5:15 PM - 6:30 PM
[SSS11-P15] Estimation of subsurface motion from surface seismic records: investigation on evaluation at deep levels
Keywords:seismic motion, plane wave, strong motion, subsurface wavefield
We need subsurface seismic motions or incident wavefields on a given subsurface level to evaluate the seismic stability of the shallow subsurface structure and the response of buildings through simulating the strong motion. It is conventional that the subsurface motion is estimated from the ground motion record at a free surface position by assuming a vertical plane-wave incidence. However, this approach requires that the vertical and horizontal components of the seismic wave independently propagate as P wave and S wave, respectively, which is theoretically incorrect, and could not represent the correct deformation in the subsurface structure. We have proposed a new method to evaluate the subsurface wavefields from a surface record without the assumption of the plane-wave incidence (Takenaka et al., 2019, SSJ Fall meeting) and applied the method to compare the evaluated motion with the subsurface records at shallow levels of around 100 – 200 m (Watanabe et al., 2020, JpGU). The evaluation level is deeper, the actual wavefields could be more different from the conventionally estimated ones based on the assumption of the vertical incident plane wave, due to the oblique incident effect and so on. In this study, we check the effectiveness of our method and the conventional one by reproducing subsurface records in depth of more than 2 km.
Our method needs a horizontally-layered model including the earthquake source. We calculate the synthetic seismograms at the surface station and the subsurface evaluation point assuming the focal mechanism and impulsive source. We then deconvolve the surface record with the surface synthetic seismogram to derive the effective source time function. We apply the water level method for the stable deconvolution. The effective source time function has the substantial source time function and the unsimulated effects by the structural model, which are included in the surface record. Then, we convolve the effective source time function with the subsurface synthetic seismogram to derive the target motion at the evaluation point.
We apply our method and conventional one to the surface records observed at the KiK-net (NIED) TKYH02 station for a source of Mj 4.8, focal depth 83 km, and epicentral distance 53.73 km. Figure 1 shows the subsurface motion observed and evaluated by three methods, our method and methods based on vertical and oblique plane-wave incidence, respectively. We select the horizontal slowness of S-wave for the oblique plane-wave incidence so as to fit the estimated subsurface waveform of the transverse component to the observed one. We could successfully recover the subsurface motion by our method. In Fig. 1(a), it is found that there are differences in the amplitude of the P-wave coda and later phases on the radial component. The subsurface waveform estimated by our method is similar to one by the method based on the oblique plane-wave incidence. We confirmed that the subsurface acceleration spectra evaluated by three methods recovered the observed one around a few Hz, while our method underestimated the spectra in lower than 1 Hz because of the effect of the water level method. We will also report about the estimation of the subsurface motion using other events and records, the evaluation of incident wavefields, and the estimation at the subsurface evaluation points with horizontal offsets from the surface station and so on.
Our method needs a horizontally-layered model including the earthquake source. We calculate the synthetic seismograms at the surface station and the subsurface evaluation point assuming the focal mechanism and impulsive source. We then deconvolve the surface record with the surface synthetic seismogram to derive the effective source time function. We apply the water level method for the stable deconvolution. The effective source time function has the substantial source time function and the unsimulated effects by the structural model, which are included in the surface record. Then, we convolve the effective source time function with the subsurface synthetic seismogram to derive the target motion at the evaluation point.
We apply our method and conventional one to the surface records observed at the KiK-net (NIED) TKYH02 station for a source of Mj 4.8, focal depth 83 km, and epicentral distance 53.73 km. Figure 1 shows the subsurface motion observed and evaluated by three methods, our method and methods based on vertical and oblique plane-wave incidence, respectively. We select the horizontal slowness of S-wave for the oblique plane-wave incidence so as to fit the estimated subsurface waveform of the transverse component to the observed one. We could successfully recover the subsurface motion by our method. In Fig. 1(a), it is found that there are differences in the amplitude of the P-wave coda and later phases on the radial component. The subsurface waveform estimated by our method is similar to one by the method based on the oblique plane-wave incidence. We confirmed that the subsurface acceleration spectra evaluated by three methods recovered the observed one around a few Hz, while our method underestimated the spectra in lower than 1 Hz because of the effect of the water level method. We will also report about the estimation of the subsurface motion using other events and records, the evaluation of incident wavefields, and the estimation at the subsurface evaluation points with horizontal offsets from the surface station and so on.