Electronic Thesis and Dissertation Repository


Doctor of Philosophy




Dr. Gail M. Atkinson


The works presented in this thesis are aimed at understanding and modeling earthquake ground motions in central and eastern North America (CENA), with an emphasis on the modeling of site effects. A number of approaches are taken, beginning with analysing of the observed ground-motion amplitudes with respect to a ground-motion prediction equation model (GMPE) of a well recorded region (California) (referenced empirical approach). In this work, we show that the ground-motion amplitudes of CENA earthquakes are very similar to the equivalent California values of Boore et al. (2014; BSSA14) at close distances, at low to moderate frequencies. At regional distances and at high frequencies the ground-motion amplitudes are larger in CENA than for the BSSA14 reference model, presumably due to lower attenuation and higher stress forCENA events relative to those in active tectonic regions.

Next, a modeling approach is taken to develop a simulation-guided generic GMPE model for southern Ontario, examining regional site effects.Then, in a collaborative study with colleagues, we use the regional model in a technique that inverts ShakeMaps parameters (e.g. response spectra at selected frequencies) to estimate moment magnitude and stress parameter in near-real-time, for earthquakes of small to moderate earthquake (M ~3 to 6).

Finally, we explore site response more deeply, building on the empirical findings of the ground-motion modeling work. We examine the applicability of the site fundamental frequency (fpeak) as a descriptive variable for site response in CENA. We introduce a new fpeak-based proxy measure for VS30 (time-averaged shear-wave velocity in the upper 30 m) for sites in CENA; this is useful because VS30 is a standard site response parameter for building code and other applications. We also examine the applicability of the Next-Generation-West2 VS30-based site effects model (Seyhan and Stewart, 2014; SS14) for the recording stations in CENA. We develop a newfpeak-based model for site amplification in CENA to address inadequacies in the VS30-based site effects model, and show that use of the new model reduced ground-motion prediction variability by a significant amount.