5:15 PM - 6:30 PM
[SCG39-P21] Activity of low-frequency tremors and regular earthquakes off Erimo, Hokkaido from 2006 to 2007, by a tremor investigation based on waveform information
Keywords:tremor, Off Erimo, Hokkaido, Slow earthquakes
Slow earthquakes implying slow fault slip have been discovered since 1990s, owing to development of seismological and geodetic observations. Low-frequency tremors, one of slow phenomena, excite seismic energy in the frequency band of 2–8 Hz and simultaneously occur with VLFEs (Very Low-Frequency Earthquakes) having energy in frequency band 0.01–0.1 Hz. Slow earthquakes are important research target, because their activities reflect the stress state at the plate boundary adjacent to the fault area of maegathrust earthquake (Obara & Kato, 2016). Off Erimo, southern Hokkaido, VLFEs have been active every several months to a year after the 2003 Tokachi-oki earthquake (Asano et al., 2003; Baba et al., 2020). On the other hand, because seismological amplitude of tremors is small, a near-field network is necessary to observe them. At the corresponding area, tremors have been observed after 2016 when the Seafloor observation network for earthquakes and tsunamis along Japan Trench (S-net) started operation (Tanaka et al., 2019; Nishikawa et al., 2019). However, in the corresponding area, there is no understanding for actual tremor activities before 2016. This study analyzes the records collected by a pop-up type ocean bottom seismometer array installed during 2006 to 2007 and tries to detect tremors. During the observation period, VLFE activities were reported off Erimo, Hokkaido, and it is probable that tremors were also active.
In our previous result, tremors with duration over 20 detected by an envelope cross-correlation method (e.g. Ide, 2010), after filtering according to epicenter errors and time residual were 1,172 events. Most of their epicenters distributes similar to tremors detected by S-net observation (Tanaka et al., 2019; Nishikawa et al., 2019). In addition, although there is some difference in the activity cycle and scale from results of S-net studies, we found 3 high activities followed by VLFE occasions (Asano et al., 2003; Baba et al., 2020) and their intermedial aseismic periods. This feature is commonly observed in S-net studies. On the other hand, we also detected tremors occurring constantly but less active between major active periods. There is a possibility mistaking regular earthquakes barely satisfying thresholds for detecting tremors. Therefore, we reinvestigated tremors by a discrimination based on the waveform and frequency analysis information.
For the discrimination, we determined regular earthquakes based on following features; (1) clear P and/or S wave arrivals in seismogram envelope, (2) exponential decay in amplitude and (3) having power even at 10 Hz or higher. We interpreted the event with (4) less difference in seismogram amplitudes between stations as a teleseismic event. We excluded the events with (5) power only at high frequency (over 8 Hz) and (6) low S/N ratio at the stations near the epicenter as noise. In order to improve the efficiency and objectivity of the discrimination that is currently being performed visually, we plan to study a method for automatically discriminating in the future.
The discrimination certified 201 events as regular earthquakes, 11 as noise, and 960 as tremors. The improved catalog made a segregation of tremor distribution from regular earthquakes clearer than Kawakubo et al. (2020, JpGU). Most of the tremors in the new catalog occurred usually belong 3 known activities, and low and constant events are mostly recognized as regular earthquakes. We newly discovered very small-scale activities with short activity cycles at intervals of main active periods, and they distribute at down-dip side of main tremors activities. Because we have been able to classify tremors and regular earthquakes, we would like to investigate the relationship between tremors and regular earthquakes in the future.
In our previous result, tremors with duration over 20 detected by an envelope cross-correlation method (e.g. Ide, 2010), after filtering according to epicenter errors and time residual were 1,172 events. Most of their epicenters distributes similar to tremors detected by S-net observation (Tanaka et al., 2019; Nishikawa et al., 2019). In addition, although there is some difference in the activity cycle and scale from results of S-net studies, we found 3 high activities followed by VLFE occasions (Asano et al., 2003; Baba et al., 2020) and their intermedial aseismic periods. This feature is commonly observed in S-net studies. On the other hand, we also detected tremors occurring constantly but less active between major active periods. There is a possibility mistaking regular earthquakes barely satisfying thresholds for detecting tremors. Therefore, we reinvestigated tremors by a discrimination based on the waveform and frequency analysis information.
For the discrimination, we determined regular earthquakes based on following features; (1) clear P and/or S wave arrivals in seismogram envelope, (2) exponential decay in amplitude and (3) having power even at 10 Hz or higher. We interpreted the event with (4) less difference in seismogram amplitudes between stations as a teleseismic event. We excluded the events with (5) power only at high frequency (over 8 Hz) and (6) low S/N ratio at the stations near the epicenter as noise. In order to improve the efficiency and objectivity of the discrimination that is currently being performed visually, we plan to study a method for automatically discriminating in the future.
The discrimination certified 201 events as regular earthquakes, 11 as noise, and 960 as tremors. The improved catalog made a segregation of tremor distribution from regular earthquakes clearer than Kawakubo et al. (2020, JpGU). Most of the tremors in the new catalog occurred usually belong 3 known activities, and low and constant events are mostly recognized as regular earthquakes. We newly discovered very small-scale activities with short activity cycles at intervals of main active periods, and they distribute at down-dip side of main tremors activities. Because we have been able to classify tremors and regular earthquakes, we would like to investigate the relationship between tremors and regular earthquakes in the future.