The Edgecumbe earthquake sequence: 1987 February 21 to March 18

Abstract

The Edgecumbe earthquake sequence consisted of more than 600 shocks of M _L ≥ 3. 0 that occurred in an area of about 70 × 50 km in the Bay of Plenty, New Zealand, between 1987 February 21 and March 18. About 130 of these occurred prior to the main shock (M _L 6. 3) on March 2. The number, distribution, and variable sensitivity of the stations of the National Seismograph Network has meant that the accuracy of the epicentres computed from data from these stations alone is excellent: by comparison with epicentres determined from temporary station data, the absolute (systematic) uncertainty of epicentres in the vicinity of the i s is estimated to be about 2 km, and the relative uncertainty of epicentres is estimated to be ± 4 km. The analysis of variance technique was applied to the magnitude data to obtain a set of self-consistent values across the whole mnge of magnitudes observed. A large discrepancy was evident between the apparent magnitudes at KRP and those at the east coast stations TUA and GNZ, probably caused by a zone of higher attenuation than allowed for in the standard magnitude model under and west of the main aftershock region. The distribution of magnitudes shows that effectively all events of M _L ≥ 3. 0 have been detected and located except during a period of about an hour after the main shock. The foreshocks occurred in two distinct zones: one just offshore from Maketu, the other near Matata and the subsequent main event. The aftershocks were similarly not confined to the vicinity of the surface faulting and main-shock rupture. The aftershock sequence may best be described as a series of swarms occurring contemporaneously with main rupture aftershocks at distances up to 35 km away. The first of these swarms commenced within half an hour of the main shock. The Gutenberg and Richter parameter b, measuring the proportions of small to large magnitude shocks, was not constant throughout the sequence but changed from b = 0. 9 ± 0. 1 prior to March 2, to 0. 7 ± 0. 1 for the frrst 2 h of the sequence, and then increased logarithmically to 1. 22 ± 0. 1 by Mattch 18, a value close to the long-term background value of b = 1. 3. This is similar to the usual pattern of b-value change in aftershock sequences and has been interpreted as indicating a high level of stress prior to, and during, the main shock, with the stress level returning to normal during the aftershock sequence. The change of b-value fits a Jeffreys-Lomnitz creep law with the same time constants as obtained for the postseismic creep on the Te Teko segment of the fault break. The low b-value for the foreshocks is one of the best such observations that has been made, but with the foreshocks being spatially distributed as they were, it could not have been used to predict the location of the main shock.