On the spatio-temporal distribution of M 7.0+ worldwide seismicity with a non-parametric statistics

The aim of this paper is to provide some constrains on the time behavior of earthquake generation mechanism, through the usage of a non-parametric statistics that leads up to the empirical estimation of the hazard function. The results indicate that the most characterizing temporal feature for large (M 7.0+) worldwide shallow earthquake occurrence is a clustering lasting few years, indicating that the probability of earthquake occurrence is higher immediately after the occurrence of an event. After that, the process becomes almost time independent, as in a Poisson process. Remarkably, this time clustering is very similar to what previously found for different spatio-magnitude windows, and it does not seem to depend on the tectonic style of the region. This may support the hypothesis of an universal law for earthquake occurrence.