1:45 PM - 2:00 PM
[SSS10-07] Estimation of subsurface motion from surface seismic records (2): improvement of the method based on the quantitative evaluation
Keywords:seismic motion, plane wave, strong motion, subsurface wavefield
Subsurface seismic motion or incident wavefield on the bedrock level in needed to evaluate the response of buildings and the seismic stability of the shallow sedimentary structure through numerical simulation of seismic wave propagation. It is conventional that the subsurface motion is estimated from a seismic record at a free surface position assuming a vertical plane-wave incidence, which means that the vertical and horizontal components of the seismic waves independently propagate as P-wave and S-wave, respectively, this assumption is theoretically incorrect. We proposed a new method to evaluate subsurface wavefields from a surface record without the assumption of plane-wave incidence (Takenaka et al., 2019, SSJ Fall meeting)and demonstrated the feasibility of the method through comparison of the actual subsurface records with estimated motions at some shallow and deep levels (Watanabe et al., 2020, JpGU; Watanabe et al., 2021, JpGU). In this study, we improve the method by employing a quantitative value to show the misfit between the estimated and observed waveforms.
Our method uses a horizontally-layered model including both the station and the source. We calculate synthetic seismograms assuming the focal mechanism and impulsive source at the surface and the subsurface points. We then deconvolve the surface record with the surface synthetic seismogram after synchronizing them by matching the S wave arrivals each other. The resultant waveform includes the actual source time function and the effect due to the difference between the structural model and real structure. To improve our model, we then extract the effective source time function (ESTF) by windowing this waveform in the time domain. We finally convolve the ESTF with the subsurface synthetic seismogram to estimate the target motion or incident wave at the evaluation point.
To select the appropriate length of the ESTF, which is controlled by the length of the window function, we adopt an index E-FDE (Envelope Frequency Domain Error) proposed by Suzuki et al. (2018) as misfit value between two waveforms. Applying to five stations with different depth subsurface receivers, we found that the appropriate length of the ESTF depends on the depth at the evaluation point. The deeper the target motion is, the shorter the length is and vice versa.
Fig. 1 is an application example to surface records at KiK-net station OSKH05, NIED for the of the MJMA 5.4 event occurring in the southern part of Wakayama Prefecture on November 19, 2016 at depth of 50 km with an epicentral distance 97 km. Fig. 1 shows the subsurface motion observed and estimated by our method and conventional assuming vertical plane-wave incidence. The misfit values indicate that our method can recover the observed ones better rather than by the conventional method.
Our method uses a horizontally-layered model including both the station and the source. We calculate synthetic seismograms assuming the focal mechanism and impulsive source at the surface and the subsurface points. We then deconvolve the surface record with the surface synthetic seismogram after synchronizing them by matching the S wave arrivals each other. The resultant waveform includes the actual source time function and the effect due to the difference between the structural model and real structure. To improve our model, we then extract the effective source time function (ESTF) by windowing this waveform in the time domain. We finally convolve the ESTF with the subsurface synthetic seismogram to estimate the target motion or incident wave at the evaluation point.
To select the appropriate length of the ESTF, which is controlled by the length of the window function, we adopt an index E-FDE (Envelope Frequency Domain Error) proposed by Suzuki et al. (2018) as misfit value between two waveforms. Applying to five stations with different depth subsurface receivers, we found that the appropriate length of the ESTF depends on the depth at the evaluation point. The deeper the target motion is, the shorter the length is and vice versa.
Fig. 1 is an application example to surface records at KiK-net station OSKH05, NIED for the of the MJMA 5.4 event occurring in the southern part of Wakayama Prefecture on November 19, 2016 at depth of 50 km with an epicentral distance 97 km. Fig. 1 shows the subsurface motion observed and estimated by our method and conventional assuming vertical plane-wave incidence. The misfit values indicate that our method can recover the observed ones better rather than by the conventional method.