09:15 〜 09:30
[SSS10-12] Ground-Motion Models Incorporating Non-Ergodic Effects from 3-D Numerical Simulations in the Miyagi-Fukushima and Kanto Regions
キーワード：Non-ergodic effect, Ground Motion Model, 3-D Numerical Simulation, Aleatory Variability, Epistemic Uncertainty
The non-ergodic ground-motion model (NGMM) has the capability to account for the systematic source, path, and site effects, while also reducing the aleatory variability (e.g., Lavrentiadis et al., 2023; Sung et al., 2023). Capturing non-ergodic results for small geographical areas requires ground motion data for specific site and source pairs, however, many regions lack sufficient empirical ground-motion data to constrain the non-ergodic terms for larger earthquakes at short distances. Numerical simulations can provide valuable insights into the characteristics of earthquake ground motions for scenarios involving large magnitudes (Mw > 6.5) and short distances (< 20 km) and can also consider the site-specific amplification from deep sedimentary basins. Some studies (e.g., Sung and Abrahamson, 2022; Sung et al., 2023) have demonstrated that non-ergodic site and path from suites of 3-D numerical simulations of ground motions can be incorporated into ground-motion models (GMMs) to capture the non-ergodic effects.
In this study, scenario ground-motion simulations from the Miyagi-Fukushima and Kanto regions (NIED, 2019, https://doi.org/10.17598/nied.0010) are used to develop a model for the spatial distribution of non-ergodic site terms (dS2S) for ground motions of M7-class crustal events. The Morikawa and Fujiwara (2013) GMM (MF13 GMM) is selected as an ergodic model. The approach outlined by Sung and Abrahamson (2022) is adopted to develop the NGMM using the 3-D simulation results: (1) the basin term of the MF13 GMM is modified to be consistent with the basin scaling from the 3-D simulations, and (2) the spatial distribution of the dS2S is estimated through the application of the varying coefficient model (VCM) within the region encompassed by the simulations. The dS2S from the simulation exhibits a stronger scaling with the basin depth as compared to the dS2S estimated by Sung et al. (2024) using the K-NET/KiK-net empirical ground-motion data. The non-ergodic aleatory variability for the NGMM is reduced by 10% to 20% compared to the ergodic GMM. The NGMM does not change the average level of ground motion from the ergodic GMM; it solely modifies the spatial distribution of the non-ergodic term and the aleatory variability. The epistemic uncertainty in the dS2S in the basins from the 3-D simulations depends on the uncertainty in the 3-D velocity model. Due to the large effort to develop 3-D velocity models and the computational costs for 3-D simulations, direct estimates of the epistemic uncertainty in the non-ergodic terms due to the uncertainty in the 3-D velocity model are not available. Without the benefit of simulation results for a suite of alternative 3-D velocity models to constrain the epistemic uncertainty, the current approach is to assume that the epistemic uncertainty is half of the standard deviation of the site terms in the region (0.5 phiS2S ).
In this study, scenario ground-motion simulations from the Miyagi-Fukushima and Kanto regions (NIED, 2019, https://doi.org/10.17598/nied.0010) are used to develop a model for the spatial distribution of non-ergodic site terms (dS2S) for ground motions of M7-class crustal events. The Morikawa and Fujiwara (2013) GMM (MF13 GMM) is selected as an ergodic model. The approach outlined by Sung and Abrahamson (2022) is adopted to develop the NGMM using the 3-D simulation results: (1) the basin term of the MF13 GMM is modified to be consistent with the basin scaling from the 3-D simulations, and (2) the spatial distribution of the dS2S is estimated through the application of the varying coefficient model (VCM) within the region encompassed by the simulations. The dS2S from the simulation exhibits a stronger scaling with the basin depth as compared to the dS2S estimated by Sung et al. (2024) using the K-NET/KiK-net empirical ground-motion data. The non-ergodic aleatory variability for the NGMM is reduced by 10% to 20% compared to the ergodic GMM. The NGMM does not change the average level of ground motion from the ergodic GMM; it solely modifies the spatial distribution of the non-ergodic term and the aleatory variability. The epistemic uncertainty in the dS2S in the basins from the 3-D simulations depends on the uncertainty in the 3-D velocity model. Due to the large effort to develop 3-D velocity models and the computational costs for 3-D simulations, direct estimates of the epistemic uncertainty in the non-ergodic terms due to the uncertainty in the 3-D velocity model are not available. Without the benefit of simulation results for a suite of alternative 3-D velocity models to constrain the epistemic uncertainty, the current approach is to assume that the epistemic uncertainty is half of the standard deviation of the site terms in the region (0.5 phiS2S ).