A test for checking earthquake aperiodicity estimates from small samples
In recent years, many new models for earthquake recurrence were proposed. Some are focusing on the clustering properties on a small time scale, while others try to model the long term behavior of large mainshocks. To this last purpose, there is a growing interest for models that take into account the aperiodicity aiming to a time-dependent hazard estimate. It is well known that a limited number of inter-event times (IETs) may lead to biased values of the distribution parameters. To overcome this problem different solutions were proposed. This paper focuses on two of them: Monte Carlo simulation of the process and aperiodicity estimated via a statistical proxy. The topics discussed are: 1) how many IETs are needed for a correct estimate, 2) to which extent a Poisson distribution is equally able to describe the process, 3) the influence of errors associated to paleoseismological IETs, and 4) the goodness of the success ratio from simulations. A simple test is proposed to discriminate real aperiodicity from apparent aperiodicity coming from undersampling.