Abstract

Constraints on the potential size and recurrence time of strong subduction-zone earthquakes come from the degree of locking between the down-going and overriding plates, in the period between large earthquakes. In many cases, this interseismic locking degree correlates with slip during large earthquakes (1-4) or is attributed to variations in fluid content at the plate interface5. Here we use geodetic and seismological data to explore the links between pore-fluid pressure and locking patterns at the subduction interface ruptured during the magnitude 8.8 Chile earthquake in 2010. High-resolution three-dimensional seismic tomography reveals variations in the ratio of seismic P- to S-wave velocities (Vp/Vs) along the length of the subduction-zone interface. High Vp/Vs domains, interpreted as zones of elevated pore-fluid pressure, correlate spatially with parts of the plate interface that are poorly locked and slip aseismically. In contrast, low Vp/Vs domains, interpreted as zones of lower pore-fluid pressure, correlate with locked parts of the plate interface, where unstable slip and earthquakes occur. Variations in pore-fluid pressure are caused by the subduction and dehydration of a hydrothermally altered oceanic fracture zone. We conclude that variations in pore-fluid pressure at the plate interface control the degree of interseismic locking and therefore the slip distribution of large earthquake ruptures.