ATKINSON, GAIL M., Dept. of Earth Science, Carleton University, Ottawa, Ont. K1S 5B6

Modified Mercalli Intensity (MMI) data are the only available ground motion data for most of the large historical earthquakes that have occurred in eastern North America (ENA). Are these historical observations consistent with expected intensity levels based on regional ground motion relations? It is important to answer this question, as it bears on the reliability of the results of seismic hazard analyses. An empirical study of California strong motion records and associated MMI values has shown that observed MMI is a function of response spectral amplitudes, magnitude, and distance (Atkinson and Sonley, 1999). Comparison of strong motion and MMI data from the 1988 Saguenay, Quebec, earthquake confirms applicability to ENA, at least for epicentral distances of 150 km or less. I therefore apply the relationships of Atkinson and Sonley (1999) to predict the MMI for historical ENA earthquakes as a function of distance, assuming that the response spectral amplitudes follow those of the ENA ground motion relations of Atkinson and Boore (1995). The ground motions are adjusted to show the expected MMI for a range of soil conditions, including hard rock, firm ground and soft soil. The predicted MMI values for firm ground to soft soil conditions are consistent with historical observations from most ENA earthquakes, including the 1811-1812 New Madrid earthquakes (M 8), the 1940/12/20 earthquake of M 5.4, the 1944 Cornwall-Massena earthquake of M 5.8 and the 1982 Miramichi earthquake of M 5.5. Observed intensities from the 1886 Charleston earthquake (M 7.3) and the 1925 Charlevoix earthquake (M 6.4) are weaker than predicted, by about 1 intensity unit, while observed intensities from the 1968 Illinois earthquake (M 5.4) are larger than predicted, by about 2 intensity units. This suggests that the 1968 Illinois earthquake, like the 1988 Saguenay earthquake, had high-frequency ground motions stronger than predicted by the Atkinson and Boore (1995) ground motion relationships. On the other hand, the high-frequency motions from the 1925 Charlevoix and 1886 Charleston earthquakes appear to be overestimated by the Atkinson and Boore (1995) relations. An alternative explanation for the 1886 Charleston earthquake could be that the magnitude was significantly smaller than 7.3; a magnitude of about 6.7 would reconcile the intensity data with the ground motion predictions. By contrast, the intensity data of the New Madrid earthquakes of 1811-1812 support a magnitude of about 8, while a magnitude of 7.5 for these events would result in underprediction of the observed intensities. On the whole, it is concluded that the historical MMI data are consistent with currently-used ENA ground motion relationships.