GEOSCAN Search Results: Fastlink


TitleEarthquake site response characterization in Kitimat, BC, via Bayesian gradient-based inversion of surface-wave dispersion
AuthorGosselin, J M; Cassidy, J F; Dosso, S E; Brillon, C
Source 2017, 1 pages
Alt SeriesNatural Resources Canada, Contribution Series 20170029
PublisherCanadian Geophysical Union
MeetingCanadian Geophysical Union Annual Meeting; Vancouver; CA; May 29-31, 2017
Subjectstectonics; earthquakes; earthquake studies; displacement; s waves; seismic velocities; modelling; computer simulations; Bernstein polynomial model; dispersion
ProgramAssessing Earthquake Geohazards, Public Safety Geoscience
AbstractEarthquake ground motions at a particular site are strongly influenced by local geology, specifically the geophysical properties of the upper 10s of metres of the soil/sediment column. Displacement is amplified as seismic waves propagate through material of decreasing impedance, such as soft, unconsolidated sediments. Hence, knowledge of the local, near-surface geophysical conditions, particularly the shear-wave velocity (Vs) structure, is important for characterizing and predicting the ground response to earthquake shaking at a particular site. Passive, non-invasive methods for estimating Vs based on recordings of ambient seismic noise are becoming increasingly popular. This study develops an inversion methodology for estimating Vs structure from surface-wave dispersion extracted from passive seismic recordings. In the inversion, the one-dimensional Vs structure is described using a Bernstein polynomial model, which efficiently represents smooth gradient structure in the geophysical properties of the sediment column. The inversion is performed within a Bayesian (probabilistic) framework in which the solution is defined in terms of the marginal posterior probability profile of Vs, which provides a quantitative measure of uncertainty in the inversion results. The approach considered here is better suited than uniform layered modelling approaches in applications where smooth gradients in geophysical properties are expected. The Bernstein polynomial is more general than other gradient-based models such that the data information constrains the form of the gradient Vs structure. This methodology is applied to dispersion data processed from passive seismic recordings collected in Kitimat, British Columbia, a region of significant proposed infrastructure development. Bayesian inversion results are compared to results from other geophysical methods, and are used in probabilistic earthquake site response characterization.
Summary(Plain Language Summary, not published)
Knowledge of near-surface properties of the soil column, in particular, the shear-wave velocity (Vs) profile over the upper 10¿s of metres, are important for characterizing the expected ground response to earthquake shaking. In this study we improve upon non-invasive and passive techniques that use recordings of ambient seismic noise (and so are low-cost, fast and minimise site disruption). We apply an advanced inversion methodology to estimate Vs structure and uncertainties from surface-wave recordings made on a seismic array. This new methodology is applied to passive seismic array recordings collected at five sites in Kitimat, British Columbia, Canada and we compare our inversion results from several sites to active-source reflection data collected nearby.