Intermediate-focus earthquakes are known to show features such as non-double-couple focal mechanisms that indicate source complexities. To characterize these features, we have systematically studied the low-frequency radiation from 108 intermediate-focus earthquakes recorded by high-performance seismic networks from 1989–1997 whose total moment was > 3 × 1018 N m. We determined frequency-dependent focal mechanisms and source phase and amplitude spectra for each earthquake, estimating the uncertainties for all parameters. Frequency-dependent focal mechanisms were obtained from vertical-component, free-oscillation data in 1-mHz bands over the range of 1–11 mHz. Source amplitude and phase-delay spectra were determined to 20 mHz from a combination of free-oscillation and surface-wave data. The population of intermediate-focus earthquakes in our catalog is not equally divided between compressional and tensional stress-release mechanisms; instead, 59% are down-dip tensional, 25% are down-dip compressional, and 16% are neither. We have assessed the statistical significance of any non-double-couple component of the source for every earthquake in the catalog. We represent a deviatoric focal mechanism by its principle axes and a scalar ξ that ranges from –1, for a compensated linear vector dipole (CLVD) with a compressional axis of symmetry through 0 for a double couple, to +1, for a CLVD with a tensional axis of symmetry. ξ varies from –0.91 to +0.64 in this dataset. Fifty-six of the 108 events have a significant CLVD component (ξ ≠ 0) at the 95% confidence level; 15 events have an unusually large CLVD component of the focal mechanism (|ξ| ≥ 0.4). We do not observe a correlation between the CLVD component and seismic moment or depth. T-type CLVD mechanisms correlate weakly with slab stress state, but no correlation was similarly found for P-type CLVD events. Seven earthquakes in the catalog are slow or compound, and three show strong frequency dependence of the moment tensor.

You do not currently have access to this article.