How do large earthquakes start? Illuminating the nucleation phase from seismology and geodesy

How do large earthquakes start? The answer to this simple question might enable improved early warning and earthquake forecasting. In one endmember model, small tremors randomly redistribute stresses that can trigger more quakes, and some of these grow unpredictably into large events. In this 'cascade model', the processes during the pre-earthquake phase are stochastic, and the eventual size of a quake is determined by the random barriers to continued rupture. The cascade model underlies current operational earthquake forecasting systems around the world and provides useful but limited predictive skill. In another endmember model, aseismic creep sweeps across a fault, generating seismic events where the aseismic slip encounters asperities, and leading to large earthquakes when the asperities are big or connected. In this 'pre-slip model', foreshocks may obey discernible patterns that betray the aseismic slip, such as a spatial migration of foreshocks or repeating tremors that re-break asperities continually reloaded by slow slip. Such patterns might help to forecast impending quakes better. Recent geodetic observations prior to some giant earthquakes in subduction zones, such as the 2011 Tohoku (Japan) and the 2014 Iquique (Chile) earthquake, seem to support the pre-slip model, but some onshore events occurred without discernible pre-slip. How large earthquakes start is of immense scientific and practical interest. The goal of this project is to gain new insights into how large earthquakes start and to determine the extent to which quakes in different tectonic settings start by cascading foreshocks or aseismic slip. The project will primarily draw on methods from observational earthquake seismology and statistical seismology to make inferences in the context of constraints from space geodesy (GPS/GNSS), fault mechanics and earthquake geology.