Dating of past Alpine Fault rupture in south Westland

Technical Abstract

We present the results of a detailed investigation of the Alpine Fault in the area between the Haast and Okuru Rivers, South Westland. This study builds upon previous investigations in order to develop a better chronology of large earthquakes generated by ruptures on the Alpine Fault.

Work during the 1990s has established that the Alpine Fault is a major source of potential seismic hazard and incorporation of data from the fault into seismic hazard maps has greatly changed the perception of earthquake hazard in the South Island.

A number of outstanding problems remain in regard to the Alpine Fault, two of which are:

• How does the chronology of fault ruptures compare for different sections of the fault, particularly for northern and southern sections is there evidence for the whole fault rupturing in a single magnitude 8 event or may it break in two or more closely spaced smaller events affecting different sections of the fault?
• What is the long-term (i.e. several thousand years) chronology of fault ruptures? Is the behaviour of the fault relatively constant through time and can we use this information to derive better estimates of future probability of ruptures and the accompanying level of hazard?

The project involved four principal avenues of investigation:

• Detailed mapping of the fault trace to determine the nature of the fault at the surface, the number of traces (if more than one), offset features such as channels, river terraces, etc. and amount of offset where determinable, and the location of suitable sites for detailed paleoseismic investigations.
• Excavation of a trench at the Turnbull River, a site identified during the fault mapping, in order to expose the fault and the sediments accumulated on the downthrown side of the fault. From these, it was hoped to identify past rupture events and constrain their age using radiocarbon dating.
• Collection of rounds from stumps of trees along the fault scarp felled during the 1960s and 70s and, using ring-counting techniques, to establish the approximate ages of forest disturbances possibly associated with ground rupture events.
• A further investigation was carried out on a rapidly subsiding swamp situated between two strands of the Alpine Fault on the south bank of the Turnbull River. The intention was to see whether the swamp stratigraphy could be linked to fault displacements and hence be used as a record of ground ruptures in the area.

All four investigations were successfully carried out and a wealth of detailed information obtained. Unfortunately, unforeseen problems with obtaining meaningful radiocarbon ages on some of the trench material and intense weathering of the trees stumps prevented us obtaining the tight age constraints we had hoped for. Nevertheless, the data obtained substantially improved knowledge of rupture history of the fault in the south and opened up further possibilities for addressing the questions raised above.

Results from the four approaches may be summarized as:

• Detailed mapping of the fault is essentially a single straight trace between the Haast and Turnbull Rivers. Slightly changes in orientation are associated with en echelon traces that overlap each other and are offset from a few to a few tens of meters. Offset channels show displacements that cluster crudely around 8 and 15m, possibly reflecting the amount of the last, and combined penultimate and last, ruptures respectively. South of the Turnbull River, the surface faulting is more complex with multiple parallel traces in places.
• The site selected for the trench at the Turnbull River was the only suitable site identified. A 22m long and 2-3m deep trench was excavated across the fault. There was clear evidence for three fault displacements within the last 1000 years and radiocarbon dating of the lowest horizon placed a maximum age on the earliest event of 1105 +/- 85 AD. Unfortunately natural contamination of organic material faulted by the penultimate event resulted in widely variable and inconsistent radiocarbon ages and prevented better age constraints being established. Deposits related to the last event had been highly disturbed during the forest clearance and road construction.
• Although over 20 samples of trees (Rimu) were collected and prepared for tree-ring counting, loss of sapwood on most of these, due to the long period of exposure in an inclement climate, prevented their use in establishing well-constrained times of forest disturbance related to possible fault rupture.
• Detailed investigation of the South Turnbull Swamp established a rapid rate of subsidence over the past 1200 years commensurate with that expected from models of fault controlled subsidence. Inundations of river-derived silt possibly reflect fault ruptures. A major inundation shortly before 1260 +/- 55 AD is almost certainty due to fault displacement. Three others since then may relate to earthquakes but it is possible to generate additional bands through other processes. Nevertheless, swamps such as this potentially contain valuable long records of sedimentation related to fault displacement.

By reviewing our new data in conjunction with existing data, we can conclude:

• Three major ground rupturing events have occurred on the Alpine Fault ion South Westland over the last 1000 years. All three had dextral offsets of approximately 8 metres. The first occurred between 1105 and 1220 AD and may correlate with a similar event in central and north Westland. The second occurred between 1400 and 1600 AD, but its correlation with northern events, if any, is not established. The last event occurred after 1665 AD and before 1826 AD and may correlate with a fault-wide rupture at 1717 AD.
• An earlier event occurred around 880 +/- 110 AD and may correlate with a widespread event to the north at around 900 AD. A major earthquake recorded by sailors in 1826 appears to have no direct fault break in the present area. The break may have extended onto land or was confined to offshore segments of the fault. The best approach for dating this and the post 1665 AD event may be dendrochronological studies on silver beech trees along the fault scarp.
• The Alpine Fault is a major potential seismic hazard throughout the South Island. Although this work has advanced our knowledge of the fault history, the questions raised at the start still require further research.