Modeling of Empirical Transfer Functions including 3D Velocity Structure

Zhifeng Hu, Daniel Roten, Kim B. Olsen, & Steven M. Day

Submitted August 14, 2019, SCEC Contribution #9628, 2019 SCEC Annual Meeting Poster #007

Site response can be an important factor in estimating seismic hazard. However, conventional simplified modeling of site amplification with assumptions of plane SH waves propagating vertically through layered homogeneous media (SH1D) often poorly predicts the empirical transfer function (ETF), particularly where large lateral variations of velocity are present. We select two well-documented sites (the KiK-net site TKCH05, Japan, and the Garner Valley Downhole Array site GVDA, CA) for our study. The low inter-event variations of the ETFs at these sites suggest that the site effects are independent of the source and path effects, so a relatively small regional model encompasses the main factors controlling the site response. Here, we use physics-based simulations that naturally incorporate the complex material properties and provide synthetic ground motions to compute theoretical transfer functions (TTF). We calibrate the 3D subsurface geometry by means of the topography near the sites and incorporate information from the Vs profiles obtained from borehole logs. By comparing TTFs to the estimated ETFs at the selected sites, we show how simulations in the calibrated 3D medium, including statistical distributions of small-scale heterogeneities, are able to adequately model the site amplification. The results indicate that the subsurface structure is more heterogeneous at TKCH05, while more laterally uniform at GVDA. Nevertheless, in both cases properly calibrated spatial variation of the sediments below the site, including the slope of the edges and the depth to the bedrock, significantly improves the fit to the ETFs. We show that the lateral velocity variations, which can be well captured by mapping topography to underground geometry, can play an important role in accurately determining both the frequency and amplitude of the site response. As a consequence, the Vs profiles derived from the borehole measurements may be unable to produce TTFs consistent with the observed ETFs. The results emphasize the importance of reliable calibration of subsurface structure and material properties in site response studies.

Citation
Hu, Z., Roten, D., Olsen, K. B., & Day, S. M. (2019, 08). Modeling of Empirical Transfer Functions including 3D Velocity Structure. Poster Presentation at 2019 SCEC Annual Meeting.


Related Projects & Working Groups
Ground Motions