9:30 AM - 9:45 AM
[SSS11-03] Changing features revealed from the non-stationary ETAS model in regional earthquake swarm activity of the Noto Peninsula
Keywords:Earthquake swarm, background seismicity, nonstationary modeling
Various statistics can be extracted from seismic time series catalog, but the background seismicity is particularly important from the viewpoint of prediction. In our research, we used a point process model that estimates the time variation of the background seismicity of the ETAS model by the Bayesian method, and applied it to the earthquake swarm events of the Noto Peninsula in Ishikawa Prefecture that occurred from the end of 2020 to show the characteristics of the events.
In the Noto Peninsula, the occurrence of earthquakes has been increasing since around 2018, and seismic activity has been active since the end of November 2020 to the present. These activities were roughly divided into four epicenter clusters. In the southeastern region where seismic activity has been observed before 2018, active region has moved to a depth of 14 km or deeper at the end of November 2020. Since then, it has been gradually activated in the other three clusters. The largest earthquake as of November 6, 2021 was M J 5.1, which occurred in the northeastern region on September 16, 2021. In GNSS observations, unsteady changes were observed from around the end of November 2020, and the baseline distance between Wajima 2-Juzhou observation point and Hegurajima-Hegurajima observation point increased by about 1 cm, and Hegurajima of Zhuzhou observation point.
The non-stationary ETAS model (Kumazawa and Ogata, 2013) was applied to these clusters. This model assumes that the background intensity and aftershock intensity parameters of the conventional ETAS model are time-dependent functions, and these functions are estimated by the Bayesian method. The background intensity is considered to be due to slow stress changes in the entire target area and fault weakening due to fluid intrusion. The aftershock intensity mainly indicates the change in the chain effect (aftershock induction rate) due to the preceding earthquake in the area.
The activity in the southeastern region was initially distributed near the center of the region, but it subsided when the activity moved to the deeper part on November 30, 2020. At the same time, in the deeper region, the seismicity becomes active like a donut so as to surround the shallow earthquake swarm, and the background intensity has been increasing until now. After the activation in the deep southeastern region, the activity in the western region increased first, followed by the seismic activity in the northern and northeastern regions, two to three months later. The background intensities in these regions are relatively compatible with the changing trends of the GNSS observation time series, which seems to reflect the crustal changes in each region.
In the Noto Peninsula, the occurrence of earthquakes has been increasing since around 2018, and seismic activity has been active since the end of November 2020 to the present. These activities were roughly divided into four epicenter clusters. In the southeastern region where seismic activity has been observed before 2018, active region has moved to a depth of 14 km or deeper at the end of November 2020. Since then, it has been gradually activated in the other three clusters. The largest earthquake as of November 6, 2021 was M J 5.1, which occurred in the northeastern region on September 16, 2021. In GNSS observations, unsteady changes were observed from around the end of November 2020, and the baseline distance between Wajima 2-Juzhou observation point and Hegurajima-Hegurajima observation point increased by about 1 cm, and Hegurajima of Zhuzhou observation point.
The non-stationary ETAS model (Kumazawa and Ogata, 2013) was applied to these clusters. This model assumes that the background intensity and aftershock intensity parameters of the conventional ETAS model are time-dependent functions, and these functions are estimated by the Bayesian method. The background intensity is considered to be due to slow stress changes in the entire target area and fault weakening due to fluid intrusion. The aftershock intensity mainly indicates the change in the chain effect (aftershock induction rate) due to the preceding earthquake in the area.
The activity in the southeastern region was initially distributed near the center of the region, but it subsided when the activity moved to the deeper part on November 30, 2020. At the same time, in the deeper region, the seismicity becomes active like a donut so as to surround the shallow earthquake swarm, and the background intensity has been increasing until now. After the activation in the deep southeastern region, the activity in the western region increased first, followed by the seismic activity in the northern and northeastern regions, two to three months later. The background intensities in these regions are relatively compatible with the changing trends of the GNSS observation time series, which seems to reflect the crustal changes in each region.