Forecasting the unpredictable

Dr Andrea Antonioli, Professor John McCloskey, Dr Sandy Steacy and Dr Suleyman Nalbant - University of Ulster

Dr Spina Cianetti, Dr Massimo Cocco, Dr Carlo Giunchi and Dr Alessio Piatanesi -Istituto Nazionale di Geofisica

Following the devastating Boxing Day earthquake and tsunami of 2004, John McCloskey and his colleagues calculated the possible impact of that event on the timing of future large earthquakes in the region. Large earthquakes change the stress state on nearby geological faults and can therefore make future events either more or less likely. In the case of the Boxing Day earthquake, the stress changes increased the likelihood of events south along two known faults - the Sunda Trench megathrust fault and the Sumatran Fault both on the island of Sumatra.

The team published their results in Nature on 17 March 2005 and 11 days later the second biggest earthquake anywhere in the world since 1965 occurred on the Sunda Trench exactly in the area highlighted in the paper. 'We were not surprised that we got the size and location nearly spot on but the timing was quite unbelievable', says John. 'Although our methods can indicate the likely location and magnitude of an earthquake, it is impossible to say when it might happen. Tying down the timing of triggered shocks is the most urgent challenge now facing earthquake scientists'.

The March earthquake increased stress further south along the Sunda Trench and along a greater extent of the Sumatran Fault. This portion of the Sunda Trench was already considered to be at high risk. Detailed work by Kerry Sieh of the California Institute of Technology has shown that the last large earthquakes in that area occurred in 1797 and 1833, and a further great earthquake is therefore a real possibility. 'Based on our calculations of stress changes combined with the work of Kerry Sieh, we are very concerned about the earthquake potential under the Mentawai Islands, off the West coast of Sumatra', says John. 'Kerry considers a reasonable model for this future earthquake to be the one that occurred in 1833 which produced a large tsunami and caused many casualties on the island of Sumatra'.

John and his colleagues are now collaborating with Kerry as well as with Alessio Piatanesi, Carlo Guinchi, Spina Cianetti, and Massimo Cocco in Rome to model the range of possible tsunamis that might be generated by this earthquake. If the earthquake experiences high slip on the earthquake fault which leads to large uplift of the seafloor then a large tsunami might be expected, whereas if the seafloor movement is small so will be the resultant tsunami. The length of the earthquake is also significant with a longer earthquake more likely to generate a large tsunami.

The calculation of these likely tsunamis involves several steps. First, John and his team generate a series of maps of possible fault slip for a number of likely earthquake lengths. They send these to Kerry in California who selects the ones that are most geologically likely given the history of previous earthquakes on this fault. These are then used to compute the seafloor displacements, and finally Alessio in Rome computes the tsunami resulting from each particular earthquake. 'Much of the loss of life on Boxing Day 2004 was avoidable', says John. 'We hope that our results will help planners and civil defence officials prepare for the next tsunami along the Sumatran Coast'.