Microzonation models for Montreal with respect to $\hbox {V}_\mathrm{S30}$

Abstract

The development of seismic microzonation maps is now a prerequisite for performing advanced seismic hazards and seismic risk assessments in urban areas. Current microzonation procedures are based on classifying site response in soil categories based on $\hbox {V}_\mathrm{S30}$. In many practical cases, information on shear wave velocity is sparse and analysts must resort to other sources of information to develop zonation maps over large regions. For Montreal, Canada, a geostatistical approach combining a large dataset of boreholes, $\hbox {f}_{0}$ data from the H/V method, and a limited number of shear wave velocity measurements is proposed. Predictive equations are developed with the different types of information and partly used to create three different $\hbox {V}_\mathrm{s30}$ models: A first model based on the total thickness of soft soil deposits (One Layer Model), a second model based on stratigraphy obtained from borehole data (Four Layers Model) and a third model (Composite Model) that combines data from the previous two models with data from seismic surveys. Variability and uncertainty associated with the choice of the models are described and estimated using a second moment analysis. The composite model which combines borehole, $\hbox {f}_{0}$ and Vs data is shown to be the most accurate compared to models derived only from borehole and Vs data (an average reduction of 40 % of the squared error) and shows good agreement with the geological map for clay and sand locations with soils classes C and D (54 % agreement on average). This approach opens new perspectives for $\hbox {V}_\mathrm{s30}$ microzonation in urban areas when budgets for Vs surveys are limited.