New observation on Seismic anisotropy in subducting slabs inferred by non-double-couple earthquake radiation patterns

Nearly one third of earthquakes are deep earthquakes with focal depths exceeding 60 km. Most of them occur in subduction zones and show non-double couple (non-DC) radiation patterns. Previously, this non-DC radiation pattern was explained in an ad hoc fashion event by event. However, in this work, we found that these non-DCs could be satisfactorily explained by assuming that (i) all earthquakes are shear dislocation faulting; (ii) they are embedded in a common but tilted transversely isotropic (TTI) medium. The TTI anisotropy is characterized by the symmetry axis (two angles) and 3 Thomsen parameters among which the S-wave anisotropy is best resolved by our algorithm. The S wave and P wave velocities parallel to the symmetry axis are derived from the preliminary reference earth model (PREM). We used the centroid moment tensors (CMT) from the Harvard's catalog to invert for both the TTI symmetry axes and anisotropy strengths for global subduction zones: Tonga, Java, Molucca, Vanuatu, Mariana, Japan, Kuriles, and Aleutians. Our inversion results show that in nearly all studied regions (19 earthquake groups) the TTI symmetry axes are perpendicular to the subducting slab interface but with a few exceptions (3 earthquake groups) where several deep-focus regions have symmetry axes parallel to the subducting slab interface. For all subduction zones and for all depths (100km to 650km), the inverted S-wave anisotropy has a typical value of about 30% (ranging from 23% to 40%) and we observed no clear dependence on depth. This shows that the rock fabrics hosting the intermediate-depth and deep-focus earthquakes are no different, pointing to a possibility that there is no difference between intermediate-depth and deep-focus earthquakes in terms of their mechanisms. Our anisotropy observation is independent of slab subduction rate, age, and dip angles, etc. Given the global nature of our observation, consistency of the anisotropy, and large S-wave anisotropy, we postulate a common process in all subduction environments. Such a process may be the metamorphic reaction in the slab, which can generate sheet silicates with layered fabrics associated with large anisotropy values. Our inverted anisotropy provides important information on mechanisms of the intermediate-depth and deep-focus earthquakes.