5:15 PM - 6:45 PM
[U15-P34] Source process on variable-dip faults of the 2024 Noto Peninsula earthquake (MJ 7.6) inverted from strong-motion records
Keywords:2024 Noto Peninsula earthquake, Source inversion, Curved fault, Near-field strong-motion waveform, Three-stage source scaling model, Aftershock relocation
The Noto Peninsula earthquake (MJ 7.6), which occurred on January 1, 2024, was an inland crustal earthquake on nearshore reverse faults with a total length of ~150 km. It is a third-stage event in the three-stage source scaling model for inland crustal earthquakes (Irikura and Miyake, 2011). We inverted near-field strong-motion waveforms (0.05-0.25 Hz) recorded by the observation networks of the National Research Institute for Earth Science and Disaster Resilience (K-NET, KiK-net, and F-net) and the Japan Meteorological Agency (JMA) to estimate the source rupture process of this earthquake.
To define the fault planes used for source inversion, we relocated the hypocenters of the foreshock (at 16:10:09 on January 1), mainshock (at 16:10:22), and aftershocks (M >1) within one week after the mainshock using arrival time data picked by JMA and the programs developed by Hirata and Matsu'ura (1987) and Waldhauser and Ellsworth (2000). Most of the relocated aftershocks are located at depths of 5-15 km, and there are variations in hypocenter depth along the fault strike direction. Furthermore, the alignment of aftershocks dips southeastward with low angles, indicating the necessity for high-angle fault planes at shallow depths to link the low-angle aftershock distribution with seafloor fault traces (Inoue and Okamura, 2010). Based on the relocated aftershock distribution, we set five fault planes with depths slightly differing from each other, all of which dip southeastward and transition from low angles (20-25°) to high angles (65°). The total length of these five segments measures 144 km. Each segment has a width of 22.5 km.
Our inversion employed the multi-time-window linear waveform inversion method (Hartzell and Heaton, 1983) and treated the foreshock and mainshock as a single continuous event. To explain the wave packets arriving with time lags observed in Suzu City (stations ISK001 and ISKH01), we assumed that the fault segment where the rupture initiation point is located could rupture twice. The first rupture was assumed to initiate at 16:10:09, and the time lag between the first and second rupture initiations was determined based on waveform residuals. It was also assumed that only the second rupture could propagate to the other segments. We used 1D layered velocity structure models, which were extracted from the 3D Japan Integrated Velocity Structure Model Version 1 (Koketsu et al., 2012), and calculated the Green’s functions for each subfault using the discrete wavenumber method (Bouchon 1981) and the reflection/transmission coefficient matrix method (Kennett and Kerry 1979). The velocity structure model for station ISK002 was iteratively adjusted using the waveform record from a small earthquake. Additionally, velocity structure models for the Sado and Joetsu regions of Niigata Prefecture took into account the influence of thick sediment layers in the offshore area.
The preliminary inversion estimated the total seismic moment to be 3.89×1020 Nm (MW 7.66). The rupture velocity triggering the first time window was 1.8 km/s, which was not as fast as those of strike-slip third-stage events, such as the 2002 Alaska, Denali, earthquake (MW 7.9) and the 2023 Turkey-Syria earthquake sequences (MW 7.8 and MW 7.5). The average slip was 4.1 m, slightly larger than the average slip of 3.3 m proposed for third-stage events by Murotani et al. (2015). The slips responsible for the two wave packets observed at stations ISK001 and ISKH01 were found at different locations. The entire source fault has three large-slip (>6 m) areas, located west of Wajima City, near the midpoint between Wajima and Suzu, and in the offshore area near longitude 137.5°. Among these, the large-slip area in the offshore area is unlikely to reach the seafloor. However, for this earthquake, the estimated slip distribution in the offshore area significantly depends on the velocity structure model used. Therefore, it is important to further validate the offshore slip.
To define the fault planes used for source inversion, we relocated the hypocenters of the foreshock (at 16:10:09 on January 1), mainshock (at 16:10:22), and aftershocks (M >1) within one week after the mainshock using arrival time data picked by JMA and the programs developed by Hirata and Matsu'ura (1987) and Waldhauser and Ellsworth (2000). Most of the relocated aftershocks are located at depths of 5-15 km, and there are variations in hypocenter depth along the fault strike direction. Furthermore, the alignment of aftershocks dips southeastward with low angles, indicating the necessity for high-angle fault planes at shallow depths to link the low-angle aftershock distribution with seafloor fault traces (Inoue and Okamura, 2010). Based on the relocated aftershock distribution, we set five fault planes with depths slightly differing from each other, all of which dip southeastward and transition from low angles (20-25°) to high angles (65°). The total length of these five segments measures 144 km. Each segment has a width of 22.5 km.
Our inversion employed the multi-time-window linear waveform inversion method (Hartzell and Heaton, 1983) and treated the foreshock and mainshock as a single continuous event. To explain the wave packets arriving with time lags observed in Suzu City (stations ISK001 and ISKH01), we assumed that the fault segment where the rupture initiation point is located could rupture twice. The first rupture was assumed to initiate at 16:10:09, and the time lag between the first and second rupture initiations was determined based on waveform residuals. It was also assumed that only the second rupture could propagate to the other segments. We used 1D layered velocity structure models, which were extracted from the 3D Japan Integrated Velocity Structure Model Version 1 (Koketsu et al., 2012), and calculated the Green’s functions for each subfault using the discrete wavenumber method (Bouchon 1981) and the reflection/transmission coefficient matrix method (Kennett and Kerry 1979). The velocity structure model for station ISK002 was iteratively adjusted using the waveform record from a small earthquake. Additionally, velocity structure models for the Sado and Joetsu regions of Niigata Prefecture took into account the influence of thick sediment layers in the offshore area.
The preliminary inversion estimated the total seismic moment to be 3.89×1020 Nm (MW 7.66). The rupture velocity triggering the first time window was 1.8 km/s, which was not as fast as those of strike-slip third-stage events, such as the 2002 Alaska, Denali, earthquake (MW 7.9) and the 2023 Turkey-Syria earthquake sequences (MW 7.8 and MW 7.5). The average slip was 4.1 m, slightly larger than the average slip of 3.3 m proposed for third-stage events by Murotani et al. (2015). The slips responsible for the two wave packets observed at stations ISK001 and ISKH01 were found at different locations. The entire source fault has three large-slip (>6 m) areas, located west of Wajima City, near the midpoint between Wajima and Suzu, and in the offshore area near longitude 137.5°. Among these, the large-slip area in the offshore area is unlikely to reach the seafloor. However, for this earthquake, the estimated slip distribution in the offshore area significantly depends on the velocity structure model used. Therefore, it is important to further validate the offshore slip.