| Upper-mantle structure beneath Canada is investigated at both large and small scales, using the broadband, three-component data set of the Canadian National Seismograph Network. In the large-scale study, 500 surface-wave waveforms are processed using partitioned wave form inversion, yielding an S-wave velocity model of the Canadian upper mantle with resolution down to 400 km. This is the first surface-wave based 3-D model of Canada and offers improved resolution over the body-wave model of Grand (1994). The model displays two large-scale anomalies: a high-velocity structure beneath the Canadian shield and platform associated with the cratonic keel, and pronounced low velocities beneath the Cordillera. The velocity contrast between these features is strong (∼10%) and sharp, occurring over distances of 600 km or less. High velocities persist to depths of ∼250 km beneath the North American craton, which I interpret to represent the base of the continental keel. Moderately low velocities beneath the St Lawrence valley region and Labrador may be related to intracontinental volcanism and rifting, respectively.; At smaller scales, seismic discontinuities in the upper mantle lead to the scattering of teleseismic P waves into Ps conversions and free-surface multiples. The teleseismic P-wave coda thus provides constraints on small-scale, layered upper-mantle structure, including anisotropy and layer dip. I develop an efficient high-frequency method for characterizing upper mantle structure beneath a single station in the presence of interface dip and anisotropy, using it to model synthetic data and seismograms from CNSN station YKW3 (Yellowknife, NWT). In order to automate the recovery of upper-mantle models from CNSN datasets, I combine the forward-modeling scheme with the Monte Carlo-based neighborhood algorithm of Sambridge (1999) to invert receiver-function data. The combined approach is applied to synthetic waveforms, successfully constraining layer velocities and thicknesses, the degree and orientation of anisotropy, and the dip of interfaces. Applied to real data, the method provides strong constraints on crustal velocities and thicknesses beneath stations GAC (Quebec), SADO (Ontario), and ULM (Manitoba). In addition, it is used to determine the orientation of anisotropic fabric beneath YKW3, and the structural dip of the Cascadia subduction zone beneath station PGC (southern Vancouver Island). |
