Use of precariously-balanced rocks to test the New Zealand seismic hazard model: A pilot study

Abstract

We undertake the first New Zealand-based pilot study to investigate the use of ancient precariously-balanced rocks, (rocks that are unstably balanced on top of a pedestal) as a criterion for testing the estimates of earthquake shaking from probabilistic seismic hazard models for long return periods. To date, research to test seismic hazard models in New Zealand have been restricted to the short historical record of earthquakes. Our survey of three sites in central Otago, a site in northwest Nelson, and a site in eastern Fiordland has yielded a total of 28 precariously-balanced rocks which, on the basis of established USA-based methodology, are used to provide estimates of the maximum ground motions that have occurred at the sites since the rocks became precarious.

Estimates of the age of the precariously-balanced rocks (10,000 to 55,000 years) are made from considerations of the geologic and climatic history of the sites, and from a cosmogenic date obtained from bedrock removed from the pedestal of a precariously-balanced rock at one of the central Otago sites. Comparison of the maximum peak ground accelerations and ages of the precarious rocks to seismic hazard curves derived from the New Zealand National Seismic Hazard Model show that the central Otago rocks provide lower estimates of peak ground accelerations than the hazard model, for equivalent return periods. In contrast, the northwest Nelson rocks correlate well with the precarious rock data. No precarious rocks were found in eastern Fiordland, consistent with the obviously-active alpine erosion processes, and regular occurrence of large earthquakes in the region. The difference between the central Otago and northwest Nelson sites is that the former are all located within 5 km of active faults, whereas the northwest Nelson site is not. Peak ground accelerations calculated with the assumption of median ground motions for these faults produce hazard curves that compare favourably with the precarious rock data. 

The variability about the median estimates of PGA for the fault sources, and/or the median PGA for the fault sources may therefore be overestimated for the central Otago sites. The implications and limitations of the pilot study for providing constraints on seismic hazard estimates are such that we propose that several aspects of follow-up work be pursued.