Poster #033, Ground Motions

Calibration of the Near-surface Seismic Structure in the SCEC Community Velocity Model Version 4

Zhifeng Hu, Kim B. Olsen, & Steven M. Day
Poster Image: 

Poster Presentation

2021 SCEC Annual Meeting, Poster #033, SCEC Contribution #11298 VIEW PDF
The near-surface seismic structure (to a depth of about 1000 m), particularly the shearwave velocity (Vs), can strongly affect the propagation of seismic waves, and therefore must be accurately calibrated for ground motion simulations and seismic hazard assessment. The Vs of the top (< 300 m) crust is often well-characterized from borehole studies, geotechnical measurements, and water and oil wells, while the velocities of the material deeper than about 1000 m are typically determined by tomography studies. However, in regions lacking information on shallow lithological stratification, typically rock sites outside the sedimentary basins, the material parameters between these two regions ar...e typically poorly characterized due to resolution limits of seismic tomography. When the alluded geological constraints are not available, models, such as the Southern California Earthquake Center Community Velocity Models (CVMs), default to regional tomographic estimates that do not resolve the uppermost Vs values, and therefore deliver unrealistically high shallow Vs estimates. A widely-used method for incorporating the
near-surface earth structure is implemented in CVMs by applying a generic overlay based on measurements of time-averaged Vs in top 30 m (Vs30) to taper the upper part of the model to merge with tomography at certain depth (e.g., 350 m). However, our 3D simulations of the 2014 M w 5.1 La Habra earthquake in the Los Angeles area using the CVM-S4.26.M01 model significantly underpredict low-frequency (< 1 Hz) ground motions at sites subject to the generic overlay (“taper”). On the other hand, extending the Vs30-based taper of the shallow velocities down to a depth of about 1000 meters improves the fit between our synthetics and seismic data at those sites, without compromising the fit at well constrained sites. We explore various tapering depths, demonstrating increasing amplification as the tapering depth increases, and the model with 1000 m tapering depth yields overall favorable results. Effects of varying anelastic attenuation are small compared to effects of velocity tapering. Although a uniform tapering depth is adopted in the models, we observe some spatial variabilities that may further improve our method.