5:15 PM - 6:30 PM
[SSS11-P20] Evaluation of near-source ground motion during the 2008 Iwate-Miyagi Nairiku earthquake by using stochastic slip models and three-dimensional velocity structures
Keywords:Iwate-Miyagi Nairiku earthquake, fault displacement, stochastic slip distribution, reciprocity theorem, finite-difference method
The detailed rupture process and heterogeneous subsurface structures should be considered in ground motion evaluations near the source fault at a large earthquake. Numerical simulation methods such as the finite difference method (FDM) and the finite element method (FEM) can incorporate the heterogenous fault slip and three-dimensional structure models into calculations and are useful for grasping seismic motion not only at individual points but also on whole ground surface and seafloor. However, these methods are computationally expensive because they treat a large number of grids and meshes in the computational domain when numerically solving the governing equation, which causes practical limitations on the number of models that can be used as inputs for simulations. However, in the case of a few number of receivers or stations to be evaluated, such as ground motion simulations at important facilities such as power plants, the Green's functions for many point sources can be efficiently calculated by interchanging the source and receiver locations based on the reciprocity theorem and the total number of numerical simulations for many source models can be significantly reduced compared with that by the straightforward approach (e.g., Eisner and Clayton). In this study, we evaluate the displacements including the static component and their deviations near the source fault of the 2008 Iwate-Miyagi Nairiku, Japan, earthquake by calculating waveforms with the reciprocal approach for various slip distribution models that are stochastically generated. We focus on KiK-net IWTH25 station that recorded the peak acceleration amplitude of 3866 gal and the static displacement of 140 cm in the vertical component at the earthquake (Aoi, et al., 2008; Aoi and Morikawa, 2009).
We calculated the Green's functions by using the 3D FDM developed by Nakamura et al. (2012). The finite-fault inversion result for the 2008 earthquake by Asano and Iwata (2011) was employed as a reference model to stochastically generate slip distributions on the fault plane. From their model, we generated 100 cases for the heterogeneous slip distribution based on the slip probability density function (SPDF) by Murphy et al. (2016). We used a point source on a subfault and set a total of 315,216 subfaults with a dense interval of 50 m on the fault plane. We assessed a smoothly spatial distribution of displacements near the fault by employing such dense interval in our simulations, whereas the distribution showed a periodic deformation pattern that is not observed by satellite geodetic surveys if the large subfault interval or size is employed. We used the same initial break point, rake angle, and rupture velocity as those presented in Asano and Iwata (2011). The rise time in subfault was set according to its subfault size. For subsurface velocity structures, we used the Japan Seismic Hazard Information Station (J-SHIS) model (Fujiwara et al., 2009). After confirming an excellent fitting of synthetic waveforms at the IWTH25 surface station for several cases between the straightforward simulation and the reciprocal approach, we generated synthetic waveforms for all slip distribution cases by postprocessing the reciprocal Green's functions.
Our calculation results show that the peak amplitude of the displacement waveform for the fault parallel and orthogonal components and the vertical one is 46±12 cm, 32±11 cm, and 88±26 cm, respectively. The static displacement, which is the mean value in a time window of 40-45 s in which the waveform amplitude converges to a constant value, for the three components is 32±13 cm, 4±10 cm, and 69±20 cm, respectively. Because the source mechanism of the reference model is a reverse fault with the rake and dip angle of 108 and 51 degrees, respectively, and IWTH25 is located just above the fault plane, both of the static displacement and the peak displacement in the vertical component are larger than those in the horizontal ones in all cases except 5 ones. The static and peak displacement in the vertical component is 6.9 and 6.4 times difference among the cases, respectively, and the maximum value of these displacements is 112 cm and 144 cm, respectively. The peak displacement amplitude in the component is 1.3 times larger on average than the static displacement. Our results also show an attenuation feature of the displacement amplitudes according to the hypocentral distance from the maximum slip position in the source fault. Because IWTH25 is several kilometers away from the surface fault areas, this attenuation relation is significantly associated with the distance to the station location projected on the fault plane compared with the depth of the slip position.
We calculated the Green's functions by using the 3D FDM developed by Nakamura et al. (2012). The finite-fault inversion result for the 2008 earthquake by Asano and Iwata (2011) was employed as a reference model to stochastically generate slip distributions on the fault plane. From their model, we generated 100 cases for the heterogeneous slip distribution based on the slip probability density function (SPDF) by Murphy et al. (2016). We used a point source on a subfault and set a total of 315,216 subfaults with a dense interval of 50 m on the fault plane. We assessed a smoothly spatial distribution of displacements near the fault by employing such dense interval in our simulations, whereas the distribution showed a periodic deformation pattern that is not observed by satellite geodetic surveys if the large subfault interval or size is employed. We used the same initial break point, rake angle, and rupture velocity as those presented in Asano and Iwata (2011). The rise time in subfault was set according to its subfault size. For subsurface velocity structures, we used the Japan Seismic Hazard Information Station (J-SHIS) model (Fujiwara et al., 2009). After confirming an excellent fitting of synthetic waveforms at the IWTH25 surface station for several cases between the straightforward simulation and the reciprocal approach, we generated synthetic waveforms for all slip distribution cases by postprocessing the reciprocal Green's functions.
Our calculation results show that the peak amplitude of the displacement waveform for the fault parallel and orthogonal components and the vertical one is 46±12 cm, 32±11 cm, and 88±26 cm, respectively. The static displacement, which is the mean value in a time window of 40-45 s in which the waveform amplitude converges to a constant value, for the three components is 32±13 cm, 4±10 cm, and 69±20 cm, respectively. Because the source mechanism of the reference model is a reverse fault with the rake and dip angle of 108 and 51 degrees, respectively, and IWTH25 is located just above the fault plane, both of the static displacement and the peak displacement in the vertical component are larger than those in the horizontal ones in all cases except 5 ones. The static and peak displacement in the vertical component is 6.9 and 6.4 times difference among the cases, respectively, and the maximum value of these displacements is 112 cm and 144 cm, respectively. The peak displacement amplitude in the component is 1.3 times larger on average than the static displacement. Our results also show an attenuation feature of the displacement amplitudes according to the hypocentral distance from the maximum slip position in the source fault. Because IWTH25 is several kilometers away from the surface fault areas, this attenuation relation is significantly associated with the distance to the station location projected on the fault plane compared with the depth of the slip position.