Electronic Thesis and Dissertation Repository

Thesis Format

Integrated Article

Degree

Doctor of Philosophy

Program

Civil and Environmental Engineering

Supervisor

El Naggar, Hesham M.

2nd Supervisor

Molnar, Sheri

Joint Supervisor

Abstract

Metropolitan (Metro) Vancouver is the largest city in British Columbia and has the highest earthquake risk in Canada due to vulnerability of persons and infrastructure located near the Cascadia subduction zone. The effects of local geology on seismic waves (site effects) play a major role in determining surface shaking levels and seismic hazard. Site effects are typically accounted for in ground motion prediction by inclusion of important seismic site characteristics. This thesis first evaluates the seismic site characteristics across Metro Vancouver then facilitates the development of a region-specific site amplification model.

To evaluate site characteristics in Metro Vancouver, a comprehensive database of seismic site condition measures is compiled to develop important predictive site characteristic models for site-specific and regional application. Non-invasive surface wave measurements are conducted to obtain the depth to glacial till (zgl) and deep shear wave velocity (Vs) profiles with their uncertainties in the Fraser River Delta (FRD). A model to predict zgl based on the 2nd peak frequency of the microtremor horizontal-to-vertical spectral ratio (MHVSR) is proposed. Seismic Cone Penetration Testing (SCPT) soundings are compiled to derive a CPT-to-Vs model using traditional linear and nonlinear regressions along with two machine learning approaches and a combination of CPT parameters. Machine learning approaches show slightly improved prediction accuracy compared to traditional regression due to the availability of a large database. A comprehensive Vs database is collected from available invasive and acquired non-invasive in situ measurements in the region. Generic Vs-depth relationships of post-glacial, glacial and rock geologies are developed. A comprehensive Vs30, the time-averaged shear wave velocity to depth 30 m, database is established from direct Vs measurements and via conversion of other in situ measurements using the developed predictive models. Region-specific Vs30 prediction models are developed based on mapped geology and topographic slope or zgl.

To facilitate the development of a region-specific site amplification model, one-dimensional (1D) site response analyses (SRA) of the Metro Vancouver region are conducted. Input time histories for the three earthquake source types are scaled at 2 and 10 % probability of exceedance in 50 years consistent with the national probabilistic seismic hazard model. The predicted linear and nonlinear SRA de/amplification spectra at 8 selected sites are compared to a site amplification model for a western North America (WNA) ground motion model (GMM). The WNA amplification model overestimates or underestimates linear site amplification for deep and shallow sites in Metro Vancouver respectively. For thick post-glacial sediments in the FRD and North Shore, stronger nonlinearity is predicted than exists in the WNA amplification model. Potential sources of 1D SRA uncertainties in Metro Vancouver are quantified via sensitivity analysis. The 1D SRA accomplished in this thesis are an important step towards developing a region-specific site amplification model that can be combined with developed site characteristics mapping to produce the first 1D site amplification map for Metro Vancouver.

Summary for Lay Audience

Seismic hazard quantification is a first step towards making informed decisions in urban planning and engineering applications to reduce the impact of earthquakes. Seismic site effects or the effect of near surface geology on the seismic waves plays a major role in determining the shaking hazard at the surface. Seismic site characteristics are metrics that correlate well with observed site effects. These characteristics include the variation of the soil stiffness with depth, shear wave velocity (Vs) depth profile, Vs30 (the time-averaged shear wave velocity to 30 m depth), and depths to glacial sediments and rock.

This thesis provides predictive models to spatially map important site characteristics in Metro Vancouver and facilitates the development of region-specific models to account for site effects. Cost-effective non-invasive measurements are conducted to obtain measures of site characteristics (Vs profiles, zgl) in the Fraser River Delta (FRD). A model to map zgl based on non-invasive measurements is proposed for shallow sites. Available geotechnical field measurements (seismic cone penetration testing, SCPT) are compiled to derive a Vs predictive model from CPTs using traditional and machine learning regression approaches. The findings promote the use of machine learning approaches in geotechnical applications when a large database is available. A comprehensive Vs database is collected from existing and acquired measurements and average Vs of different sediments are presented. Further, the predictive models are applied to available measurements to propose a model to produce the first Vs30 map for Metro Vancouver. A Vs30 map is an essential input to many applications involving emergency response, urban planning, and risk studies.

Finally, numerical simulations are conducted to predict site-specific surface ground motions inclusive of 1D site effects. Earthquake loading for different types of earthquakes is determined and propagated through soil models to predict site amplification. Comparison of the site-specific amplification with those predicted from a ground motion prediction equation applicable to the region is inconsistent and confirms our site-specific modelling is required to predict accurate ground motions in Metro Vancouver. Recommendations for developing a new model that better captures Metro Vancouver soil response during earthquakes are provided.

Download

Share

COinS