Source spectra are obtained for the 1988 Saguenay, Quebec, earthquake (mN = 6.5), its foreshock (mN = 4.8), and largest aftershock (mN = 4.1), using recordings from analog and digital seismographs on rock sites at epicentral distances from 40 to 700 km. The source spectra of the foreshock and the aftershock are well matched by a single-corner-frequency ω−2 model, with seismic moments of 2.2 × 1022 and 2.2 × 1021 dyne-cm, respectively, and a stress parameter of 65 bars for both events. By contrast, the spectrum of the mainshock differs dramatically from this simple shape. A stress parameter of 500 bars is required to match the observed high-frequency amplitudes, using the seismic moment determined from teleseismic observations as a constraint. This single-corner-frequency model, however, overestimates amplitudes at frequencies near 1 Hz by a factor of about 2. The relative radiation of short- and long-period energy for the mainshock is inconsistent with that from other intraplate earthquakes; the fore- and aftershocks appear typical in this regard.
The extraction of source spectra from the regional data required a careful examination of attenuation. Azimuthal variations were observed, with the attenuation being most severe to the southeast (normal to the St. Lawrence Valley and the dominant structural grain of the region). The St. Lawrence River coincides in part with the boundary between the Grenville and Appalachian tectonic provinces, and the difference in attenuation may be related to the difference in tectonic province. The distribution of data, however, makes it difficult to obtain good estimates of the attenuation in each province. Lumping the data from both provinces and the three earthquakes together results in an anelastic attenuation factor given by Q = 755f0.5 for frequencies, f, from 0.6 to 18 Hz, using data at distances, R, beyond 100 km from the sources and assuming geometrical spreading. This attenuation is similar to that obtained in a number of other studies in the general area. Three-component recordings were available at a few stations, and from these the following frequency-dependent ratio of horizontal-to-vertical motion was obtained: HIV = 1.14 + 0.118f − 0.00638f2, for frequencies from 0.2 to 15 Hz.
Different methods of correcting observations for attenuation, in order to derive source spectra, were examined. Source spectra derived by simple regression analyses of observed data, assuming a uniform medium, are very similar to those based on a more complicated synthetic seismogram correction, which accounts for wave propagation in a layered Earth.