5:15 PM - 6:45 PM
[SSS08-P11] Spatio-Temporal Variation of Okinawa Trough Earthquake swarms Extracted by Matched Filter Method
Keywords:Okinawa Trough, earthquake swarm, b value, tectonics, dike intrusion
The Okinawa Trough is an active back-arc basin located northwest of the Ryukyu Arc, where earthquake swarms frequently occur. The earthquake swarm activity in the Okinawa Trough is concentrated around the central axis of the trough, and the pattern of the earthquake swarm activity differs between the southern and central parts of the Okinawa Trough. In the south part, the seismic swarm areas tended to expand in an east-west direction with time. However, such phenomena have rarely been observed in the central region. In addition, the areas of particularly active seismic swarms have tended to be concentrated in the same areas over the past 20 years. In most cases, activity ceases within a few days. However, there are cases in which the activity continues for more than a week or even nearly a year.
More seismic activity must be detected to view changes in seismic swarm activity. However, when earthquake swarms occur in the Okinawa Trough, the detection capability temporarily decreases (Nakamura, 2022). Therefore, we used a matched filter method to detect events whose epicenters had not been determined when the earthquake swarm occurred. We used seismic data from the JMA catalog as a template to detect events of major earthquake swarms that occurred around the rift axis of the Okinawa Trough between 2002 and 2022. We calculated the correlation between the template earthquake's S-wave and the horizontal component's continuous waveform at several stations near the earthquake swarm. We used ±2.0s of S-wave reading time for the correlation. The waveforms were bandpass filtered at 2.0 - 4.0 Hz. The detection threshold was set to 13 times the absolute standard deviation. We employed the Rivalta (2010) equation to determine the time constant and final migration distance of seismic activity if migration was observed. In this model, the seismic swarm activity area expands exponentially owing to dyke intrusion.
The analysis showed that all cluster earthquakes, including the events listed in the JMA catalog, were detected about 2-3 times larger than those in the JMA catalog. Even in the northern part of Ishigaki Island in 2002 and near Yonaguni in 2013, where the completeness magnitude (Mc) temporarily increased, the Mc values dropped to 3.2 and 2.0, respectively, improving the detection capability. In the 2013 Yonaguni and 2014 Amami-Oshima seismic swarms, the seismic area expanded significantly with time. We calculated the time constant and final migration distance for the starting point of the earthquake swarm using the equation of Rivalta (2010). As a result, in the 2013 Yonaguni earthquake swarm, seismic activity spread over a length of approximately 26±0.3 km in the northeast-southwest direction. The migration velocities in the northeast and southwest directions were about 0.55 km/h and 1.54 km/h, respectively. The time constants were about 6.5 ± 0.5 h and 5.5 ± 1.0 h, respectively. In the 2014 earthquake swarm near Amami-Oshima Island, seismic activity extended southeastward over a length of approximately 10 ± 0.2 km. The migration velocity and time constant of the southeastward movement were about 0.60 km/h and 5±0.5 h, respectively.
This is similar to the migration distances and velocities observed in the 1978 seismic swarm in Iceland (33.1 ±2.3 km and 12.3 ±2.3 h) and in the two swarms in Afar (1st: 8.0 ±1.0 km and 1.5 ±0.5 h; 2nd: 10.9 ±1.3 km and 3.2 ±1.2 h). However, assuming that the movement of earthquake swarms is fluid diffusion, the diffusion coefficient is 807.2 using the formula of Shapiro et al. (1997).
This value is larger than the diffusion coefficient obtained for a nonvolcanic earthquake swarm. This suggests that the seismic swarm activity in the central and southern Okinawa Trough may have been caused by magma movement due to dike intrusion. The fact that crustal deformation (equivalent to Mw 6.9 in fault movement) was larger than the seismic moment (Mw 5.4 for the largest earthquake) released by swarm earthquakes (Nakamura and Kinjo, 2018; Tu and Heki, 2017) also suggests the influence of magma intrusion.
More seismic activity must be detected to view changes in seismic swarm activity. However, when earthquake swarms occur in the Okinawa Trough, the detection capability temporarily decreases (Nakamura, 2022). Therefore, we used a matched filter method to detect events whose epicenters had not been determined when the earthquake swarm occurred. We used seismic data from the JMA catalog as a template to detect events of major earthquake swarms that occurred around the rift axis of the Okinawa Trough between 2002 and 2022. We calculated the correlation between the template earthquake's S-wave and the horizontal component's continuous waveform at several stations near the earthquake swarm. We used ±2.0s of S-wave reading time for the correlation. The waveforms were bandpass filtered at 2.0 - 4.0 Hz. The detection threshold was set to 13 times the absolute standard deviation. We employed the Rivalta (2010) equation to determine the time constant and final migration distance of seismic activity if migration was observed. In this model, the seismic swarm activity area expands exponentially owing to dyke intrusion.
The analysis showed that all cluster earthquakes, including the events listed in the JMA catalog, were detected about 2-3 times larger than those in the JMA catalog. Even in the northern part of Ishigaki Island in 2002 and near Yonaguni in 2013, where the completeness magnitude (Mc) temporarily increased, the Mc values dropped to 3.2 and 2.0, respectively, improving the detection capability. In the 2013 Yonaguni and 2014 Amami-Oshima seismic swarms, the seismic area expanded significantly with time. We calculated the time constant and final migration distance for the starting point of the earthquake swarm using the equation of Rivalta (2010). As a result, in the 2013 Yonaguni earthquake swarm, seismic activity spread over a length of approximately 26±0.3 km in the northeast-southwest direction. The migration velocities in the northeast and southwest directions were about 0.55 km/h and 1.54 km/h, respectively. The time constants were about 6.5 ± 0.5 h and 5.5 ± 1.0 h, respectively. In the 2014 earthquake swarm near Amami-Oshima Island, seismic activity extended southeastward over a length of approximately 10 ± 0.2 km. The migration velocity and time constant of the southeastward movement were about 0.60 km/h and 5±0.5 h, respectively.
This is similar to the migration distances and velocities observed in the 1978 seismic swarm in Iceland (33.1 ±2.3 km and 12.3 ±2.3 h) and in the two swarms in Afar (1st: 8.0 ±1.0 km and 1.5 ±0.5 h; 2nd: 10.9 ±1.3 km and 3.2 ±1.2 h). However, assuming that the movement of earthquake swarms is fluid diffusion, the diffusion coefficient is 807.2 using the formula of Shapiro et al. (1997).
This value is larger than the diffusion coefficient obtained for a nonvolcanic earthquake swarm. This suggests that the seismic swarm activity in the central and southern Okinawa Trough may have been caused by magma movement due to dike intrusion. The fact that crustal deformation (equivalent to Mw 6.9 in fault movement) was larger than the seismic moment (Mw 5.4 for the largest earthquake) released by swarm earthquakes (Nakamura and Kinjo, 2018; Tu and Heki, 2017) also suggests the influence of magma intrusion.