Artificial Intelligence for Induced Earthquake Prediction.

The ability to predict the evolution of earthquakes, particularly those induced in geothermal areas, represents a crucial challenge in the field of geophysics. The study conducted by the MODAL research group coordinated by Francesco Piccialli, professor of Computer Science at the Department of Mathematics and Applications "Renato Caccioppoli," in collaboration with the National Institute of Geophysics and Volcanology (INGV) and the University of Salerno, led to the design and development of PreD-NET (Precursor Detection Network). This deep neural network, through training on large seismic catalogs, demonstrates a high prediction capability with 98 percent accuracy, marking a significant step forward in earthquake risk reduction.

The objective of the study was to develop a reliable and robust Artificial Intelligence model for monitoring and early warning of seismic events in geothermal areas, with a focus on minimizing false positives and negatives.

Through the use of advanced Deep Learning methodologies and multi-parameter analysis, PreD-Net can identify "precursors" of potentially dangerous earthquakes, providing crucial support for planning risk mitigation interventions.

This research, the first tangible output of the Project of National Significance (PRIN) called D.I.R.E.C.T.I.O.N.S., combines geophysical, seismological and computer skills, and opens new avenues in understanding induced earthquakes by offering concrete tools for their management, with significant implications for the safety of geothermal areas and the energy industry.

The study, published in the journal Scientific Reports in the Nature series lays the foundation for numerous future applications, including monitoring of natural seismicity, and is an excellent model of interdisciplinary collaboration in seismological research.