10:00 〜 10:15
[SCG39-23] 広帯域海底地震観測で捉えた北海道襟裳沖に沈み込んだ海山周辺の地震活動
キーワード：千島・日本海溝会合部で起こる浅部VLFEと微動、広帯域海底地震計アレイ観測、沈み込んだ海山
Many shallow very-low-frequency earthquakes (sVLFE) have been observed in the Kuril and Japan trench-trench junction, southern off Erimo, Hokkaido, by the inland F-net (Asano et al., 2008). In this region, many tectonic tremors have been also discovered by the S-net after 2016 (Tanaka et al., 2019; Nishikawa et al., 2019) and the long-term offline ocean bottom seismometer (OBS) network between 2006 and 2007 (Kawakubo et al., 2020). These studies reported simultaneous occasion of sVLFE and tremors. Kawakubo et al. additionally indicated that intervals of activities and epicenters distribution are roughly consistent with that before and the 2011 Tohoku-oki earthquake. However, there is no broad-band observatory above sVLFEs, and known hypocenters only by F-net are less accurate. Therefore, a spatial comparison of sVLFEs with tremors and regular earthquakes is difficult. In addition, because a spatial relation between a seamount subduction from the junction (Tsuru et al., 2005; Makino, 2005) and sVLFE occasion is expected, a near-field observatory providing precise spatio-temporal distribution of earthquakes is essential for understanding a state of the slow earthquake activities. From above backgrounds, we aim to investigate a seismicity of slow and regular earthquakes by deploying a near-field broad-band OBSs array.
This study uses seismograms recorded by one broad-band (120 seconds) OBS and four shot-period (1 Hz) OBSs deployed around the subducted seamount during July 6th, 2019 to October 8th, 2020. The deployment and recovery of OBSs were operated by the R/V Shinsei-maru (KS19-12 and KS20-16 cruises), the Japan Agency for Marin-Earth Science and Technology, and the R/V Ryoufu-maru (RF20-07), the Japan Meteorological Agency. Seismograms were recorded by a recorder featuring a 24-bit A/D converter with a 200 Hz sampling rate.
To investigate sVLFE, we firstly search tremors occurring simultaneously with sVLFEs by an envelope correlation method (Ide, 2010) which calculates correlation between station pairs and estimates hypocenters by using residuals of arrival times between pairs with maximum correlation coefficient over an optional value. We calculated envelopes from three components of seismograms between 2–8 Hz, took the squared sum of components, low-pass filtered below 0.5 Hz, and finally resampled at 1 Hz. We then took cross correlation for each 120 s window with 60 s overlap and adopted events with a maximum correlation coefficient over 0.6 at all pairs of stations. Finally, we extracted reliable events in accordance with a hypocenter error < 5 km, and arrival time residual < 3 s.
As a preliminary result, we obtained 378 events satisfying thresholds for errors, and 176 of them have a duration over 20 seconds, identified as tremors referring to Ohta et al. (2019). Followings are of interest in our result.
(1) A time series of all events projected along the seismic survey line (Tsuru et al., 2005; Makino et al., 2005) shows a highly activating period accompanied by long duration events on middle February, 2020, so that simultaneous sVLFE occasion is expected. We will process a waveform analysis for this period as a first step of sVLFE investigation.
(2) Most of determined events locate at the west half of the array, that is down-dip side of the subducted seamount. This may indicate a stress accumulation by the forearc–seamount collision.
(3) The resultant epicenter distribution seems to be isolated from that by previous works (Nishikawa et al., 2019; Kawakubo et al. 2020). Such a state of the seismic gap would be apparent related with a limit of detectable area of S-net and our arrays. Therefore, to discuss the isolation, the analysis with S-net data to expand the detectable area would be effective.
As future tasks, careful investigation of accuracy of hypocenter and duration, and confirmation of events on seismograms would bring an understanding of the state of seismic activity and the relationship with the subducted seamount.
This study uses seismograms recorded by one broad-band (120 seconds) OBS and four shot-period (1 Hz) OBSs deployed around the subducted seamount during July 6th, 2019 to October 8th, 2020. The deployment and recovery of OBSs were operated by the R/V Shinsei-maru (KS19-12 and KS20-16 cruises), the Japan Agency for Marin-Earth Science and Technology, and the R/V Ryoufu-maru (RF20-07), the Japan Meteorological Agency. Seismograms were recorded by a recorder featuring a 24-bit A/D converter with a 200 Hz sampling rate.
To investigate sVLFE, we firstly search tremors occurring simultaneously with sVLFEs by an envelope correlation method (Ide, 2010) which calculates correlation between station pairs and estimates hypocenters by using residuals of arrival times between pairs with maximum correlation coefficient over an optional value. We calculated envelopes from three components of seismograms between 2–8 Hz, took the squared sum of components, low-pass filtered below 0.5 Hz, and finally resampled at 1 Hz. We then took cross correlation for each 120 s window with 60 s overlap and adopted events with a maximum correlation coefficient over 0.6 at all pairs of stations. Finally, we extracted reliable events in accordance with a hypocenter error < 5 km, and arrival time residual < 3 s.
As a preliminary result, we obtained 378 events satisfying thresholds for errors, and 176 of them have a duration over 20 seconds, identified as tremors referring to Ohta et al. (2019). Followings are of interest in our result.
(1) A time series of all events projected along the seismic survey line (Tsuru et al., 2005; Makino et al., 2005) shows a highly activating period accompanied by long duration events on middle February, 2020, so that simultaneous sVLFE occasion is expected. We will process a waveform analysis for this period as a first step of sVLFE investigation.
(2) Most of determined events locate at the west half of the array, that is down-dip side of the subducted seamount. This may indicate a stress accumulation by the forearc–seamount collision.
(3) The resultant epicenter distribution seems to be isolated from that by previous works (Nishikawa et al., 2019; Kawakubo et al. 2020). Such a state of the seismic gap would be apparent related with a limit of detectable area of S-net and our arrays. Therefore, to discuss the isolation, the analysis with S-net data to expand the detectable area would be effective.
As future tasks, careful investigation of accuracy of hypocenter and duration, and confirmation of events on seismograms would bring an understanding of the state of seismic activity and the relationship with the subducted seamount.