Semi-automated Template Matching and Machine-Learning Based Analysis of the August 2020 Castelsaraceno Microearthquake Sequence (southern Italy)

The accurate characterization of microearthquake sequences allows seismologists to better understand the physical processes involved in earthquake nucleation and rupture propagation and to gain insights on fault geometry at depth. Standard procedures for seismic sequences analysis are based on manual detection and phase-picking, requiring a huge amount of work from expert seismologists, particularly in the case of microseismic events. Here we show how the investigation of a low-magnitude foreshock-mainshock-aftershock sequence, occurred in August 2020 close to Castelsaraceno village (southern Italy), greatly benefited from the application of a semi-automated template matching and machine-learning based workflow. The phase-picking was automatically performed through a deep-learning algorithm on 202 microearthquakes detected between July and October 2020, followed by an automatic multi-step absolute and relative earthquake location procedure. The 72 relocated events of the seismic sequence were clustered in time (7-12 August) and in a narrow range of depths (10-12 km). The Ml 2.1 foreshock doublet and the Ml 2.9 mainshock identified a persistent asperity. The joint analysis of aftershocks distribution, the mainshock focal mechanism, and the geology of the study area suggest the occurrence of the sequence along a NNE-SSW left-lateral, transtensional fault in the brittle portion of the crystalline basement.