Issues

A paucity of digital seismic data from eastern North America has in the past necessitated a reliance on stochastic models to predict ground motion from large eastern events (e.g. Boore and Atkinson, 1987). A key assumption in applications of this technique has been that a stress drop of 100 bars is appropriate for eastern events. Although Somerville et al. (1987) conclude that there is no systematic difference in stress drop between inter- and intraplate events in North America, other studies have concluded that events in stable cratonic environments do have higher stress drops than those along plate boundaries (e.g. Scholz et al., 1986). Because this issue remains unresolved, it is important to test the stochastic model predictions with comparisons to newly-collected digital data from eastern North America, as Atkinson (1990) has done.

As noted by Atkinson (1990), there is no doubt that theoretical models underpredict the observed ground motions from the M=6, 11/25/88 Saguenay, Quebec event, especially for frequencies around 5 Hz. This earthquake was the largest event to occur south of 50 degrees latitude in eastern North America in over 50 years, providing the only observational constraints to-date for events of this magnitude. Atkinson suggests that the model predictions are validated because, while the Saguenay amplitudes are underpredicted, amplitudes of the magnitude 6.8 Nahanni earthquake are overpredicted.

Clearly, it is impossible at this point to assess whether the Saguenay event is representative or exceptional. Questions have been raised both about the high-frequency radiation at the source and about . . .

I thank Susan Hough, Leonardo Seeber and Klaus Jacob for the opportunity to discuss the implications of the Saguenay ground motion data for ENA ground motion relations. Perhaps the most contentious issue of the ground motion relations developed by Dave Boore and myself has been our assumption that the seismological source can be modeled by a simple ‘constant stress drop’ model, with a mean stress parameter of 100 bars for eastern events (Atkinson, 1984; Boore and Atkinson, 1987; Atkinson and Boore, 1990). (This compares with a stress parameter value of 50 bars for western events (Boore, personal communication; Somerville, personal communication).) The adoption of any simple model to characterize the ‘average’ seismological source parameters will inevitably lead to significant discrepancies in the case of individual events, Saguenay being an excellent example. Because earthquakes are subject to great inter-event variability, it is not necessarily the particular model or value that is selected which causes disagreements between theory and observation. A single earthquake can neither validate nor disprove a median prediction.

Hough et al. have two specific comments regarding the analyses and conclusions presented in my paper:

• they believe that the NCEER strong motion data were excluded on the basis of an incorrect assumption, and that these excluded data might change the conclusions reached; and

• they believe that Saguenay is more ‘typical’ of large ENA events than Nahanni; including Nahanni in the ground motion data set has the effect of downweighting the Saguenay data, which they feel is unjustified. . . .