Evaluation of Historical Earthquake Interaction and Seismic Risk to Western Sumatra

On 26 December 2004 the great Sumatra Andaman Earthquake ruptured the Sunda megathrust for a length of more than 1200km resulting in a magnitude 9.3 earthquake and a massive tsunami which devastated the entire circum-Indian ocean coastline. Over 250000 people died. Earthquakes do not occur in isolation but one earthquake deforms the earth's crust around it and transfers stress onto other structures which may also generate other, so called triggered earthquakes. Over the last ten years or so, scientists have tried to understand this effect and have developed mathematical tools for the calculation of these interaction stresses in the crust. This work has lead to the idea that if the interaction stresses could be calculated immediately after an earthquake, the structures which would be likely to produce other events might be identified, thus locating areas of particularly high seismic hazard. Following the Sumatra-Andaman earthquake, the University of Ulster geophysics research group illustrated the usefulness of this method by pointing out the increase risk of another great (ie M>8) earthquake under the Islands of Simeulue and Nias to the west of Sumatra. The forecast event occurred (M8.7) 11 days after their paper was published in Nature While this work clearly illustrated the power of the technique there is, unfortunately not one-to-one mapping between the calculated stresses and the following earthquakes. So, while we can say where another earthquake is likely we cannot say when, or even how likely. It is of great importance that more work is done if we are to be able confidently and consistently to use the interaction stress technique in a practical way. Another crucial and closely linked question is: what earthquake are we trying to forecast? While some scientists believe that big earthquakes recur with more or less the same magnitude in more or less the same place, others point out significant failures in this idea. Big earthquakes sometimes overlap, occur in different places, combine into even bigger earthquakes or break more than one fault. In this project, the team, which brings together scientists from the universities of Ulster and Edinburgh as well as from the National Institute for Geophysics and Vulcanology in Rome, will examine how good the interaction stress idea is, not by looking into the future, but by looking into the past. Due to the concerted effort of scientists for over 20 years, there are few places on earth for which we have a better picture of past big (M>8) earthquake activity. We are confident that we know the locations and approximate sizes of all earthquakes (M>6) which have occurred there since about 1600. We also know a lot about plate convergence - the driving force for earthquake generation. In this project we will put this information together in a complex numerical simulation to reconstruct the entire stress field in the western Sumatran region for the last 400 years. The resulting 3D stress history will allow us to see if the interaction stress idea could have been used to explain the last 400 years of earthquakes; this will help us to understand what is likely to happen next. In addition we will look at the recent records of smaller earthquakes and see if we can use these to tell us about the dependability of the simple recurrence behaviour for the big ones. This problem is complex and requires very careful application of statistical techniques to be sure we reach the right conclusions. We will develop many of these during the project. Western Sumatra is currently one of highest earthquake risk areas in the world. Coastal populations around the Indian ocean, and on the western coast of Sumatra in particular live with the threat of another great earthquake which has the potential to kill on the same scale as the boxing day event. This project will not stop the earthquake but it will contribute to a body of scientific knowledge which will ultimately help to prepare people for it.