In conventional models of frictional sliding, faults slip in one of two ways. Cold, shallow faults slip episodically in earthquakes, and warm, deep faults slip slowly and steadily. But in the past few decades, observations have revealed that many faults slip show an intermediate behaviour: slow slip events. In slow slip events, initially accelerate but then puzzlingly stall well before they start generating large seismic waves.

Currently we are studying some large (300-km-long) slow slip events which rupture the subduction zone plate interface in Cascadia. In these events, the fault mostly slips slowly, at rates around 1 cm/day, but the slip triggers millions of tiny earthquakes. I am currently identifying those earthquakes in order to determine which part of the fault is slipping and how the slipping location evolves through time.

We will compare the spatial growth of a number of slow slip events with models of fault slip in order to determine which fault zone process drives slow slip. For instance, in one model, changes in fluid pressure in the fault zone pull the fault tightly shut when it tries to slip quickly. Such a model predicts a strong resistance to rapid slip and would predict a limited range of slip and growth rates, while other models, with less resistance to rapid slip, would allow a wider range of slip rates.