P

[S15P-02] Features of horizontal-to-vertical spectral ratios of ambient noises and earthquake motions at the S-net sites

〇Yadab Prasad DHAKAL¹, Atsushi Wakai¹, Hiroaki Yamanaka², Takashi Kunugi¹, Shin Aoi¹ (1.National Research Institute for Earth Science and Disaster Resilience, 2.Tokyo Institute of Technology)

Studies about the horizontal-to-vertical spectral ratios (HVSRs) of ambient noises and earthquake motions are popular in Japan and many other earthquake-prone countries to estimate the dynamic properties of soils for earthquake disaster mitigation. In this study, we analyzed the HVSRs of ambient noises, referred to as noises, and earthquake motions at the S-net seafloor sites to understand the site effect on the recorded motions for the reliable prediction of ground motions for earthquake early warning. We processed the strong-motion records at the 150 stations of S-net from over 1000 earthquakes of Mj >= 4 and JMA focal depth < 70 km recorded between 2016 and 2019. We focused on the relatively low amplitude records (maximum value of peak accelerations of three components < 50 gals) to minimize the effects of rotations of sensor vessels and nonlinear site responses that may occur during large amplitude earthquake motions. The horizontal and vertical components of motions were obtained by finding the roll and pitch angles from the pre-event parts of the records and matrix operations described in Takagi et al. (2019). The S-wave onset times were estimated from the JMA2001 travel-time table (Ueno et al. 2002) to expedite the data processing, and a time window of 40.96 s was selected beginning 5 s before the S-onset time. We obtained the Fourier spectral amplitudes for each component, smoothed them using the smoothing function of Konno and Ohmachi (1998), and obtained the HVSRs as the square root of the sum of squares of the two horizontal components of motions divided by the vertical component of motions. The HVSRs for the noises were also similarly computed from the parts of the records before the earthquake origin time. We show the median values of the HVSRs for the noises and S waves (more accurately, extended S-wave portion) at all sites of the S-net in the attached figure. The plots of HVSRs of the noises gave gentle hump-shaped curves at frequencies between about 0.1 and 2 Hz at many sites, while they were flat at higher frequencies. In contrast, the plots of the S-wave HVSRs gave higher and wider hump-shaped curves including between 1 and 10 Hz. Some of the curves showed prominent peaks with steep slopes between 1 and 10 Hz. The maximum values of the HVSRs for the noises ranged between about 1.5 and 7, with the median value about 3.5. In contrast, the maximum values of the HVSRs for the S-waves ranged between about 5 and 40, with the median value about 10. The peak frequencies of the HVSRs for the noises mainly ranged between 0.1 and 2 Hz, with the median frequency of about 0.6 Hz, while they were mostly between 0.2 and 10 Hz for the S-waves, with the median frequency of about 3 Hz. The HVSRs of the noises at few sites displayed sinuous pattern with frequencies lower than about 1-2 Hz, especially in the shallow water regions (depths < 600 m). The S-wave HVSRs showed a steep decrease of the values after about 7-8 Hz. Some sites gave peak values of the S-wave HVSRs near the resonance frequency of the P waves in the water layer, while such peaks did not exist for the noise HVSRs. It has been argued that the HVSRs at the ground stations primarily reflect the effect of sediments. On the other hand, the values at the seafloor sites may reflect the combined effect of the sediments and water layer because the vertical components of motions are affected by the water layer. Further analysis is necessary to explain these observed features of the HVSRs to extract useful information for site characterization. Acknowledgments: This study was supported by "Advanced Earthquake and Tsunami Forecasting Technologies Project" of NIED and JSPS KAKENHI Grant Number JP20K05055.