12:00 〜 12:15
[SCG45-21] Single-station detection of seismic slow earthquakes using their broadband characteristics
キーワード：スロー地震、広帯域スロー地震、テクトニック微動、超低周波地震、南海沈み込み帯、メキシコ
Slow earthquakes radiate tiny signals and are observed only in some limited frequency bands with low noise. Those observed in 1-10 Hz are called tectonic tremors or low-frequency earthquakes and those observed in 0.01-0.05 Hz are called very-low-frequency earthquakes (VLFE). These seismic slow earthquakes are spatiotemporally correlated, and the seismic-moment rate calculated from VLFE is almost always proportional to the seismic-energy rate calculated in the tremor frequency band (Ide et al., 2008). Brownian motion-type models (e.g., Ide, 2008) can explain this relationship and also have predicted the radiation of broadband signals through 0.01-10 Hz. In fact, broadband detections of seismic slow earthquakes including microseism frequency of 0.05-1 Hz support the idea (e.g., Kaneko et al., 2018; Masuda et al., 2020). In this study, we utilize this characteristic of slow earthquakes to design a method to monitor slow-earthquake activities even from a single-station data.
From a broadband seismogram recorded at a single station, we measure seismic energy rate as the squared-velocity waveform in 2-8 Hz and, and seismic moment rate as the displacement waveform in 0.02-0.05 Hz. Then we calculate the correlation coefficient between the seismic energy rate and seismic moment rate, every 10 s in a time window of 100 s, and take a moving average of 10000 s. We regard the timing when this value exceeds a certain threshold is “detection”.
The first application of this method is to the continuous records of F-net, a broadband seismic network of 75 stations in Japan for 18 years. The detected activities are spatiotemporally consistent with a tremor catalog (Mizuno and Ide, 2019), and the false-positive rate is low. Though some tremor activities, especially far from the station, are missed, this limitation of the detectable distance constrains the locations of detected seismic slow earthquake activities. Next, we apply this method to the Guerrero, Oaxaca, and Jalisco regions in the Mexican subduction zone. In the Guerrero and Oaxaca regions, we expand the number of usable stations as well as obtain detections consistent with a previous study (Husker et al., 2019). Furthermore, in the Jalisco region, this is the first long-term analysis of seismic slow earthquakes using a permanent station, and the large-scale activities detected by this study are not only consistent with previous a geodetic study but also newly detected one.
From a broadband seismogram recorded at a single station, we measure seismic energy rate as the squared-velocity waveform in 2-8 Hz and, and seismic moment rate as the displacement waveform in 0.02-0.05 Hz. Then we calculate the correlation coefficient between the seismic energy rate and seismic moment rate, every 10 s in a time window of 100 s, and take a moving average of 10000 s. We regard the timing when this value exceeds a certain threshold is “detection”.
The first application of this method is to the continuous records of F-net, a broadband seismic network of 75 stations in Japan for 18 years. The detected activities are spatiotemporally consistent with a tremor catalog (Mizuno and Ide, 2019), and the false-positive rate is low. Though some tremor activities, especially far from the station, are missed, this limitation of the detectable distance constrains the locations of detected seismic slow earthquake activities. Next, we apply this method to the Guerrero, Oaxaca, and Jalisco regions in the Mexican subduction zone. In the Guerrero and Oaxaca regions, we expand the number of usable stations as well as obtain detections consistent with a previous study (Husker et al., 2019). Furthermore, in the Jalisco region, this is the first long-term analysis of seismic slow earthquakes using a permanent station, and the large-scale activities detected by this study are not only consistent with previous a geodetic study but also newly detected one.