Definition of the locked zone of the shallow plate interface in the Hikurangi subduction zone using clusters of small earthquakes

Abstract

Beneath the eastern North Island, the Pacific plate is being thrust beneath the Australian plate in what is known as a subduction zone. In such zones, the fault between the plates often locks up where it is between about 10 km and 30 km deep. As the plates continue to converge, this leads to a build up of strain which is eventually released by a large earthquake which ruptures the fault. The magnitude 9.3 Sumatran earthquake on Boxing Day 2004 was such an event. The magnitude of the earthquake produced depends on the area of the fault that locks up. So if we can determine this area, we can get a good estimate on the magnitude of future earthquakes between the plates that will impact the eastern North Island.

Here we test a concept developed in Japan that clusters of small earthquakes on the fault between the plates can provide an estimate of this locked area. The idea is that outside this locked area, the plates are more or less slipping freely. However, if there are a few rough spots on the fault in this slipping area, they will repeatedly break causing clusters of small earthquakes. So where we see small earthquake clusters we would not expect the plates to be locked – locking would be more likely in areas without clusters. We find that this concept does not seem to work in the North Island. The small earthquake clusters that we identify are mostly in the top of the lower plate, rather than on the fault between the plates. Nevertheless, these clusters do seem to concentrate beneath the edges of the locked zone of the fault between the plates, as determined with accurate GPS measurements. Such small earthquake clustering is more intense in the Wellington region than in Hawke’s Bay, suggesting that currently the plates are more strongly coupled in the Wellington region.

When small earthquakes in a cluster are very close together, the seismic waves they produce are often very similar. We use this fact to accurately locate earthquakes within clusters. This shows up faults that are moving in the top of the lower plate. In particular, we can clearly see the faults that moved during the earthquakes that occurred beneath Upper Hutt in 2004 and 2005. Movement on these faults in turn provides information on how the top of the lower plate responds to the locking of the fault between the plates. Taken together, information we have gleaned from small earthquake clusters corroborates the pattern of locking of the fault between the plates beneath the eastern North Island determined from accurate GPS measurements.

---

Technical Abstract

Subduction thrusts produce the largest and potentially the most destructive earthquakes and tsunamis worldwide. The size of such earthquakes scales with the area of the plate interface which locks up during the interseismic period. It is thus important to estimate the limits of this locked zone, particularly in New Zealand where our short historical record provides no clues as to the size of such events. Here we investigate whether we can use clusters of small earthquakes at the plate interface to provide constraint on the limits of the locked part of the interface. Recent overseas research (particularly in northeastern Japan) has shown that persistent clusters of small earthquakes can be used to map the transition between locked and creeping parts of the plate interface.

To robustly define clusters, we implement high-precision waveform cross-correlation and relative arrival time (“double-difference”) techniques. We find that the Japanese model of clusters of repeating small earthquakes being indicators of quasi-static slip on the plate interface does not seem to be applicable to the Hikurangi subduction zone. We observe earthquake clusters within the crust of subducted plate, rather than on the plate interface. This difference in behaviour may be due to different frictional conditions at the plate interface arising from thicker than normal subducted crust in the Hikurangi subduction zone. Alternatively, because our results may be more precise than those from Japan (where most earthquake clusters occur well outside the seismograph network), our results may represent a more realistic model of the relationship between plate coupling and small earthquake clustering. The earthquake clustering that we do see within the subducted crust appears to concentrate beneath the edges of that part of the plate interface which is thought to be strongly coupled from GPS measurements.

In Hawke’s Bay, earthquake clustering within the subducted crust is less intense than that in the Wellington region. Also, small thrust earthquakes at the plate interface occur throughout the seismogenic zone in Hawke’s Bay. Both these observations are consistent with GPS results which suggest that the plate interface is more strongly coupled in Wellington than in Hawke’s Bay. Earthquakes in the crust of the subducted plate in the Wellington region beneath the edges of the locked zone of the plate interface show mechanisms consistent with plate unbending, while those beneath the locked zone show mechanisms consistent with plate bending. Unbending earthquakes may thus prove useful in further defining the limits of the locked zone of the plate interface. Some earthquake clustering within the crust of the subducted plate appears to be temporally related to slow slip events at the plate interface in the surrounding region. This suggests a novel method for defining the locked zone of the plate interface – tracking the regional development of seismicity during such slow slip.

Double-difference relocation of earthquakes has significantly sharpened up the distribution of seismicity in the Wellington region, and revealed individual faults at depth. We have identified the faults involved in both the April-May 2004 Upper Hutt swarm and subsequent January 2005 ML 5.5 mainshock. This has provided insight into the physical relationship between mainshocks and precursory swarms.