THE 1986 KERMADEC EARTHQUAKE AND ITS RELATION TO PLATE SEGMENTATION

To evaluate the tectonic significance of the October 20, 1986 Kermadec earthquake (M(W) = 7.7), we performed a comprehensive analysis of sou rce parameters using surface waves, body waves, and relocated aftersho cks. Amplitude and phase spectra from up to 93 Rayleigh waves were inv erted for centroid time, depth, and moment tensor in a two-step algori thm. In some of the inversions, the time function was parameterized to include information from the body-wave time function. The resulting s ource parameters were stable with respect to variations in the velocit y and attenuation models assumed, the parameterization of the time fun ction, and the set of Rayleigh waves included. The surface wave focal mechanism derived (phi = 275-degrees, delta = 61-degrees, lambda = 156 -degrees) is an oblique-compressional mechanism that is not easy to in terpret in terms of subduction tectonics. A seismic moment of 4.5 x 10 (20) N-m, a centroid depth of 45 +/- 5 km, and a centroid time of 13 /- 3 s were obtained. Directivity was not resolvable from the surface waves. The short source duration is in significant contrast to many la rge earthquakes. We performed a simultaneous inversion of P and SH bod y waves for focal mechanism and time function. The focal mechanism agr eed roughly with the surface wave mechanism. Multiple focal mechanisms remain a possibility, but could not be resolved. The body waves indic ate a short duration of slip ( 1 5 to 20 s), with secondary moment rel ease 60 s later. Seismically radiated energy was computed from the bod y-wave source spectrum. The stress drop computed from the seismic ener gy is about 30 bars. Sixty aftershocks that occurred within three mont hs of the mainshock were relocated using the method of Joint Hypocentr al Determination (JHD). Most of the aftershocks have underthrusting fo cal mechanisms and appear to represent triggered slip on the main thru st interface. The depth, relatively high stress drop. short duration o f slip, and paucity of true aftershocks are consistent with intraplate faulting within the downgoing plate. Although it is not clear on whic h nodal plane slip occurred, several factors favor the roughly E-W tre nding plane. The event occurred near a major segmentation in the downg oing plate at depth, near a bend in the trench, and near a right-later al offset of the volcanic arc by 80 km along an E-W direction. Also, a ll events in the region from 1977 to 1991 with CMT focal mechanisms si milar to that of the mainshock occurred near the mainshock epicenter, rather than forming an elongate zone parallel to the trench as did the aftershock activity. We interpret this event as part of the process o f segmentation or tearing of the subducting slab. This segmentation ap pears to be related to the subduction of the Louisville Ridge, which m ay act as an obstacle to subduction through its buoyancy.