MONIFAULTS · Monitoring real faults towards their critical state

The last seismic sequence in Italy, responsible for 298 fatalities and important economic loss, remind us how urgent it is to improve our knowledge about earthquake physics to advance earthquake forecasting. While direct observations during laboratory earthquakes permit us to derive exhaustive physical models describing the behaviour of rocks and to forecast incoming lab-earthquakes, the complex physics governing the nucleation of earthquakes remain poorly understood in real Earth, and so does our ability to forecast earthquakes. I posit that this ‘ignorance’ emerges from our limited ability to unravel information about fault physics from geophysical data.The objective of this proposal is to introduce a new and integrated methodology to monitor the spatiotemporal evolution of elastic properties on real faults using seismological and geodetic data. We will apply machine learning and covariance matrix factorization for improved earthquake detection, and to discover ‘anomalous’ seismological signals, which will reveal unknown physical processes on faults. These novel observations will be integrated with time dependent measurements of rheology and deformation, obtained from cutting-edge techniques applied to continuous seismological and geodetic data. Our integrated monitoring approach will be applied to study how faults respond to known stress perturbations (as Earth tides). In parallel, we will analyse periods preceding significant earthquakes to assess how elastic properties and deformation evolve while a fault is approaching a critical (near rupture) state. Our natural laboratory will be Italy, given its excellent geodetic and seismological instrumentation, deep knowledge about faults geometry and the relevant risk posed by earthquakes. Our research will provide new insights about the complex physics of faults at critical state, necessary to understand how real earthquakes nucleate. This project will also have a major impact on observational earthquake forecast.