5:15 PM - 6:45 PM
[U15-P31] Source inversion using empirical Green’s functions for the 2024 Noto Peninsula earthquake
Keywords:The 2024 Noto Peninsula earthquake, Source inversion analysis, Empirical Green's function, Strong ground motion
The Noto Peninsula earthquake (Mj 7.6) that occurred on January 1, 2024 was the second largest event in Japan since the 1891 Nobi earthquake (M8.0), and several stations in the northern part of Noto Peninsula recorded a seismic intensity of 7 on the JMA scale. In the northern Noto Peninsula, locally active seismic swarms have continued for the past several years. In this earthquake, the length of the aftershock zone reached approximately 150 km, far exceeding the area of the earthquake swarm. In this study we estimated the source slip distribution model by applying the source inversion method using the empirical Green’s function to the strong-motion observation records around the epicenter.
A fault plane model of the main shock was constructed from the aftershock distribution for 24 hours after the main shock. We assumed a total of four fault planes, referring to changes in the trend of the strike direction of the aftershock distribution and existing submarine fault traces. Of these, only the easternmost fault plane was assumed to be a northwest dipping reverse fault that is conjugate with others, and the remaining three faults show opposite southeast dipping. The westernmost fault plane is modeled to have a strike along the Ama-Misaki-Oki fault zone, which has a different strike angle from the Sasanami-Oki fault zone, identified as the source of the 2007 Noto Peninsula earthquake (Mj 6.9). A foreshock of Mj 5.9 occurred just before the main shock near the hypocenter, and the strong-motion records near the source region clearly show a moderate wave packet at a time earlier than the arrival of S wave from the origin time of the main shock. On the other hand, however, these wave packets show a significant time delay compared to the arrival of S wave of the foreshock’s rupture. Based on such observations, a separate foreshock fault plane with different depth and dip angle was set at the epicentral area of the main shock, and the rupture was initiated at the origin time of the foreshock. For other fault planes, inversion analysis was conducted under conditions that allows a large time delay including the origin time of the main shock, while ensuring the causal low of the rupture propagation. For the empirical Green’s function, we employed four small earthquake records, ranging from Mj 4.9 to 5.3, for each fault plane. For the analysis we used two horizontal components of the velocity waves observed at 24 K-NET and KiK-net stations in the 0.1-1Hz frequency band, confirming no significant nonlinear ground amplification effects. Inversion process is performed using the fast simulated annealing optimization method by Shiba and Irikura (2005).
The obtained slip distribution model is shown in the figure. The seismic moment is estimated to be 2.01E20 Nm with Mw 7.5. The largest slip reaches 7.0 m. The slip on the northeastern conjugate fault and the southern part of the Ama-Misaki-Oki segment is relatively small and is not strongly contributed to the strong-motion generation. In terms of the rupture time distribution, rupture on the fault plane having the main shock hypocenter starts close to the origin time of the main shock itself, about 13 seconds after the foreshock occurrence, while the rupture on the western adjacent fault starts before the rupture front of the main-shock fault arrives, as it propagated from the foreshock fault. Therefore, the motion on the western fault can be interpreted as a part of the foreshock. On the other hand, there is a time delay of about 10 seconds after the foreshock rupture ends, which is considered as a fluctuation in the rupture propagation during the optimization. The calculated waveforms from the inverted source model agree well with the observation records at many stations, but the wave packets observed at earlier timings for stations near the faults cannot be reproduced adequately. So there remains issues for improvement in the modeling of rupture propagation.
A fault plane model of the main shock was constructed from the aftershock distribution for 24 hours after the main shock. We assumed a total of four fault planes, referring to changes in the trend of the strike direction of the aftershock distribution and existing submarine fault traces. Of these, only the easternmost fault plane was assumed to be a northwest dipping reverse fault that is conjugate with others, and the remaining three faults show opposite southeast dipping. The westernmost fault plane is modeled to have a strike along the Ama-Misaki-Oki fault zone, which has a different strike angle from the Sasanami-Oki fault zone, identified as the source of the 2007 Noto Peninsula earthquake (Mj 6.9). A foreshock of Mj 5.9 occurred just before the main shock near the hypocenter, and the strong-motion records near the source region clearly show a moderate wave packet at a time earlier than the arrival of S wave from the origin time of the main shock. On the other hand, however, these wave packets show a significant time delay compared to the arrival of S wave of the foreshock’s rupture. Based on such observations, a separate foreshock fault plane with different depth and dip angle was set at the epicentral area of the main shock, and the rupture was initiated at the origin time of the foreshock. For other fault planes, inversion analysis was conducted under conditions that allows a large time delay including the origin time of the main shock, while ensuring the causal low of the rupture propagation. For the empirical Green’s function, we employed four small earthquake records, ranging from Mj 4.9 to 5.3, for each fault plane. For the analysis we used two horizontal components of the velocity waves observed at 24 K-NET and KiK-net stations in the 0.1-1Hz frequency band, confirming no significant nonlinear ground amplification effects. Inversion process is performed using the fast simulated annealing optimization method by Shiba and Irikura (2005).
The obtained slip distribution model is shown in the figure. The seismic moment is estimated to be 2.01E20 Nm with Mw 7.5. The largest slip reaches 7.0 m. The slip on the northeastern conjugate fault and the southern part of the Ama-Misaki-Oki segment is relatively small and is not strongly contributed to the strong-motion generation. In terms of the rupture time distribution, rupture on the fault plane having the main shock hypocenter starts close to the origin time of the main shock itself, about 13 seconds after the foreshock occurrence, while the rupture on the western adjacent fault starts before the rupture front of the main-shock fault arrives, as it propagated from the foreshock fault. Therefore, the motion on the western fault can be interpreted as a part of the foreshock. On the other hand, there is a time delay of about 10 seconds after the foreshock rupture ends, which is considered as a fluctuation in the rupture propagation during the optimization. The calculated waveforms from the inverted source model agree well with the observation records at many stations, but the wave packets observed at earlier timings for stations near the faults cannot be reproduced adequately. So there remains issues for improvement in the modeling of rupture propagation.