waves with a period of 1s, as observed at the KiK-net Mashiki stations. However, the distribution of peak ground velocity in the fault-normal-direction line showed a peak in the North direction at the KiK-net Mashiki station; this distribution differed from the damage concentration area of the wooden houses.
3) Third, using the above characterized fault model, long-period pulses with large amplitudes as observed at Nishihara Village were successfully simulated. Surface slip rupture due to the nearby LMGA and normal fault slip of the Idenoguchi faults, both of which are parallel to the main faults, were found to largely contribute to the long-period pulses that were recorded at the Nishihara Village.
4) Fourth, in order to reproduce the distribution of the strong ground motions corresponding to the local damage area, we proposed another characterized fault model (Model-C2) in which the top of the SMGA was located at a relatively shallow area just under Mashiki town. Using this analysis, the pulse velocity in Mashiki town was simulated. However, the reproducibility of strong ground motion records at stations that are distant from the hypocenter is not sufficiently good.
5) Finally, the distant observed ground motion records have been successfully reproduced in a broad area including the vicinity of the seismic faults through the combination of the heterogeneous fault model and the characterized fault model (Model-C2). The distribution of the strong ground motions corresponding to the local damage area in Mashiki town was also reproduced.
抄録全体を表示