Popis: |
SUMMARY Observations of short-period S-P converted phases from the 670 km discontinuity (S670P) provide unique constraints on the nature of the discontinuity and the seismic structure of deep subducted slabs. Because this is a subtle phase in the P codas from deep earthquakes, great care is necessary in its identification. In past work we identified S670P phases from a large number of deep events beneath Tonga primarily by showing that the arrival times for S670P had the correct dependence upon hypocentral depth. Here we offer further confirmation of our interpretation of this S-P converted phase by showing that the occurrence of the phase is a predictable consequence of the earthquake source mechanism. We show this in two ways: (1) independently determined moment tensor solutions for the earthquakes generally predict relatively high-amplitude S-P conversions where they are observed; and (2) the observed amplitudes of S670P, ScP, PP, and P arrivals vary systematically in a way which confirms that S670P, is generated with large amplitude when the S-wave energy pattern is optimal. Comparison of ScP observations with moment tensor solutions also suggests that much of the S-wave energy results from the non-double-couple or, for CMT solutions, linear vector dipole (LVD) part of the source mechanism for the northern part of the Tonga subduction zone, where we observe the strongest S670P, phases. Because these LVD components tend to be directed along fault planes inferred from deep seismicity patterns, we suggest that they may be caused by rupture propagation directivity. We conclude that the occurrence of S670P, is primarily controlled by the source radiation pattern in a way that is consistent with our previous interpretation, thus confirming that S670P, is an S-P conversion from an almost undeformed phase change boundary at about 670 km depth. |