Heterogeneous Coupling on the Sumatran Megathrust and the Nature of Triggering of Seismicity in the Inter-seismic Period

The recent sequence of great earthquakes on the Sunda megathrust off western Sumatra has provided seismologists with a large quantity of modern quality data, both seismic and geodetic. Sumatra has, as a result, become a focus of study for seismologists, as it is where the next insights are likely to be made in our understanding of the constraints on earthquake triggering, the details of the distribution of slip in large events and the long term evolution of slip in these tectonic settings. Observations of anomalous interseismic geodetic deformation (Chlieh et al 2008) on Sumatra and its islands have been interpreted to be the result of a spatially heterogeneous distribution of interseismic slip rate at the interface. Spatially, this is parameterised by a 'coupling' field so that maps of the distribution of coupling are indicators, to first order, of the amount of interseismic strain that has been accumulated at each location on the fault surface. This field alone, without making any assumptions about its specific physical causes, may therefore be useful to predict the characteristics of the seismicity we expect in terms of where events trigger, the extent of earthquake rupture and the distribution of slip. A study of the distribution with respect to the coupling field of hypocentre locations (as published in the Bulletin of the International Seismological Centre) of Sumatran inter-seismic seismicity is presented, where the statistical parameterisation of the earthquake catalogue with respect to the coupling is based not only on the published locations but also on the uncertainties in the solutions and the resolution of the coupling field. Any location on the fault surface, which has an associated value for coupling, may be used as a model hypocentre for each observation. The likelihood for the model location to have produced the data is used to weight the contribution of its value of coupling to a parameter η characterising the asymmetry of the catalogue with respect to the coupling field or, equivalently, with respect to the uniform distribution. An estimate of the statistical significance of the result uses the null hypothesis of randomly distributed seismicity. We discuss possible systematic location errors in the region and what their likely effect on the results would be. We interpret the results in the context of interseismic slip rates at the resolution that is available at present from geodetic observations. However, the inherent limits to the spatial resolution of geodetic techniques for processes occurring at depth in the crust mean that the observed coupling distribution will be filtered in high frequencies. We suggest some further work to examine how these unresolved heterogeneities would influence the character of the seismicity we expect, and whether it could be used to explain deviations in the real data from the predictions of a more homogeneous model.