鈥淵OU can鈥檛 rely on media reports,鈥 says Vasily Titov. 鈥淵ou have to ask eyewitnesses: what was the height of the first wave, and did the water recede first? You have to go into the field and assess damage to buildings, look at water marks, use any information you can about the size and speed of the waves.鈥
Titov, a researcher at the Pacific Marine Environmental Laboratory in Seattle, Washington, is one of the small community of tsunami experts who, three weeks on, are just beginning to understand the waves that wreaked havoc around the Indian Ocean. It used to take years to model a tsunami. Now it can be done in almost real time, but not without crucial data missing from the Indian Ocean on 26 December. So Titov and his ilk are having to reconstruct what happened, and if past tsunamis are anything to go by, the smallest details, culled from sources ranging from eyewitness accounts to satellite images, will be crucial.
For a start, seismologists are still struggling to understand the earthquake that triggered the waves (see 鈥淎natomy of a Quake鈥). But so far the best evidence suggests that a swathe of the sea floor lifted by up to 5 metres, while other parts dropped by 2.5 metres. The body of water above, 5 kilometres high, instantly moved with it, and from that moment a tsunami was inevitable.
Advertisement
Titov is the first scientist to run a computer simulation of the movement of this water. The speed of a tsunami in the ocean is directly proportional to the depth of the water, so at the deepest points of the Indian Ocean the water raced forward at up to 900 kilometres per hour. At this stage the wave was broad but shallow, 100 kilometres from front to back but less than 0.5 metres high. When the displaced water reached the shallower continental shelf it was forced to slow, allowing the back to catch up with the front to create the destructive tsunami waves.
鈥淭he wavelength would have gone from 100 kilometres to five kilometres. It is such a long wave it is difficult to grasp the scale,鈥 says Titov. Sketchy field data suggests wave heights of up to 10 metres, although tsunami experts are expecting evidence for waves of 30 metres. As eyewitness accounts and other field data are assembled, they will be fed into computer models to refine the simulations.
Experience suggests that this process of refinement is likely to continue for years. Simulations suggested that a 17-metre high tsunami that struck Papua New Guinea in 1998 was caused by an underwater landslide, triggered by a relatively small 7.1 magnitude earthquake. But in 2003, that was called into question. If the landslide simulations were correct the tsunami would have struck at dusk. Yet eyewitness interviews revealed that the wave had arrived slightly earlier, but while it was still daylight. 鈥淟ate afternoon activities were continuing, touch football, people painting a new canoe,鈥 says Hugh Davies of the University of Papua New Guinea in Port Moreseby.
To simulate a tsunami, scientists begin by feeding computer models with the location and shape of the initial earthquake or underwater landslide, and details of the ocean floor topography. The simulation then predicts how fast the tsunami will travel and in which direction. 鈥淢ountain ranges or depressions on the ocean floor will refract the waves, change their direction, alter their speed,鈥 says Elena Suleimani, a tsunami modeller at the University of Alaska in Fairbanks.
鈥淎n experimental computer program successfully recreated the tsunami鈥檚 passage on 26 December through open seas鈥
Titov鈥檚 experimental computer program, called MOST for 鈥淢ethod of Splitting Tsunami鈥, successfully recreated the tsunami鈥檚 passage on 26 December through open seas. The simulation lagged behind the actual event by a few hours, but the picture it produced proved remarkably accurate when cross-checked against the shape of the ocean鈥檚 surface, as determined by researchers at the US National Oceanic and Atmospheric Administration and NASA, using altimetry measurements from four satellites (see Graphic).
Under the circumstances, the results are amazing, Titov says. That鈥檚 because recordings from deep-sea pressure detectors called 鈥渢sunameters鈥 or DART buoys are usually needed for such a rapid simulation. The results demonstrate the robustness of the MOST program, but for real time simulations, tsunameters, which are deployed as part of warning systems in the Pacific Ocean, would be essential, says Titov. But there are none in the Indian Ocean.
As well as modelling the wave in open water, MOST has a second, 鈥渋nundation鈥 stage, designed to predict the size of waves that will hit the coast and how far inland they will run. Titov was unable to run this stage of the simulation because it needs tsunameter recordings plus high-resolution representations of details of the coastal topography, which are also largely unavailable for the Indian Ocean.
