10:00 〜 10:15
[SCG45-15] A comprehensive earthquake catalog for the 2020 seismic swarm in the central Japan
From the middle of April 2020, a seismic swarm initiated near the boundary between the Nagano and Gifu Prefectures, in the central Japan. The seismicity had been particularly intensive from the end of April to May 2020, including the largest magnitude event of 5.5. Most of major events show focal mechanism solutions of strike-slip faulting. During this activity, the seismicity extended toward the north in the stepwise manner. In 1998, a similar seismic swarm occurred in the same area (Aoyama et al., 2002). To reveal the spatio-temporal evolution of this swarm sequence, we present the more comprehensive earthquake catalog in this study.
We used ~13 months continuous three-component velocity seismograms recorded at 18 seismic stations in and around the target region, starting from 30 March 2020. To pick P- and S-wave arrival times from the continuous waveforms, we applied a deep-neural network-based picker, PhaseNet, developed by Zhu and Beroza (2019). We then associated the phase picks with individual events using the Rapid Earthquake Association and Location package (Zhang et al., 2019). The earthquakes with a sufficient number of arrival times (i.e., the numbers of P- and S-wave arrivals must be greater than 4 and 2, respectively) were then selected for the relocation. We applied a Hierarchical Clustering Algorithm for Relative Earthquake Relocation (Trugman and Shearer, 2017) to differential times constructed by waveform cross-correlation method and relocated around 48000 events. Then, we used these relocated hypocenters as template events and searched the continuous records for events that strongly resembled these template events, by applying a matched filter technique accelerated by Tensor Cores with TensorFloat-32 precision on GPU (Yamaguchi et al., 2019). Our comprehensive catalog provides an opportunity to extract the geometry of complex fault structures that were activated during the swarm. The swarm area extends north-south with a length of about 30 km, and many vertically dipping faults with north-south or east-west trends are clearly observed. During some episodes, diffusion-like migrations of hypocenters can be identified during the seismic swarm, implying fluid involvement.
We used ~13 months continuous three-component velocity seismograms recorded at 18 seismic stations in and around the target region, starting from 30 March 2020. To pick P- and S-wave arrival times from the continuous waveforms, we applied a deep-neural network-based picker, PhaseNet, developed by Zhu and Beroza (2019). We then associated the phase picks with individual events using the Rapid Earthquake Association and Location package (Zhang et al., 2019). The earthquakes with a sufficient number of arrival times (i.e., the numbers of P- and S-wave arrivals must be greater than 4 and 2, respectively) were then selected for the relocation. We applied a Hierarchical Clustering Algorithm for Relative Earthquake Relocation (Trugman and Shearer, 2017) to differential times constructed by waveform cross-correlation method and relocated around 48000 events. Then, we used these relocated hypocenters as template events and searched the continuous records for events that strongly resembled these template events, by applying a matched filter technique accelerated by Tensor Cores with TensorFloat-32 precision on GPU (Yamaguchi et al., 2019). Our comprehensive catalog provides an opportunity to extract the geometry of complex fault structures that were activated during the swarm. The swarm area extends north-south with a length of about 30 km, and many vertically dipping faults with north-south or east-west trends are clearly observed. During some episodes, diffusion-like migrations of hypocenters can be identified during the seismic swarm, implying fluid involvement.