14:00 〜 14:15
[SSS02-02] Vp and Vs structure estimation from ambient noise record observed by distributed acoustic sensing with a submarine fiber optic cable
キーワード：DAS、南海、地震波速度構造
Distributed acoustic sensing (DAS) with fiber optic cable is now a powerful tool for detecting signals of various wave propagations due to e.g., earthquakes and microseisms, because the station densities of cables are higher than those of typical seismic observations using individual seismic sensors. A previous study estimated a S-wave velocity (Vs) structure within sediment layers below a submarine cable off Tohoku region, Japan (Spica et al. 2020). However, P-wave velocity (Vp) structure has not been estimated by DAS techniques, because observed wavefield by the DAS is mainly the horizontal component in which S wave and surface waves are dominant. In this study, we attempted to estimate Vs and Vpstructures below an ocean-bottom cable in the Nankai subduction zone, Japan, with Rayleigh and P waves, respectively.
We applied a frequency-wavenumber (f-k) analysis for measuring propagation velocities of waves in the ambient noise records. The starting points of sub-arrays are distributed with an increment of 2.5 km along the cable. The cross-correlation functions (CCFs) were calculated within 5 km from the starting points. Here, because the cable is not linear, the distances between two stations are defined by the line between the two stations. The station spacing was 5.1 m, but we stacked the continuous records within ±2 stations in order to increase signal-to-noise ratios. The wavefield of each sub-array was interpolated to an increment of 25 m within 5 km. The dispersion curves of the Rayleigh wave are measured in the f-k analysis, and one-dimensional (1D) velocity inversion was performed with a software of DISPER80 (Saito, 1988). Also, to retrieve P wave information, waves with velocities greater than 1.6 km/s were filtered in the f-k domain. We then performed tau-p stacks using the filtered CCFs at a frequency band of 0.4–0.8 Hz, and 1D velocity inversion was performed by tau-sum inversion (Stoffa et al. 1981; Shinohara et al. 1994).
Our results show Vs structure below the cable down to a depth of 2–3 km, in which slow and high velocity at the topmost part of the sediments were observed in the northern and southern region of the cable. This difference may be caused by the geologic settings between the two regions: the northern and southern parts of the cable consist of basin and ridge (Tosa-bae), respectively. The 1D Vs structure at the northern edge of the cable shows higher velocity than that at other sub-arrays, because it is considered that this high velocity reflects seismic velocity structure in land. Moreover, such a high velocity structure can be observed in Vp structure. This ensures that the both approaches show reliable velocity structure.
We applied a frequency-wavenumber (f-k) analysis for measuring propagation velocities of waves in the ambient noise records. The starting points of sub-arrays are distributed with an increment of 2.5 km along the cable. The cross-correlation functions (CCFs) were calculated within 5 km from the starting points. Here, because the cable is not linear, the distances between two stations are defined by the line between the two stations. The station spacing was 5.1 m, but we stacked the continuous records within ±2 stations in order to increase signal-to-noise ratios. The wavefield of each sub-array was interpolated to an increment of 25 m within 5 km. The dispersion curves of the Rayleigh wave are measured in the f-k analysis, and one-dimensional (1D) velocity inversion was performed with a software of DISPER80 (Saito, 1988). Also, to retrieve P wave information, waves with velocities greater than 1.6 km/s were filtered in the f-k domain. We then performed tau-p stacks using the filtered CCFs at a frequency band of 0.4–0.8 Hz, and 1D velocity inversion was performed by tau-sum inversion (Stoffa et al. 1981; Shinohara et al. 1994).
Our results show Vs structure below the cable down to a depth of 2–3 km, in which slow and high velocity at the topmost part of the sediments were observed in the northern and southern region of the cable. This difference may be caused by the geologic settings between the two regions: the northern and southern parts of the cable consist of basin and ridge (Tosa-bae), respectively. The 1D Vs structure at the northern edge of the cable shows higher velocity than that at other sub-arrays, because it is considered that this high velocity reflects seismic velocity structure in land. Moreover, such a high velocity structure can be observed in Vp structure. This ensures that the both approaches show reliable velocity structure.