Monitoring megathrust faults with abyssal distributed acoustic sensing

Earthquakes have caused more than half a million of fatalities in the past 20 years. A large fraction of this death toll arises from the current lack of systematic predictive signals. While some theories describe earthquakes as intrinsically stochastic processes, challenging or impossible to predict, evidence from laboratory and numerical experiments indicates that earthquakes could be preceded by a preparatory phase. But despite all efforts, only incomplete observations of such phase have been achieved. This observational gap is mainly due to our inability to deploy extensive sensor networks near the earthquake nucleation zone, especially for large earthquakes that are the most likely to produce precursors detectable at the surface.I propose to probe the mechanical state of a fault prior to large earthquakes using seismic waves recorded on the largest and densest seismic array ever deployed directly above one of the most active faults on Earth: the Chilean subduction zone. Long term monitoring of its vigorous activity will allow my team to observe the preparatory phase of several strong earthquakes (M>6).To achieve this, I will use a revolutionary technology, Distributed Acoustic Sensing, to convert several ~100 km long segments of fiber optic telecommunication cables that run offshore along the 4200 km of the Chilean subduction zone, into a large and dense ocean-bottom seismic observatory. The unique data produced by this new observatory will enable the detection of weak earthquakes and changes in the crustal properties with a sensitivity that has never been achieved before.This transformative capability, augmented by the development of real time data processing workflows, will enhance the early warning system in Chile by improving the timeliness and accuracy of earthquake warnings. The expected outcomes of this project will have a transformative impact on earthquake science as well as on the reduction of societal vulnerability to natural hazards.