Title: Scientist
Country/Region: U.S.A
Period: 2023/05/08-2023/8/6
Theme: Japan non-ergodic ground motion models
Host: Hiroe Miyake
Introduction: Chih-Hsuan Sung worked as an assistant project scientist in Civil and Environmental Engineering at UC Berkeley and collaborated closely with Prof. Norman Abrahamson. Her primary research is to implement and develop ground-motion models (GMMs) in Probabilistic Seismic Hazard Analysis (PSHA), focusing on the non-ergodic models for multiple regions (e.g., CA, Seattle, France, Taiwan, Turkey) based on the empirical data and the 3-D simulation. Currently, she built a new approach using the varying coefficient model (VCM) to analyze path terms related to the 3-D velocity structure from the CyberShake simulation results (Sung et al., 2023).
Her time at ERI will be spent working with Prof. Hiroe Miyake. The main goal of the research in the ERI is to build the completed process of a methodology to develop the fully non-ergodic model incorporating the empirical data from the NIED and 3-D simulations that can be used in probabilistic seismic hazard analyses.
Research Report:
Research Report for the ERI Visit from 8th May to 6th
August 2023
1 People
• Chih-Hsuan Sung: the short visit scholar at the ERI and the assistant project scientist of the
Civil and Environmental Engineering at UC Berkeley.
• Hiroe Miyake: the host and professor at the ERI of the University of Tokyo.
• Norman Abrahamson: the professor of the Civil and Environmental Engineering at UC
Berkeley.
• N. Morikawa: the senior researcher of the NIED, who provides the crustal and subduction
zone earthquakes in this research.
2 Research
At present, the ground-motion models (GMMs) for the subduction zones (interface and slab)
and crustal earthquakes of Japan were developed in the National Research Institute for Earth
Science and Disaster Resilience (NIED) (Morikawa and Fujiwara, 2013) and NGA-Sub project
1
(e.g., Si et al., 2022). These GMMs are ergodic models in that they apply to all of Japan. However,
many studies have proven that the non-ergodic effects can capture the systematic source, path,
and site effects from the ground motion scaling via the varying coefficient model and reduce the
aleatory variability (e.g., Sung et al., 2022a; Lavrentiadis et al., 2021). In this study, we capture
the non-ergodic behavior for the subduction zone and crustal earthquakes in Japan using the
NIED dataset based on the different ergodic models (Morikawa and Fujiwara, 2013; Abrahamson
et al., 2014; Si et al., 2022; Abrahamson and Gulerce, 2022; Abrahamson et al., 2016). For the
non-ergodic path effects, the cell-specific linear-distance scaling mimics the effects from a 2-D Q
structure, ϕP2PQ, (Dawood and Rodriguez-Marek, 2013; Kuehn et al., 2019) and the path effect
related to the 3-D velocity structure, ϕP2PV , (Sung et al., 2023) are all considered in our model.
The model is a smooth spatially varying non-ergodic source, site, and path terms that can be
applied to any source/site pair. For the region without the data, the adjustment terms would
be close to zero and provide higher epistemic uncertainty of the non-ergodic term. The result
shows that the full path variability (ϕP2P ) of the non-ergodic path term is about 0.3-0.4 in natural
logarithm units for all periods. Overall, with the non-ergodic terms, the fully non-ergodic models
lead to an aleatory variance of residual values for the GMMs that is reduced by 65%, which can
significantly affect seismic hazard calculations for the Japan region. In addition, our future work
will focus on the ground motion model based on the 3D numerical simulation, which will improve
the results for the shorter distance and great magnitude.
August 2023
1 People
• Chih-Hsuan Sung: the short visit scholar at the ERI and the assistant project scientist of the
Civil and Environmental Engineering at UC Berkeley.
• Hiroe Miyake: the host and professor at the ERI of the University of Tokyo.
• Norman Abrahamson: the professor of the Civil and Environmental Engineering at UC
Berkeley.
• N. Morikawa: the senior researcher of the NIED, who provides the crustal and subduction
zone earthquakes in this research.
2 Research
At present, the ground-motion models (GMMs) for the subduction zones (interface and slab)
and crustal earthquakes of Japan were developed in the National Research Institute for Earth
Science and Disaster Resilience (NIED) (Morikawa and Fujiwara, 2013) and NGA-Sub project
1
(e.g., Si et al., 2022). These GMMs are ergodic models in that they apply to all of Japan. However,
many studies have proven that the non-ergodic effects can capture the systematic source, path,
and site effects from the ground motion scaling via the varying coefficient model and reduce the
aleatory variability (e.g., Sung et al., 2022a; Lavrentiadis et al., 2021). In this study, we capture
the non-ergodic behavior for the subduction zone and crustal earthquakes in Japan using the
NIED dataset based on the different ergodic models (Morikawa and Fujiwara, 2013; Abrahamson
et al., 2014; Si et al., 2022; Abrahamson and Gulerce, 2022; Abrahamson et al., 2016). For the
non-ergodic path effects, the cell-specific linear-distance scaling mimics the effects from a 2-D Q
structure, ϕP2PQ, (Dawood and Rodriguez-Marek, 2013; Kuehn et al., 2019) and the path effect
related to the 3-D velocity structure, ϕP2PV , (Sung et al., 2023) are all considered in our model.
The model is a smooth spatially varying non-ergodic source, site, and path terms that can be
applied to any source/site pair. For the region without the data, the adjustment terms would
be close to zero and provide higher epistemic uncertainty of the non-ergodic term. The result
shows that the full path variability (ϕP2P ) of the non-ergodic path term is about 0.3-0.4 in natural
logarithm units for all periods. Overall, with the non-ergodic terms, the fully non-ergodic models
lead to an aleatory variance of residual values for the GMMs that is reduced by 65%, which can
significantly affect seismic hazard calculations for the Japan region. In addition, our future work
will focus on the ground motion model based on the 3D numerical simulation, which will improve
the results for the shorter distance and great magnitude.
View All
Fiscal Year: 2023
Fiscal Year: 2023