[ooyala id=”libDVvMzqUt-T9854XFlHLbJvgyR3QTj” ]
From today’s paper: A year after the earthquake in Japan, Tom Chivers reports that scientists still have no reliable method of prediction.
One year ago this Sunday, a huge earthquake – the fifth largest ever recorded – ripped through the seabed about 40 miles off the coast of Japan. At magnitude 9.0, the Great Tohoku earthquake shifted Honshu Island 8ft to the east, and made the Earth wobble on its axis. It sent a series of tsunamis, some 130 feet high, crashing into the coast, tearing a swath of destruction six miles inland and wrecking the nuclear power station at Fukushima. The total energy released was 600 million times that of the Hiroshima bomb. Fifteen thousand people are known to have died; thousands are still missing a year later. And nobody knew it was coming.
Even in Japan – a highly advanced society with access to the best scientists and engineers, located in one of the most seismologically active regions of the planet – there was not one minute’s warning before this vast rupture shattered the crust of the Earth. The country had been devastated by earthquakes before, most famously in 1923, when the Great Kanto quake killed at least 100,000 people. If any nation needed to be able to predict an earthquake, it was Japan. But it couldn’t.
People have been looking for ways to predict earthquakes for centuries, and all sorts of myths have grown up. In Sri Lanka, ahead of the 2004 Boxing Day tsunami (following an earthquake in the Indian Ocean), reports came of elephants fleeing the coast. Indeed, there have been stories of animals behaving strangely before earthquakes since classical times: Greek historians said that rats and snakes left the city of Helice days before a devastating quake in 373BC.
For geologists, finding any reliable method of prediction is a priority – a literal matter of life and death. Last year, six Italian scientists and a government official were charged with manslaughter over their failure to give sufficient warning of the 2009 L’Aquila earthquake, which killed 309 people; the trial is currently under way.
In the days before the quake, smaller tremors shook the region, which is sometimes an indicator of danger. But while half of all major quakes in the area are preceded by such “foreshocks”, the reverse is not true: only about one in 50 clusters of small quakes is followed by a big one. “What people don’t understand is just how low the risk was. These swarms of earthquakes happen all the time,” John Vidale, a University of Washington seismologist, told Scientific American last year.
So what can scientists do to predict a quake? According to Brian Baptie, a seismologist at the British Geological Survey, the answer is nothing. “Nobody has been able to predict an earthquake in terms of where, when and how big it might be,” he says. And without any one of those three facts, any prediction is essentially useless.
“When you’re talking about earthquakes that affect tens of thousands of people, the upheaval of evacuating the area is huge. You need a very reliable means of predicting them for it to be of use,” he says. In L’Aquila, much was made – after the fact – of a local technician called Gioacchino Giuliani, who had warned of a disaster. But his prediction of where and when the quake would happen, based on radon gas emissions, was out by a week and several miles.
There are several reasons why prediction is so difficult. Like weather forecasting, seismology is a study of a complex and chaotic system – in this case, the fluid dynamics of Earth’s viscous core and mantle and their interaction with the rigid tectonic plates above them. But unlike the weather, about which we have vast amounts of data from satellites and weather stations, large earthquakes are rare events, and what causes them is hidden from view.
What geologists can do is predict, in general terms, how often earthquakes of a given size will occur in particular parts of the world. “We can put together statistical models of earthquake activity – we know they happen at the boundaries between tectonic plates. But it’s quasi-random,” says Dr Baptie.
That seeming randomness is a product of both the complexity of an earthquake’s origins, and our own ignorance. A rupture, or fault, will start to run through the crust of the Earth, and, Dr Baptie adds, “the longer that fault propagates, the bigger the earthquake is. But we don’t understand what starts and what stops that rupture.”
To complicate matters further, not all earthquakes happen on tectonic plate boundaries, where slabs of crust jostle and slide against each other. There are also rare but catastrophic “intraplate” quakes, such as the one that killed more than 12,000 people in Gujarat, India, in 2001. These remain a geological mystery. While a tectonically active country such as Japan is as well prepared as it can be for a big quake, intraplate events come out of the blue.
Despite efforts to study what happens underground, at places such as Parkfield, on the San Andreas fault in California, the process is slow. “There have been lots of subsurface measurements of things like strain rates,” says Dr Baptie. “None of them have been particularly successful in predicting when the next event might occur. There’s no probe that you can send down a borehole that will tell you this is what the stress [building up in the rocks] is.”
That’s not to say there aren’t lots of suggestions for indirect ways of seeing what’s going on beneath the surface. One is Giuliani’s radon method; according to this hypothesis, the radioactive gas, trapped in rock cavities in the Earth’s crust, is released when that crust is under high levels of stress. When radon levels in soil or groundwater are elevated, an earthquake may be imminent. But this hypothesis is a far from reliable indicator, and there are huge technical challenges involved in monitoring radon levels across the world’s faultlines.
If it falls by the wayside, it won’t be the first faddish prediction method to have gained attention. “In the 1980s something called the ‘VAN method’ became popular,” says Dr Baptie. “It looked at changes in electromagnetic fields in different places. But there was no real physical mechanism: it was just a series of measurements which may or may not have shown something. In the end, it was shown to be unreliable.” Other methods under examination include “seismic anisotropy”, which involves studying variations in seismic waves. This has been successful in experiments, but has not been demonstrated outside the laboratory.
And there is little evidence to support the folklore that animals do behave erratically before an earthquake. “Animals behave weirdly all over the world, and at strange times,” says Albert Yue Yuan, a Chinese science writer who recently addressed the American Association for the Advancement of Science on the science, and pseudoscience, of earthquake prediction. “People make the wrong connection afterwards.”
This phenomenon is called “psychological focusing”: people remember their dogs behaving strangely if, shortly afterwards, there’s an earthquake, but forget those times when dogs get excited and nothing happens. It would be surprising, says Yuan, if animals could predict earthquakes, since “there’s no advantage to evolve that ability, no pressure of natural selection”. It’s simply a myth.
Earthquakes are probably not becoming more frequent, but they are killing more people. The highest levels of human population growth are, coincidentally, in some of the most tectonically active parts of the planet. Before the next big one strikes, even a few minutes’ notice could save tens of thousands of lives by allowing people to dive under desks – and bring glory to the scientist who finds a way to tell when the Earth is about to go into spasm. But so far, that goal seems as remote as ever.