Variability of recurrence interval for New Zealand surface-rupturing paleoearthquakes

Abstract:

Recurrence interval (RI) for successive earthquakes on individual faults is recorded by paleoseismic datasets for surface-rupturing earthquakes which, in New Zealand, have magnitudes of >Mw ~6 to 7.2 depending on the thickness of the brittle crust. New Zealand faults examined have mean RI of ~130 to 8500 yrs, with an upper bound censored by the sample duration (<30 kyr) and an inverse relationship to fault slip rate. Frequency histograms, probability density functions (PDFs) and coefficient of variation (CoV₌ standard deviation/arithmetic mean) values have been used to quantify RI variability for geological and simulated earthquakes on >100 New Zealand active faults. RI for individual faults can vary by more than an order of magnitude. CoV of RI for paleoearthquake data comprising 4-10 events ranges from ~0.2 to 1 with a mean of 0.6±0.2. These values are generally comparable to simulated earthquakes (>100 events per fault) and suggest that RI ranges from quasi periodic (e.g., ~0.2-0.5) to random (e.g., ~1.0). Comparison of earthquake simulation and paleoearthquake data indicates that the mean and CoV of RI can be strongly influenced by sampling artefacts including; the magnitude of completeness, the dimensionality of spatial sampling and the duration of the sample period. Despite these sampling issues RI for the best of the geological data (i.e. >6 events) and earthquake simulations are described by log-normal or Weibull distributions with long recurrence tails (~3 times the mean) and provide a basis for quantifying real RI variability (rather than sampling artefacts). Our analysis indicates that CoV of RI is negatively related to fault slip rate. These data are consistent with the notion that fault interaction and associated stress perturbations arising from slip on larger faults are more likely to advance or retard future slip on smaller faults than visa versa.