What can the limited and uncertain geological record tell us about the earthquake cycle?

The geological record of past earthquakes on a fault provides a basis for forecasting the probability of another earthquake occurring within some future timeframe. Yet paleoearthquake records are typically limited to the most recent few events, and dating uncertainties are often large. This creates uncertainty in the application of statistical models to these data, both in terms of model parameterisation and in the choice of model itself. Consequently, there are challenges linking observations of large earthquake recurrence to theoretical models of the earthquake cycle.

In this study we use paleoearthquake records from more than 90 faults globally to investigate the earthquake cycle and how it varies across different tectonic regions, fully accounting for data uncertainties. We find that earthquake recurrence is weakly to moderately periodic for most faults, while low activity-rate faults exhibit more strongly aperiodic recurrence behaviour. Fitting four different renewal models (Weibull, gamma, lognormal and Brownian passage time distributions) to the data, we show that there is no single model that universally best describes earthquake recurrence. We find that diversity in recurrence characteristics exists both between different tectonic regions and for different fault segments within the same fault system. Finally, we investigate how observations of cumulative fault displacements due to multiple earthquakes can help constrain earthquake cycle models when paleoearthquake data is limited and uncertain.