5:15 PM - 7:15 PM
[SSS07-P03] Detection and location of seismic sources near Sofugan volcano in October 2023 using the broadband triad subarrays
Seismic sources other than earthquakes, such as landslides, glacial ice collapses, and ocean infragravity waves, are known to occur and excite various seismic waves. However, compared to the earthquake signals, these non-earthquake signals often have lower amplitudes and unclear arrivals of waves, and the conventional hypocenter determination methods may not be suitable for identifying such signals. For example, a series of events have been reported in the adjacent seas of Torishima Island around 19:00–21:00 (UTC) on October 8, 2023, and tsunami was observed from the Izu and Ogasawara Islands as well as along the coast from Chiba Prefecture to Okinawa. However, because of the low amplitudes and unclear arrivals of waves such as P-waves, the Japan Meteorological Agency (JMA) has pointed out the difficulty in identifying these events. The U.S. Geological Survey (USGS) determined a total of 15 related source locations and magnitudes. Furthermore, previous studies have reported 14 events, occurring intermittently between approximately 20:00 and 21:30. Regarding the mechanism of tsunami generation, previous studies have indicated multiple submarine deformation events near Sofugan volcano. There are discrepancies in the source locations determined by the USGS and the previous studies, as well as in the number of submarine deformation events estimated by tsunami modeling. Moreover, the seismic waveforms observed in these events are not only low in amplitude but also exhibit unclear phase arrivals and blurry onsets. Consequently, this event may have been difficult to locate using conventional methods which rely on precise phase picking, arrival times, and Earth structure and travel time models.
Here we utilized the Automated Event Location Using a Mesh of Arrays (AELUMA) method to effectively detect and locate the non-earthquake seismic sources. In this method, a seismic network is reconstructed as an assemblage of triad subarrays (triangular arrangements of three adjacent observation points) using Delaunay triangulation. By utilizing waveform correlations between pairs of observation points at both ends of each side of the triad arrays, the method detects signals originating from non-earthquake sources that are dominant in the intermediate-to-long period waveforms. We applied the developed method to 24-hour continuous seismic records on October 8, 2023 (UTC) of the F-net broadband seismic network of the National Research Institute for Earth Science and Disaster Resilience (NIED) to search for and determine source locations and analyzed the series of seismic activities that occurred near Torishima Island. Furthermore, to examine the period band effective for our detection, we testified multiple ranges of periods (10–30 s, 20–50 s, 30–60 s).
On the application date, 47 events were detected and located. Among them, we newly determined 27 events that were not listed in the earthquake catalogs, including the ones from the JMA and USGS. 3 of those 27 events occurred earlier than those documented primary events in the previous studies. The period band of the seismic waveform data used for detection was examined by testing a different set of period bands. It was found that the period band of 20–50 s, which is less violated by oceanic noises, was effective in detecting the events. In this presentation, we narrate these detection results described above, and we will also examine three events that are potentially precursory phenomena by evaluating the similarity between their seismic waveforms and those of the tsunami-triggered events. We will also perform additional detections for one month before and after the application date to discuss the relevant long-term activity in the adjacent seas of Torishima Island region.
Here we utilized the Automated Event Location Using a Mesh of Arrays (AELUMA) method to effectively detect and locate the non-earthquake seismic sources. In this method, a seismic network is reconstructed as an assemblage of triad subarrays (triangular arrangements of three adjacent observation points) using Delaunay triangulation. By utilizing waveform correlations between pairs of observation points at both ends of each side of the triad arrays, the method detects signals originating from non-earthquake sources that are dominant in the intermediate-to-long period waveforms. We applied the developed method to 24-hour continuous seismic records on October 8, 2023 (UTC) of the F-net broadband seismic network of the National Research Institute for Earth Science and Disaster Resilience (NIED) to search for and determine source locations and analyzed the series of seismic activities that occurred near Torishima Island. Furthermore, to examine the period band effective for our detection, we testified multiple ranges of periods (10–30 s, 20–50 s, 30–60 s).
On the application date, 47 events were detected and located. Among them, we newly determined 27 events that were not listed in the earthquake catalogs, including the ones from the JMA and USGS. 3 of those 27 events occurred earlier than those documented primary events in the previous studies. The period band of the seismic waveform data used for detection was examined by testing a different set of period bands. It was found that the period band of 20–50 s, which is less violated by oceanic noises, was effective in detecting the events. In this presentation, we narrate these detection results described above, and we will also examine three events that are potentially precursory phenomena by evaluating the similarity between their seismic waveforms and those of the tsunami-triggered events. We will also perform additional detections for one month before and after the application date to discuss the relevant long-term activity in the adjacent seas of Torishima Island region.