Near-Surface Shear Wave Velocities for the Charleston Area: Models Derived from Seismic Land Streamer Data Using a Grid Search Approach

Session: Near-Surface Effects: Advances in Site Response Estimation and Its Applications [Poster]
Type: Poster
Date: 4/28/2020
Time: 08:00 AM
Room: Ballroom
Description: 

For the Charleston, South Carolina area, predominant peaks in the acceleration response spectra match fundamental periods of many existing buildings. The spectral differences result from variations in soil properties and are strongly dependent on shear wave velocities (Vs) of the near-surface materials. We acquired 14 km of seismic land streamer seismic data to map faults related to the 1886 Charleston earthquake. Complementary to the reflection data, we match fundamental surface wave dispersion curves, spaced every few meters, to obtain two-layer, two-dimensional Vs models. Here Pleistocene soils, upwards of 20 m thick, lie directly upon Tertiary strata. Previous downhole studies show little depth dependence within Pleistocene strata. Our 1-D inversion approach is to minimize the misfit between observed and calculated curves. We estimate Poisson’s ratio, p-wave velocity and density for each layer using body wave (first arrival) constraints and past rock property measurements. We initiate our grid search by estimating Pleistocene Vs values using a direct phase velocity/Poisson’s ratio relationship. We estimate Pleistocene soil thickness by matching inflection and slope on the phase velocity-frequency curve. Lastly, we constrain Tertiary Vs using the low frequency portion of the dispersion data for a range of previously published values. We perturb each parameter and identify the model with the lowest misfit. While Vs for Tertiary strata are poorly constrained with this approach, we note that resonant frequencies are strongly influenced by Pleistocene deposits. We compare our models to borehole measurements, surficial geology and ground penetrating radar data that identify a top of Tertiary reflector. We also compare our models to a standard 10-layer deterministic inversion approach to assess model uncertainties and extrapolate Vs with depth using Vp/Vs relationships.

Presenting Author: Lee M. Liberty

Authors