A detailed earthquake catalog using Machine Learning-based methods for Tuscany, Italy

Seismic activity in Tuscany, Italy, is driven by the interplay between complex tectonics and local geological processes. Fluid-driven seismic sequences may occur in high-enthalpy geothermal regions, such as the Larderello-Travale Geothermal Field (LTGF), the oldest and among the most productive geothermal systems of the world. To better understand the regional tectonic setting, we build a detailed seismic catalog of earthquake hypocenters and magnitudes from a composite seismic network consisting of 30 temporary stations deployed in Tuscany in the framework of a temporary experiment (TEMPEST), during a period of one year (from September 2020 to September 2021) and 30 permanent seismic stations from the Istituto Nazionale di Geofisica e Vulcanologia (INGV).

We applied an automated processing routine including a machine learning (ML) phase picker, PhaseNet, and the Gaussian Mixture Model Association (GAMMA) algorithm, a sequential earthquake association and location workflow. We initially obtain nearly 1 million P-phases and 2 million S-phases, yielding in around 5 thousand detected events. We then located the events with NonLinLoc and applied a quality factor metrics to filter out potential false detections (22%) and recognize the high quality solutions which represents 30% of the initial 5 thousand locations with moment magnitudes (Mw) ranging between 0.5 to 2.9, and depths generally shallower than 15 km. Further steps involve the location analysis of the remaining events from the initial catalog. Moreover, we will apply relative earthquake location methods to better constrain already evident seismicity clusters. We also plan to calculate focal mechanisms from first-motion polarities and Moment Tensor (MT) inversion to investigate the earthquake sources in the highlighted tectonic features.

This work represents the starting point of the project “Multidisciplinary and InteGRated Approach for geoThermal Exploration” (MIGRATE). The goal of MIGRATE is to streamline passive seismic exploration methods for the investigation of geothermal resources, while addressing relevant scientific questions. This will result in the development of an automatized end-to-end tool to prospect the upper crust and identify potential geothermal targets.