Multifractal analysis of earthquake catalogues

We show that standard statistical analyses of earthquake Catalogues are not very useful tools for describing temporal variations of the seismic event clustering. A more appropriate approach is the rnultifractal analysis. In fact, if interarrival times between earthquakes are fractally distributed, this means that the behaviour of the catalogue is scale-invariant and it can be described in terms of a Poissonian or a stationary clustered distribution. On the contrary, if we observe a multifractal distribution, the clustering in time cannot be considered constant. We analyse the earthquake catalogues of some Italian regions and of Kaoiki Island (Hawaii), showing that in general we have multifractal distributions, and therefore a varying degree of clustering. This result suggests that the fractal dimension as well as the correlation dimension are the best parameters to be monitored for enhancing earthquake prediction.