Identification and characterisation of earthquake clusters: a comparative analysis for selected sequences in Italy and adjacent regions

Identification of earthquake clusters has a twofold scope: a) characterisation of clustering features and their possible relation to physical properties of the crust; b) declustering of the earthquake catalogues, to allow for space-time analysis of main-shocks occurrence. Since different methods, relying on different physical/statistical assumptions, may lead to diverse classifications of earthquakes into main events and related events, we investigate the classification differences for three different declustering techniques: the nearest-neighbour approach (NN) and the two widely used windows methods by Gardner-Knopoff and Uhrhammer. A formal selection and comparative analysis of earthquake clusters is carried out for selected earthquakes in north-eastern Italy and adjacent regions, as reported in the OGS catalogue since 1977. The comparison is, then, extended to earthquake sequences associated with strong earthquakes in central Italy, occurring in a different seismotectonic setting, by making use of INGV data over the period 1981-2017. The NN data-driven approach turns out well consistent with classical window approach for large events, while improving clusters identification in areas characterised by low to moderate seismic activity, where windowing methods necessitate adequate optimisation. Moreover, the declustering performed by NN method preserves the features of inhomogeneous and possibly non-stationary background seismicity, relevant for several studies.