5:15 PM - 6:30 PM
[SSS11-P21] Source rupture process of the 2018 Hokkaido eastern Iburi earthquake estimated from waveform inversion with empirical Green’s functions
Keywords:the 2018 Hokkaido eastern Iburi earthquake, rupture process, waveforms inversion, empirical Green’s functions, strong ground motions, large pulse velocity
The Hokkaido Eastern Iburi earthquake (MJMA6.7) on September 6:03:08 (JST), 2018, occurred in the eastern Iburi region of Hokkaido, Japan, which is reported by the JMA (Japan Meteorological Agency). More than 6000 landslides occurred in the early hours after the earthquake (Yamagishi and Yamazaki, 2018). Tragically, more than 40 people lost their lives, along with 689 people injured (Fire and Disaster Management Agency, Japan, as of 5 October 2018). During this earthquake, large ground motions have been observed, e.g., ground motions at Kik-net Atsuma- (IBUH03) and K-net Mugawa (HKD126), the peak velocity is approximately 150cm/s, and ground motions in the frequency components 1~3Hz are predominant, which are very important to the engineering facilities. In general, the strong crustal earthquakes are with 5-20km depth, e.g., the mainshock of the 2016 Kumamoto earthquake (Mw7.1, depth 12km), the 1995 Kobe (Mw6.9 depth 16km); the 1989 Loma Prieta (Mw6.9, depth 19 km), the 1994 Northridge (Mw6.7, depth 18.2 km) earthquake. However, the 2018 Hokkaido eastern Iburi earthquake was in contrast to the usual situations; it is with Mj6.7 but with hypocentral depth of 37 km but caused large ground motions.
Such special characteristics, the large strong ground motions it generated, its impact on lifeline facilities and buildings, and the serious geological hazards it caused have motivated us to study its rupture features, especially focusing on the generation mechanism of the strong ground motions (e.g., the large pulse velocity). This event was well recorded by permanent strong-motion networks K-net and KiK-net, forming a rich waveform database, which enables us to investigate the rupture process of the earthquake in detail. Therefore, in this analysis, the source rupture process of this event was estimated from the waveform inversion of strong-motion data, and with the adoption of empirical Green’s functions, which could avoid the possible uncertainties caused by the assumption of subsurface velocity structures, as well as to consider relatively higher frequency ranges of strong ground motions typically up to 2.0 Hz.
Our preferred slip model is illustrated in Figure 1, and the comparisons of observed and synthesized velocity waveforms (0.2 - 2.0 Hz) are illustrated in Figure 2. Figure 1(a) indicates that one large slip region with a maximum slip of approximately 3.4 m was centered ~ 2 km SE of the hypocenter; and in Figure 1(b), two regions with large peak slip velocity were identified, the first one with a peak value of 2.0 m/s was located ~ 6 km SE of the hypocenter, which was mainly contributed to the large pulse velocity of the K-net station HKD126, and the region with serious landslides; and the second one was centered ~ 5 km NW of the hypocenter. The large slip region (with 23~28km in depth) and both the regions with peak slip velocity (with 21~26 km in depth) are located shallower than the hypocenter. The rupture velocity of 2.0 km/s was identified, and the preferred source model corresponds to Mw 7.0 for this event.
The robustness of our preferred slip model was verified through different combinations of EGF events and near-fault stations, which resulted that the locations of the large slip regions are not significantly affected by the selection of the EGF events and the near-fault station. Additionally, based on the preferred source model, the ground motions are synthesized for strong-motion stations that were not used in the inversion analyses, the synthesized ground motions are consistent with the observed ones. Furthermore, our preferred slip model is consistent very well with the broadband source model proposed by Satoh (2020).
Such special characteristics, the large strong ground motions it generated, its impact on lifeline facilities and buildings, and the serious geological hazards it caused have motivated us to study its rupture features, especially focusing on the generation mechanism of the strong ground motions (e.g., the large pulse velocity). This event was well recorded by permanent strong-motion networks K-net and KiK-net, forming a rich waveform database, which enables us to investigate the rupture process of the earthquake in detail. Therefore, in this analysis, the source rupture process of this event was estimated from the waveform inversion of strong-motion data, and with the adoption of empirical Green’s functions, which could avoid the possible uncertainties caused by the assumption of subsurface velocity structures, as well as to consider relatively higher frequency ranges of strong ground motions typically up to 2.0 Hz.
Our preferred slip model is illustrated in Figure 1, and the comparisons of observed and synthesized velocity waveforms (0.2 - 2.0 Hz) are illustrated in Figure 2. Figure 1(a) indicates that one large slip region with a maximum slip of approximately 3.4 m was centered ~ 2 km SE of the hypocenter; and in Figure 1(b), two regions with large peak slip velocity were identified, the first one with a peak value of 2.0 m/s was located ~ 6 km SE of the hypocenter, which was mainly contributed to the large pulse velocity of the K-net station HKD126, and the region with serious landslides; and the second one was centered ~ 5 km NW of the hypocenter. The large slip region (with 23~28km in depth) and both the regions with peak slip velocity (with 21~26 km in depth) are located shallower than the hypocenter. The rupture velocity of 2.0 km/s was identified, and the preferred source model corresponds to Mw 7.0 for this event.
The robustness of our preferred slip model was verified through different combinations of EGF events and near-fault stations, which resulted that the locations of the large slip regions are not significantly affected by the selection of the EGF events and the near-fault station. Additionally, based on the preferred source model, the ground motions are synthesized for strong-motion stations that were not used in the inversion analyses, the synthesized ground motions are consistent with the observed ones. Furthermore, our preferred slip model is consistent very well with the broadband source model proposed by Satoh (2020).