A Bernstein Polynomial Model in Bayesian Inversion of Surface-Wave Dispersion for Earthquake Site Response in British Columbia, Canada

Abstract eng:
Knowledge of near-surface properties of the soil column, in particular, the shear-wave velocity (Vs) profile over the upper 10s of meters, is important for characterizing the expected ground response to earthquake shaking at a specific site. Noninvasive and passive methods based on recording ambient seismic noise are increasingly popular for estimating Vs structure with minimal cost and site disruption. This paper applies a fully nonlinear Bayesian inversion methodology, based on parallel tempering, to estimate Vs profiles and uncertainties using surface-wave dispersion data processed from passive seismic array recordings. In the inversion, the Vs profile is parameterized using a Bernstein polynomial basis, which efficiently characterizes general depth-dependent Vs gradients in the soil column. Bernstein polynomials provide a stable parameterization in that small perturbations to the model parameters (basis function coefficients) result in only small perturbations to the Vs profile. Shear-wave velocity profiles and uncertainties are obtained from the velocity-depth marginal posterior probability density. This methodology is applied to passive seismic array recordings collected at several sites in Kitimat, British Columbia, Canada, a region where earthquake site response is of significant interest. The probabilistic Vs profile results ultimately provide probabilistic estimates of site response factors such as peak ground velocity/acceleration and Vs30.

• Export as: BibTeX | MARC | MARCXML | DC | EndNote | NLM | RefWorks
• Add to your basket