17:15 〜 18:30
[SSS11-P19] Characterized source model of M7.3 2016 Kumamoto eq. by 3D reciprocity GFs inversion with special reference to velocity pulse at KMMH16
キーワード：2016 Kumamoto earthquake, source inversion, seismic reciprocity, characterised source model, strong motion generation area, Mashiki
2016 Kumamoto earthquakes caused severe damage centering on Mashiki residential area. The velocity waveforms at KMMH16 station in Mashiki during the M7.3 event show a large velocity pulse observed both by the surface and downhole sensors. Kawase et al., 2017, found that severe damage in Mashiki may be result of strong westward velocity pulse. Question raised by the 2016 Kumamoto earthquakes is how the near-source ground motions with strong velocity pulses at KMMH16 were generated during M7.3 event.
In this study, we focus on the characterized SMGA source model. EGF method is widely used for source modelling in this case. However, in case of target site is located just near the fault in the nodal plane of the source mechanism (like KMMH16), the mechanism of the EGF event should perfectly fit the mechanism of the mainshock, which is a rare case. Otherwise, large variations of amplitudes in the vicinity of the nodal plane will produce large errors of ground motion synthetics. Instead, we will use theoretical 3D GFs.
Our approach is non-linear source inversion. This method requires calculation of waveforms and comparison with observations for many source models. To accelerate these calculations, we use precalculated Green’s functions (GF) by the reciprocity method and the JIVSM velocity structure model. By comparison with aftershock record, we validated this structure for as short as 2 s in periods. For mainshock inversion we expand the target period to 1.5 sec. Target sites are limited to sites close to the fault: KMM005, KMM006, KMMH06, KMMH14, and KMMH16.
First, we explore the parameter space with the simplex search method with the initial SMGA model of Somei et al. (2019). In this process we were able to reduce a number of search parameters by limiting to effective parameters only. Also, necessary physical constraints for rupture velocity, rupture starting point, etc., as well as their strength, were estimated here. Then, we try to look for an alternative initial SMGA source model by the grid search method, and finally, estimate source model with the simplex search method again.
Resulting model and waveforms for KMMH16 are shown in figure below. Location of SMGAs is deeper than in initial model of Somei et al. (2019). For SMGA2 the deep location agrees well with other source inversions (e.g., Yoshida et al., 2017). SMGA rupture initiation points have physically reasonable location allowing outward propagation from the mainshock hypocenter, without any reverse ruptures. Waveform reproduces short period westward pulse at KMMH16.
Acknowledgements. We use strong motion data from K-NET, KiK-net, and CMT solutions from F-net, provided by the NIED.
In this study, we focus on the characterized SMGA source model. EGF method is widely used for source modelling in this case. However, in case of target site is located just near the fault in the nodal plane of the source mechanism (like KMMH16), the mechanism of the EGF event should perfectly fit the mechanism of the mainshock, which is a rare case. Otherwise, large variations of amplitudes in the vicinity of the nodal plane will produce large errors of ground motion synthetics. Instead, we will use theoretical 3D GFs.
Our approach is non-linear source inversion. This method requires calculation of waveforms and comparison with observations for many source models. To accelerate these calculations, we use precalculated Green’s functions (GF) by the reciprocity method and the JIVSM velocity structure model. By comparison with aftershock record, we validated this structure for as short as 2 s in periods. For mainshock inversion we expand the target period to 1.5 sec. Target sites are limited to sites close to the fault: KMM005, KMM006, KMMH06, KMMH14, and KMMH16.
First, we explore the parameter space with the simplex search method with the initial SMGA model of Somei et al. (2019). In this process we were able to reduce a number of search parameters by limiting to effective parameters only. Also, necessary physical constraints for rupture velocity, rupture starting point, etc., as well as their strength, were estimated here. Then, we try to look for an alternative initial SMGA source model by the grid search method, and finally, estimate source model with the simplex search method again.
Resulting model and waveforms for KMMH16 are shown in figure below. Location of SMGAs is deeper than in initial model of Somei et al. (2019). For SMGA2 the deep location agrees well with other source inversions (e.g., Yoshida et al., 2017). SMGA rupture initiation points have physically reasonable location allowing outward propagation from the mainshock hypocenter, without any reverse ruptures. Waveform reproduces short period westward pulse at KMMH16.
Acknowledgements. We use strong motion data from K-NET, KiK-net, and CMT solutions from F-net, provided by the NIED.