Detecting and Characterizing Swarm-Like Seismicity

Swarm-like seismicity manifests as earthquake clusters driven by aseismic transients. The investigation of the mechanisms behind their occurrence is generally based on automatic detection and characterization of swarm-like clusters. In this study, we investigate four different (de-)clustering algorithms to identify earthquake clusters, and then classify these clusters as either swarm-like or mainshock–aftershocks sequences. The classification uses the clusters’ distribution of seismic moment over time, quantified by standardized central moments. Synthetic catalogs from an epidemic-type aftershock sequence model are used to establish confidence bounds for swarm classification. The workflow is applied to the swarm-dominated regions Húsavík–Flatey fault, Iceland, and Pollino range, Italy. Our workflow effectively detects/classifies earthquake clusters, but their inherent variability in duration and seismic moment release can bias automated swarm/mainshock–aftershocks labeling. The results provide benchmarks for future swarm-like seismicity analyses, highlighting the importance of a posteriori careful inspection of clusters to understand the underlying physical mechanisms.