Seismic records from the GDSN (SRO, ASRO, DWWSSN, RSTN) and IDA networks were used to obtain long-period (T > 45 sec) focal mechanisms for 35 earthquakes in Western North America (longitude: −140° to −110°; latitude: +30° to +60°) employing the centroid-moment tensor method of Dziewonski et al. (1981) and Dziewonski and Woodhouse (1983). The same algorithm was applied to all events, the smallest (M0 = 6.4 × 1023 dyne-cm) being an aftershock to the 1979 Imperial Valley, California, earthquake and the largest (M0 = 8.5 × 1026 dyne-cm), the 1980 Eureka, California, earthquake. All events were analyzed with and without the constraint Mrθ = Mrφ = 0 on the moment tensor; an ad hoc criterion on the relative variance reduction was employed to determine a favored solution. Our solutions show general agreement with other published results, as well as with tectonic structures. Agreement with results obtained using the abundant local and regional data confirms the adequacy of the method in determining long-period source parameters and verifies its applicability in regions where local and regional data are scarce or nonexistent.
In some cases, disagreement between our solution and results from first motion studies appears to be related to source complexities. Our moment tensor represents the total stress release of the composite event.
A consistent westward movement of the centroid locations reflects lateral heterogeneities in the crust and mantle.