Constellation, le dépôt institutionnel de l'Université du Québec à Chicoutimi

Résumé

A novel probabilistic methodology for regional seismic site characterization is proposed and applied to a region with highly heterogeneous surficial geology and varying soil sediment thickness and stiffness. The method combines various sources of geological and geotechnical uncertainties to develop a three-dimensional (3D) shear-wave velocity (Vs) model and evaluate the associated uncertainties. A 3D geological model of the unconsolidated deposits was developed using geostatistical interpolation and simulation methods. Sequential indicator simulations produced a quantitative geologic model that explicitly quantified geological uncertainties based on the likelihood of specific soil types occurring. In situ measurements and multivariate statistical analysis allowed the development of empirical correlations between Vs, geotechnical parameters, depth, and soil types. The resulting 3D Vs values were estimated on the basis of Vs-depth correlations and the probability of occurrence of each soil type. In this approach, the propagated uncertainty was also quantified by considering the combined variance. Seismic microzonation mapping was then conducted by transforming the 3D Vs model into two-dimensional (2D) maps that represent the spatial distributions of the time-averaged shear-wave velocity of the top 30 m (Vs,30) and the fundamental site period (T0), along with their respective uncertainties using Monte Carlo simulations. The results indicate that microzonation maps and their uncertainties are influenced by the thickness, occurrence probability, and geotechnical properties of soils. The proposed method can be used to assess the probabilistic seismic risk at local and regional scales in areas with geologically and geotechnically complex soil properties.