Seismic reflection profile derived from the natural earthquakes recorded with seafloor DAS data

Abstract

　Using distributed acoustic sensing (DAS), we can obtain seismic data from optical fibers. Compared to conventional seismometers, DAS is cost-effective because it is easy to observe seismic data in high spatial resolution. Furthermore, once optical fibers are installed, seismic data can be easily obtained many times. In this study, we applied autocorrelation analysis with spectral whitening to natural seismic data obtained from DAS observations using optical fibers installed off the Sanriku coast and estimated the subsurface structure beneath the seafloor by producing a seismic reflection profile. To increase signal-to-noise ratio, we performed a stacking process of autocorrelation functions from 23 earthquakes. Typically, extracting shallow reflection surface using autocorrelation analysis is difficult because the resulting seismic reflection profile has high amplitude close to 0 second in travel time. Therefore, we calculated a moving average of every 5 km along the fiber from the autocorrelation function obtained at each observation point and suppressed the apparent shallow reflective surfaces by calculating their difference from the original autocorrelation function. The resulting autocorrelations show the basement structure at some of the observation locations in this study. The results also show a disturbance in the reflected wave affected by a fault and reflections affected by a shallow reflective surface, which had not appeared in the previous study using ambient noise. This is probably because natural earthquakes have strong energy over a wide range of frequencies, and S-waves reflected from fault surfaces reach the ground surface and can be observed by the DAS. This study shows that DAS observations can provide high-resolution subsurface structure.