Quantifying the hazard from New Zealand's shallow intraslab earthquakes

Abstract

One of the major challenges in modelling seismic hazard is how to deal with hidden faults – that is, those with no surface expression. This is especially so in subduction zones, where earthquake faulting can occur not only in the overlying plate and at the plate interface, but also within the lower subducted plate. In recent years, there has been a growing appreciation that faults in the subducted plate can contribute significantly to the seismic hazard in the shallow parts of subduction zones. Because the shallow part of the Hikurangi subduction zone underlies or is close to the east coast of the North Island of New Zealand, it is important to quantify the hazard that such faults pose to population centres in this region.

Here we use new results on the three-dimensional structure of the Hikurangi subduction zone determined from seismic tomography (the earthquake wave equivalent of a medical CAT scan) to better define larger faults in the crust of the subducted Pacific plate. We look at faults that moved in eight large earthquakes, from the magnitude 6.8 Wairarapa earthquake in August 1942 through to the magnitude 5.5 Hastings earthquake in August 2008. We find that faults in the subducted plate get larger with increasing depth below the plate interface. These larger faults were formed when the subducted plate began to bend at the offshore trench. The faults are reactivated as pressure and temperature increase in the subducted plate. Rupture of the faults initiates at rough spots on the fault where a geochemical process known as dehydration embrittlement takes place. This model explains why aftershocks for such events are sparse, and die out very quickly.

Our modelling of these faults in the subducted plate provides an explanation for the type of shaking that earthquakes on the faults produce at the surface. In addition, the distribution of these larger earthquakes in the subducted plate is consistent with a model derived from small earthquakes that suggests that seismicity in the subducted plate may be modulated by the state of locking of the overlying plate interface. In other words, this study has given us a better understanding of how, why and where larger earthquakes will occur in the subducted plate at the Hikurangi subduction zone. It provides important new information for refining seismic hazard models in this region.

---

Technical Abstract

We investigate the habitat of large intraslab earthquakes shallower than 50 km in the shallow part of the Hikurangi subduction zone in New Zealand by relocating such earthquake sequences which occurred during the period 1942-2008, using a new three-dimensional seismic velocity model. Three of the largest intraslab earthquakes coincided in space and time with large earthquakes in the overlying plate or significant slow slip at the plate interface. We find that faults get larger with increasing depth (and increasing seismic velocity) below the plate interface. We also find that average stress drops, as estimated from aftershock zone dimensions, decrease with increasing depth below the plate interface. Given that fault slip inversions suggest high stress drops for deeper intraslab events, our results suggest a model where these earthquakes involve rupture of a localized asperity (the locus of dehydration embrittlement) on an otherwise weak, reactivated fault. This model explains why aftershocks for such events are sparse, and die out very quickly.

We also use the distribution of these larger intraslab earthquakes to test a model derived from small earthquakes that suggests that intraslab seismicity may be modulated by the state of coupling of the overlying plate interface. The spatial distribution of the larger intraslab earthquakes appears consistent with this model. An important seismic hazard implication of this is that there may be a time-dependence of large intraslab events related to plate coupling. If a large rupture on the strongly coupled part of the plate interface allows the egress of fluid from the subducted slab, this may stimulate previously suppressed intraslab earthquakes.