We determine the frequency-dependent amplification inherent in hard-rock sites across Canada under the assumption that it can be estimated from the ratio of the horizontal-to-vertical (H/V) components of ground motion. The use of H/V as an estimate of amplification was originally proposed by Nakamura (1989) for microtremors and then generalized to earthquake ground motions by Lermo and Chavez-Garcia (1993). The H/V method was applied to a Fourier spectra database compiled from 424 earthquakes of magnitude greater than or equal to 2, recorded on 32 three-component stations of the Canadian National Seismograph Network (CNSN), all sited on rock (shear-wave velocity > 1500 m/sec); in each case we analyzed the S-wave portion of the seismograms, including the direct S and other strong S phases such as Lg.
The average amplification (H/V) for rock sites does not vary significantly across the country. It is weakly frequency dependent, increasing from a factor near unity at 0.5 Hz to a maximum in the range from 1.2 to 1.6 at 10 Hz. This amplification is consistent with what would be expected from a gradual decrease in shear-wave velocity as the seismic waves approach the surface, due to factors such as near-surface weathering. The interpretation of the H/V ratio as a measure of site response is consistent with the general geological conditions of the recording sites. In particular, the 5 Hz H/V ratio can be correlated with local geological conditions. The central Canada sites have the lowest estimated site amplification; this is consistent with competent hard-rock site conditions with high near-surface shear-wave velocities. There are two stations that show anomalous H/V ratios that suggest significant local site effects; these are EDM near Edmonton, which may be influenced by topographic amplification, and A61, a station in the Charlevoix seismic zone.