Probabilistic seismic-hazard site assessment in Kitimat, British Columbia, from Bayesian inversion of surface-wave dispersion

Abstract

This paper applies rigorous quantitative inversion methods to estimate seismic-hazard site classification and amplification factors in Kitimat, British Columbia, due to near-surface geophysical conditions. Frequency-wavenumber seismic-array processing is applied to passive data collected at three sites in Kitimat to estimate surface-wave dispersion. The dispersion data are inverted using a fully nonlinear Bayesian (probabilistic) inference methodology to estimate shear-wave velocity (VS) profiles and uncertainties. The VSresults are used to calculate the travel-time average of VSto 30 m depth (VS30) as a representation of the average sediment conditions, and to determine seismic-hazard site classification with respect to the National Building Code of Canada. In addition, VS30-dependent site amplification factors are computed to estimate site amplification at the three Kitimat sites. Lastly, the VSprofiles are used to compute amplification and resonance spectra for horizontally polarized shear waves. Quantitative uncertainties are estimated for all seismic-hazard estimates from the probabilistic VSstructure. The Kitimat region is the site for several proposed large-scale industrial development projects. One of the sites considered in this study is co-located with a recently deployed soil seismographic monitoring station that is currently recording ground motions as part of a 5 year campaign. The findings from this work will be useful for mitigating seismic amplification and resonance hazards on critical infrastructure, as well as for future seismological research, in this environmentally and economically significant region of Canada.