A comparative analysis of methods for identification and characterization of earthquake clusters

Identification and statistical characterization of seismic clusters may provide useful insights about the features of seismic energy release and their relation to physical properties of the crust within a given region. Moreover, in seismology some studies on hazard assessment require to remove secondary events (aftershocks) from the catalogue, leaving only main shocks; that is called declustering. Seismicity shows clustering in space, highlighted by the concentration of earthquakes along regional fault networks, and clustering in time primarily, but not only, associated with the increase of seismic activity immediately after large earthquakes leading to aftershock sequences. An attempt of combining both these aspects is done through the notion of generalized distance defined on the space-time-size domain, which measures the correlation between earthquakes in terms of Gutenberg-Richter law for the magnitude and of fractal dimension of the epicentres distribution. We aim at comparing declustering techniques, ranging from classical space-time windows methods to methods based on nearest-neighbour distance. A comparative analysis of resulting clusters is carried out for the most relevant earthquakes in North-Eastern Italy, as reported in the bulletins compiled at the National Institute of Oceanography and Experimental Geophysics (INOGS) since 1977. The examination is then extended to selected earthquake sequences associated with a different seismotectonic setting, namely to events that occurred in the region struck by the recent Central Italy destructive earthquakes, making use of data recorded by the National Institute of Geophysics and Volcanology (INGV). The study shows that the data-driven approach, based on the nearest-neighbour distances, can be satisfactorily applied to decompose the seismic catalog into background seismicity and individual sequences of earthquake clusters, also in areas characterized by moderate seismic activity, where the standard declustering techniques may turn out rather gross approximations.