Earthquake supercycle in subduction zones controlled by the width of the seismogenic zone

A supercycle describes a long-term cluster of differently-sized megathrust earthquakes, leading up to the final complete failure of a subduction zone segment(1,2). The precise controls on supercycles are unclear, although structural and frictional heterogeneities are proposed(1). We recognize that supercycles are suggested to occur in those regions(1-4) where the estimated downdip width of the seismogenic zone(5-7) is larger than average. Here we investigate the link between supercycles and the seismogenic zone downdip width using a two-dimensional numerical model(8). In our simulations, the first megathrust earthquakes in a supercycle generally rupture only the outermost parts of the seismogenic zone. These partial ruptures are stopped owing to a large excess of strength over stress, and transfer stresses towards the centre of the seismogenic zone. In addition to the continued tectonic loading, they thereby gradually reduce the strength excess so that the largest megathrust events finally rupture the entire seismogenic zone and release most of the accumulated stress. A greater width increases the average strength excess and thus favours supercycles over ordinary cycles of only similarly sized complete ruptures. Our results imply that larger than thus far observed earthquakes could conclude a supercycle where seismogenic zone widths are larger than average.