A statistical approach to foreshocks discrimination in Italy

Large earthquakes often occur unexpectedly, mostly in previously recognized high seismic hazard regions. Some of them are forewarned by small and moderate magnitude events classified as “foreshocks” once the major one happens. However, most seismic clusters are swarms, i.e., they do not lead to large mainshocks. The real time discrimination of foreshocks from swarms would be an outstanding achievement in short-term seismic forecasting. Here, we analyze clusters of swarms and foreshocks until the occurrence of the event with the largest magnitude in Italy from 1990 to 2024. Our results suggest that seismicity spreads over larger and larger areas, is featured by higher and higher magnitude variance and entropy, and it tends to grow both in number and cumulative seismic moment as the magnitude of the impending mainshock increases. We propose a possible explanation considering progressive destabilization of crustal volumes marked by some characteristics of seismic clusters. Our investigation provides a first attempt to assess the probability that an ongoing seismic cluster is likely to culminate with a large mainshock or to wane in a swarm, grounded on the statistical properties of observed seismicity, with all the limitations it entails.