Indicators of paleoseismic activity along the western Hope Fault

Technical Abstract

The report presents the results of a comprehensive study of paleoseismic, landscape features, and dendrochronologic data to assess the timing of the most recent faulting events along the Hurunui section of the Hope Fault, South Island. Geomorphic observations supported by radiocarbon dates and relative soil age estimates show that the investigated sites along the fault in the upper Hurunui and Hope areas are all late Holocene in age.

Several radiocarbon ages are used to define the age of events in the landscape. Data from a hand-dug trench at Matagouri Flat showed evidence for at least 2 faulting events. The most recent faulting event there is dated from a colluvial wedge developed across a fresh fault scarp, and probably occurred since AD 1458-1655. The age of this event is supported by other radiocarbon ages on buried or fallen trees. This wedge appears to be weakly reorganised, perhaps suggesting some minor, sympathetic refaulting since the most recent faulting event. The penultimate faulting event is recognised by warped sand and paleosol units toward the base of the trench, and probably occurred before AD 1297-1425.

One of the main objectives of the research was to determine the timing and history of forest change along the fault, as a test of the fault rupture data. Approximately 250 Red, Silver and Mountain Beech (Nothofagus) trees were cored at 14 tree plots at selected alluvial and fault scarp sites. Ten of the plots were located in the west close to data from our paleoseismic and slip rate sites at Matagouri Flat and McKenzie Stream. The other 4 plots were located farther east near the western end of the AD 1888 North Canterbury earthquake rupture.

In general, we observed that the forest had an even (sized?) appearance comprising trees ranging in size from 20-70 cm in diameter, giving core ages of 90-280 yr. Three main peaks of tree colonisation are indicated by the tree dataset. The youngest peak occurred near the turn of the 20th century, and is interpreted to be related to damage from rupture of the Hope River segment of the Hope Fault in the 1888 earthquake. At one site in particular (Plot 3), we were able to demonstrate that it took c.21 ± 4 yr time to recolonize this open canopy site. Five of the pots show a strong signal of this event. An older peak in the tree ring data occurred at c.150 yr ago. Based on the 21 yr colonisation offset, this second peak has been attributed to an earthquake of age c.AD 1830 ± 5 yr. Eight of the tree plots show evidence of this event. The oldest peak occurs at c.230-240 rings (yr) ago and is attributed to an earthquake at c.AD 1760 ± 5 yr. Similarity diagrams of the tree plots show that each of the three peaks is observed at specific plots that have clustered age data. Another minor tree age peak may precede this one. We infer that such a peak is evidence of shaking from the AD 1717 rupture of the Alpine Fault in at Plot 4 in our study area.

One of the clearest results from the dendrochronologic data is that the forest has not undergone wholesale change through damage since at least the time of the 1888 earthquake. It appears that the 1929 Arthur’s Pass earthquake had little or no impact at the plots we investigated. However, before this there are up to 4 instances (1888, c.1830, c.1760 and c.1717) where the forest underwent significant change through tree loss and recolonization. The trench and landscape records indicate 2 and possibly 3 earthquake ruptures between c.AD 1460-1888, while the forest change record indicates 3 and possibly 4 major forest damaging events during the period AD 1717-1888 (171 yr). These give simple recurrence ranges of 214-428 yr and 57-85 yr, respectively.

The difference between these two recurrence ranges is related to the recurrence of surface faulting earthquakes along the Hurunui section of the Hope Fault, compared to the return time for Intensity VIII or greater shaking throughout the study area. In other words, the forest plots we investigated have been subject to strong shaking and other agents of forest change besides simply the single ‘Hurunui’ fault source. The short recurrence period exhibited by the forest change data is indicative of repeated, high intensity damage by earthquakes in this very active part of the South Island plate boundary.

There is a considerable level of uncertainty about the results of this study, mainly because it is difficult to inter-relate the two types of data produced by this study, ie, imprecisely aged but accurate (paleoseismic) results from trenching and landscape modification, versus, precisely aged but less easily interpretable data from forest change. We infer that large earthquakes are the strongest agents of forest change at the sites we studied and other agents such as storm, windthrow, and fire were of much lower importance.

A new model of synthetic seismicity was developed to test the effects of rupture probability in the western Marlborough Fault System following an Alpine Fault rupture. The elastic Coulomb Failure Stress models give low probabilities of triggered ruptures along the western Hope Fault within 3 years of an Alpine Fault rupture. This is in agreement with our forest damage results that indicate there are (several years to) decades between strong shaking and/or surface faulting, large earthquake events.

This project has been instrumental in demonstrating the value of dendrochronological studies along the very active faults of the Marlborough Fault System. In future, larger, more comprehensive datasets need to be collected to provide more robust interpretations of tree data that are discussed in this report. In addition, studies of tree growth suppression may be a useful means to determine the age of damaging events in these areas.