Seismic velocities and composition of the Canadian crust

- Crustal thickness and Vp/Vs measured at 343 stations across Canadian landmass
- Thickness and velocity ratios vary from 25 to 48 km, and 1.65 to 1.95, respectively.
- Results are consistent with bulk compositions between granite gneiss and diorite.
- Reflectivity profiles consistent with heterogeneity scale lengths of < 800 m

To investigate variations in crustal structure and composition, we analyze approximately 80,000 teleseismic P wave seismograms recorded at 343 broadband stations distributed about the Canadian landmass. These data are binned by horizontal slowness and simultaneously deconvolved into receiver functions. We apply stacking methods to retrieve estimates of the bulk crustal velocity ratio VP/VS and thickness H from this receiver function dataset under the assumption of locally 1D structure. These crustal parameters are analyzed together with a compilation of velocity and thickness estimates from previous active source experiments and global crustal thickness models to investigate competing models of crustal composition and evolution. Crustal thicknesses vary between ~ 25 km and ~ 48 km with an area-weighted average of 37 km for all stations. VP/VS estimates range from 1.65 to 1.95 and display a systematic increasing trend from NW to SE across the Churchill, Superior and Grenville provinces. Analysis of intra-crustal conversions indicates little evidence for a well-defined, widespread Conrad discontinuity, although the lower crust appears to be more transparent to teleseismic converted waves than the upper crust. This observation runs contrary to that of P-reflectivity in many active source studies, and may be reconciled through a dominant scale-length of heterogeneity that is less than ~ 800 m. The VP/VS observations and VP constraints from active source studies are compared with laboratory measurements for common crustal rocktypes and found to be consistent with bulk compositions falling between granite gneiss and diorite. Previously reported systematic variations in crustal thickness and VP/VS with age in a global data set are not supported by our data set.