Abstract
The identification of seismic clusters is essential for many applications of statistical analysis and seismicity forecasting: an incorrect cluster identification leads to incorrect results. However, there are several methods to identify clusters, and their results are not always compatible. We tested different approaches to analyze the clustering: a traditional window-based approach, a complex network-based technique (Nearest Neighbor - NN), and a novel approach based on fractal analysis. The case study is the increase in seismicity observed in Molise, southern Italy, from April to November 2018. To analyze the seismicity in detail with the above-mentioned methods, an improved template-matching catalog was created. A stochastic declustering method based on the Epidemic Type Aftershock Sequence (ETAS) model was also applied to add probabilistic information. We explored how the significant discrepancies in these methods’ results affect the result of NESTORE (NExt STrOng Related Earthquake) algorithm – a method to forecast strong aftershocks during an ongoing cluster – previously successfully applied to the whole Italian territory. We performed a further analysis of the spatio-temporal pattern of seismicity in Molise, using the principal component analysis (PCA), the ETAS algorithm, as well as analyses, aimed at detecting possible migration and diffusion signals. We found a relative quiescence of several months between the main events of April and August, the tendency of the events to propagate upwards, and a migration of the seismicity consistent with a fluid-driven mechanism. We hypothesize that these features indicate the presence of fluids, which are also responsible for the long duration of the sequence and the discrepancies in cluster identification methods. Such results add to the other pieces of evidence of the importance of the fluid presence in controlling the seismicity in the Apennines. Moreover, this study highlights the importance of refined methods to identify clusters and encourages further detailed analyses of anomalous sequences.