Multidisciplinary analysis of the Tohoku-Oki Earthquake and implications for the estimation of strain accrual

Japan has the densest countrywide GPS network anywhere in the world and the techniques for estimating the location and magnitude of slip on a fault from geodetic data such as GPS are now well-established. However, the first published fault models from GPS data alone or GPS and seafloor pressure gauge data locate most of the slip around the hypocentral area, whereas results based on broadband seismic data locate most of the slip at shallower depths, closer to the trench. Interseismic analyses of GPS data acquired prior to the earthquake indicate that the coupled zone, where strain was accumulating, approximately coincides with the slip area from the geodetic studies. If the geodetically-estimated slip is correct, we have a chance of predicting the size of future earthquakes from this kind of interseismic study, but if the results from the seismic data are correct, we apparently do not. We take a multi-disciplinary approach to estimating the slip that occurred during the earthquake. We use measurements of deformation from GPS and seafloor geodesy together with measurements of tsunami propagation from seafloor pressure gauges to constrain the slip, and validate our results with measurements of tsunami run-up and satellite altimetry. Contrary to previous geodetic studies, our results indicate that the vast majority of slip occurred close to the trench and not where strain was estimated to be building. Given our incomplete record of where large earthquakes have occurred in the past, modelling of where strain is currently accruing is essential for assessing the earthquake potential of subduction zones. It is now clear, however, that the measurements used to constrain these models, which come from displacements on land alone, are not sufficient.