Robinson, R. 1997 Recent New Zealand seismicity and models of accelerating precursory activity. Lower Hutt: Institute of Geological & Nuclear Sciences. Institute of Geological & Nuclear Sciences science report 97/27 76 p.
Abstract: Recently proposed models of accelerating seismic moment release before and around large earthquakes have been tested on three recent New Zealand earthquakes, and an attempt to identify any such regions currently ''active'' has been made. A non-linear fitting procedure, based on the Levenberg-Marquardt method, was developed in order to fit the proposed models to the observed seismicity data. The event catalogue, January 1964 - August 1996, had aftershocks removed and was limited to shallow earthquakes (depth <=40 km) of magnitude 5.0 or more. The past earthquakes used for testing the models were the magnitude 7.0 East Cape event of February 5, 1995, the magnitude 6.7 Arthur's Pass event of June 18, 1994, and the magnitude 6.7 Secretary Island earthquake of August 10, 1993. For the former two events, when the date was fixed, the theory did fit the data reasonably well for precursory regions of the appropriate radius, either centered on the mainshock or displaced by 40 to 50 km. For the Secretary Island event the results were not so good, and complicated by the nearby 1976 Milford Sound earthquake. For the East Cape event the test indicated that swarm events in the Taupo Volcanic Zone (TVZ) significantly degraded the results: in later tests the TVZ was excluded. When the data were limited to periods prior to the mainshocks, but the locations known, predicted mainshock dates and magnitudes were too late/too big for the East Cape event. For the Arthur's Pass event the results were mixed, but the real time/magnitude was usually within the estimated, rather large, errors. For the Secretary Island event the results were poor. To see if the precursory areas around these three test events could have been recognized beforehand, a grid search procedure was developed to systematically search for regions showing an accelerating moment release. Surprisingly, precursory regions near all three mainshocks were successfully identified using data through 1992. But forward predictions based on these results were generally not very good. No currently active precursory regions were defined, but a region off the SE coast of the North Island is perhaps ''suspicious''. Because of limitations to this first grid search procedure, a further search was done for larger precursory regions, for which events such as the East Cape, Arthur's Pass, and Secretary Island earthquakes are merely sub-processes. If the accelerating seismicity model is correct, then two regions would be identified as potentially precursory for very large events (M~7.5): (1) East Cape and nearby offshore areas; and (2) southwest South Island. The results for the smaller test earthquakes means that the predicted mainshock times should not be given a lot of weight, but that the identified regions may indicate the location of future large events. This study should be updated as new data become available, and in light of results elsewhere. Also, there are several clear lines of research that could potentially impro ve the results. (auth)