Potential earthquake triggering in a complex fault network: The northern South Island, New Zealand

Abstract

A synthetic seismicity computer model of multiple, interacting faults has been used to investigate potential earthquake triggering in the Alpine-Marlborough-Buller fault network, northern South Island, New Zealand. Of particular interest is whether large, characteristic earthquakes on the central 300-km-long segment of the Alpine fault (average recurrence time of 240 yr) might trigger events to the north and northeast, possible temporal clustering of other large events and stress shadows. The synthetic seismicity model generates long, homogeneous catalogues that sample a wide range of possible stress interactions and stress states allowing statistical analysis. The catalogues used here include ∼1500 central Alpine fault events of M > = 7.8 and approximately 2 000 000 other events of magnitudes down to 4.8. These other events occur on four additional segments of the Alpine fault, on 30 segments of the major Marlborough and Buller faults and on 4650 randomly distributed smaller faults. The synthetic seismicity model is of the quasi-static type, governed by the Coulomb failure criterion, with extensions to better account for rupture propagation and fault healing. True rate and state friction and viscoelastic relaxation are not yet included. Mechanical properties are adjusted so that most events rupture an entire fault segment, i.e. the faults behave characteristically. The regional b-value of ∼1 thus arises mainly from the fault size distribution. The driving mechanism and fault properties are iteratively adjusted so that the resulting long-term slip rates, single event displacements and recurrence times match the observed, real world values. The major results are as follows. (i) Large Alpine fault earthquakes, M >= 7.8, on the central segment, have recurrence times with a broad distribution, from 81 to 601 yr (average 234 yr). (ii) The regional moment release rate is constant over periods of ∼1000 yr, but is quite variable on scales of a few hundred years. (iii) Large events on the central Alpine fault will usually propagate at least part way along strike to the northeast, past the junctures with the Marlborough faults that splay off to the east. (iv) Thirty-seven per cent of such large Alpine fault events (LAFEs) will also instantaneously trigger significant activity (M >= 6.5) on one or another of the other major faults (here instantaneous means with a 0 to approximately 3 yr delay). (v) Characteristic events initiating on other major faults also have a high chance of triggering activity elsewhere. (vi) Following the short period of increased risk of triggered events, a stress shadow effect predominates, probably reflecting the mutually inhibitory nature of parallel strike-slip faults and the need for a constant long-term moment release rate. © 2004 RAS.