Characteristics of well-determined non-double-couple earthquakes in the Harvard CMT catalog

Abstract

This paper evaluates earthquakes in the Harvard centroid moment tensor (CMT) catalog which possess well-determined, deviatoric non-double-couple mechanisms, i.e. mechanisms with significantly large compensated linear vector dipole (CLVD) components. For a CMT M , we measure the strength of the non-double-couple component with the statistic Γ norm = 2.6det[M] | M | 3 , which is 0.0 for a pure double couple and ±1.0 for a pure CLVD. For each CMT, we measure whether Γ norm is well determined by allowing individual elements of M to vary within their range of uncertainties as reported in the Harvard catalog, then determining the range of variation of | Γ norm| as M varies. The most well-determined 11% of the catalog has a range for | Γ norm| of 0.2 or less; of these earthquakes, about one-quarter have predominantly CLVD mechanisms with | Γ norm| of 0.4 or greater—these 283 mechanisms make up the principal data for this study. These CLVD earthquakes occur commonly in almost all tectonic environments — along mid-ocean ridges and transforms, in shallow subduction zones, within Wadati-Benioff zones, and in the interiors of plates in both oceanic and continental regions. Along ridge-transforms, the majority have horizontal T axes; in shallow subduction zones the majority have horizontal P axes which are approximately perpendicular to the trench. These observations are consistent with a schematic model which explains CLVD mechanisms as the sum of two or more elementary, double-couple subevents with orientation fixed by the type and geometry of the plate boundary. Within ridge-transforms, subduction zones, and Wadati-Benioff zones where the plate motion, lithospheric structure, etc., is sufficiently uniform, about 75% of the CLVD mechanisms conform to this model.