2:00 PM - 2:15 PM
[MIS11-02] Estimation of the source fault model for the 2024 Noto Peninsula earthquake based on tsunami inundation traces
Keywords:The 2024 Noto Peninsula earthquake, Numerical tsunami simulation, Tsunami trace height
The 2024 Noto Peninsula earthquake (Mw 7.5) generated tsunami inundation along the Noto Peninsula coast in Ishikawa Prefecture, Japan (JMA 2024). Field surveys documented tsunami wave heights ranging from 3 to 6 meters on the northeastern coast of the peninsula (Yuhi et al. 2024). Source fault models, proposed based on observed tsunami waveforms and crustal deformation, have provided insights on the slip distribution and tsunami propagation characteristics (e.g., Fujii and Satake 2024; GSI 2024; Masuda et al. 2024). However, there are not enough examples to validate these estimations based on the terrestrial inundation conditions. In this study, we examined tsunami sources by comparing the observed tsunami trace distribution on land with simulation results. This approach is expected to contribute to a more accurate estimation of tsunami sources by evaluating the reproducibility of onshore inundation, which has not been extensively addressed in previous studies.
The numerical simulations were conducted using JAGURS (Baba et al. 2015, 2017) with high-resolution topographic data (2/9 and 2/27 arc-seconds grids) to simulate tsunami inundation in three areas where significant inundation was observed: Horyu in Suzu City, and Uchiura and Shiromaru in Noto Town. The initial tsunami sources were based on two existing models: the fault model of Fujii and Satake (2024), which was determined by joint inversion of tsunami waveforms and crustal deformation, and the fault model of Masuda et al. (2024), which was derived by integrating the fault models of the Sea of Japan (MLIT 2014) and comparing them with tsunami waveforms. The simulation results were validated against tsunami trace height data from 76 points along the northeastern coast of the peninsula, obtained by Yuhi et al. (2024) and our surveys. The comparison between the simulation results and the tsunami trace heights was performed using high-resolution simulation results. The reproducibility of tsunami trace heights was evaluated by calculating the geometric mean K, which measures the validity of the tsunami energy content, and the geometric standard deviation κ, which reflects the accuracy of the spatial distribution of wave sources and errors in topographic data. According to JSCE (2002), the indices considered to indicate high reproducibility were 0.95 ≦ K ≦ 1.05 and κ ≦ 1.45.
For the Fujii and Satake (2024) model, simulated tsunami heights tended to be lower than the observed trace heights, with K = 1.60 and κ = 1.26. These results suggest that the Fujii and Satake (2024) model lacks sufficient energy as a wave source. In contrast, the fault model of the Masuda et al. (2024) yielded K = 0.90 and κ = 1.19, providing a relatively better explanation for the magnitude and distribution of the tsunami trace heights. However, a comparison between the crustal deformation calculated using the Okada (1985) model and the observed data (GSI 2024) revealed that the calculated values in Wajima City exceeded the observed data by more than 1 m in the positive vertical direction. To enhance model accuracy, we constructed a subfault model by dividing the aftershock distribution into 134 segments of 5 km × 5 km rectangular subfaults. A trial-and-error approach was applied to estimate a more accurate slip distribution. Near-land subfaults along the Noto Peninsula were adjusted to match observed crustal deformation. A large slip zone was set for the offshore fault northeast of the peninsula to align with observed earthquake magnitudes. The results yielded K = 1.00 and κ = 1.17, effectively explaining the magnitude and distribution of tsunami trace heights. The resulting slip distribution showed a larger slip for the offshore fault compared to the Fujii and Satake (2024) model. These results underscore the importance of utilizing field survey data, such as tsunami trace heights, when estimating tsunami sources, particularly in cases where waveform records are unavailable.
The numerical simulations were conducted using JAGURS (Baba et al. 2015, 2017) with high-resolution topographic data (2/9 and 2/27 arc-seconds grids) to simulate tsunami inundation in three areas where significant inundation was observed: Horyu in Suzu City, and Uchiura and Shiromaru in Noto Town. The initial tsunami sources were based on two existing models: the fault model of Fujii and Satake (2024), which was determined by joint inversion of tsunami waveforms and crustal deformation, and the fault model of Masuda et al. (2024), which was derived by integrating the fault models of the Sea of Japan (MLIT 2014) and comparing them with tsunami waveforms. The simulation results were validated against tsunami trace height data from 76 points along the northeastern coast of the peninsula, obtained by Yuhi et al. (2024) and our surveys. The comparison between the simulation results and the tsunami trace heights was performed using high-resolution simulation results. The reproducibility of tsunami trace heights was evaluated by calculating the geometric mean K, which measures the validity of the tsunami energy content, and the geometric standard deviation κ, which reflects the accuracy of the spatial distribution of wave sources and errors in topographic data. According to JSCE (2002), the indices considered to indicate high reproducibility were 0.95 ≦ K ≦ 1.05 and κ ≦ 1.45.
For the Fujii and Satake (2024) model, simulated tsunami heights tended to be lower than the observed trace heights, with K = 1.60 and κ = 1.26. These results suggest that the Fujii and Satake (2024) model lacks sufficient energy as a wave source. In contrast, the fault model of the Masuda et al. (2024) yielded K = 0.90 and κ = 1.19, providing a relatively better explanation for the magnitude and distribution of the tsunami trace heights. However, a comparison between the crustal deformation calculated using the Okada (1985) model and the observed data (GSI 2024) revealed that the calculated values in Wajima City exceeded the observed data by more than 1 m in the positive vertical direction. To enhance model accuracy, we constructed a subfault model by dividing the aftershock distribution into 134 segments of 5 km × 5 km rectangular subfaults. A trial-and-error approach was applied to estimate a more accurate slip distribution. Near-land subfaults along the Noto Peninsula were adjusted to match observed crustal deformation. A large slip zone was set for the offshore fault northeast of the peninsula to align with observed earthquake magnitudes. The results yielded K = 1.00 and κ = 1.17, effectively explaining the magnitude and distribution of tsunami trace heights. The resulting slip distribution showed a larger slip for the offshore fault compared to the Fujii and Satake (2024) model. These results underscore the importance of utilizing field survey data, such as tsunami trace heights, when estimating tsunami sources, particularly in cases where waveform records are unavailable.