Wednesday, Jan. 13, 2010

Could the Haiti Earthquake Have Been Predicted?

The tragedy of the earthquake that struck Haiti on Tuesday is easy to measure in the lives lost, homes destroyed and infrastructure wrecked. The paradox of the quake is equally evident: when a natural disaster so devastating hits, oughtn't we have some way of predicting it? Hurricanes, blizzards, even volcanoes can be forecast well before their arrival, after all, allowing governments and people to make lifesaving preparations. Earthquakes, however, are stealth disasters, geological phenomena largely undetectable until just seconds before they occur. What scientists have long wanted to know is why quakes are so sneaky and what, if anything, can be done to read their warning signs better.

If any earthquake ought to have been predictable, it was the one that just struck. Haiti sits over two clashing tectonic plates, the North American and the Caribbean, which form what's known as the Enriquillo-Plantain Garden Fault. Geologists know the fault well and have studied it for decades, and well they should: it has shaken the region violently and repeatedly over history, though yesterday's quake, measuring 7.0 on the Richter scale, is the worst in a century.

"The plates are shearing the island, grinding it and crushing it," says geophysicist Michael Blanpied of the U.S. Geological Survey (USGS). "As that happens, earthquakes pop off."

Monitoring zones like this around the world to get a general sense of where the next such pops may happen is not that difficult, mostly because tectonic activity is hard to conceal completely. There are three types of quakes: the dip-slip fault (in which one clashing plate slides under the other), the reverse dip-slip (in which they pull apart) and the strike-slip (a sideways grinding of the plates). Haiti's quake was a strike-slip, but all three varieties are preceded by years or centuries of accumulated stress, in which the slowly drifting slabs of planet try to shift their position but remain hung up on one another. The quake occurs when that rock lock is broken. Seismometers can detect the years of stirrings that lead up to that moment, providing some help to geologists.

"We can say things generally about earthquakes," says Blanpied. "We look at faults and patterns of quakes over many years and say where on the landscape they're likeliest to occur next. This forecasting is usually in the long term." Indeed, scientists predicted as recently as 2008 that a fault zone on the south side of Haiti's island of Hispaniola posed a "major seismic hazard."

The problem is, long-term forecasting does not do much to keep people from putting themselves in harm's way. If it did, nobody would live in California or Mexico City or parts of Japan. What's needed is short-term forecasting on the order of weeks, days or hours. And this has stymied scientists again and again.

In the 1980s, for example, the USGS partnered with the state of California to study a particularly unstable stretch of the San Andreas fault near the town of Parkfield. Quakes had been occurring in the area every 20 to 25 years or so. The previous one had been in 1966, so scientists predicted another should hit before 1990. They threaded the region with seismometers and other sensors, then sat back and waited for the telltale signal that would give them notice of the coming rumble. The quake never occurred in the 1980s, nor in the 1990s. It was not until 2004 that the earth in the area at last moved again. And what did all the high-tech hardware reveal? Nothing.

"We did capture wonderful information about the earthquake afterwards," says Blanpied. "But the earth gave no indication with a foreshock or an electrical signal or a water signal or anything else. It demonstrates to us what we've already learned: earthquake prediction, if it can be done at all, is very difficult."

That's not to say geologists aren't making headway. Instruments that detect P waves are a good example. Earthquakes set off two kinds of seismic vibrations: body waves, which move through the interior of the earth, and surface waves, which move on top. The fastest of the body waves is the P wave, and it's thus the first to travel from the epicenter to a seismic station where it can be detected. P waves don't give you a whole lot of notice, but even a little bit can help. In California, gas lines in many homes are now equipped with P-wave detectors, which sense a coming tremor and shut off the inflow valve to prevent fires and explosions if the house is shaken severely.

Much to the disappointment of pet lovers, the celebrated ability of animals to predict quakes before they happen is not all it's said to be. It's true that animals are more sensitive to shaking than humans are, so they may feel a tremor a few seconds before we do. But you need the right animal in the right spot at just the right moment, and even then, that detection doesn't buy you much extra time. Nonetheless, Blanpied says, the research goes on.

For now, the best the USGS and other organizations can do is use the global web of seismometers and other instruments that are already in place and always being improved to build the best map possible of the planet's interior and at least narrow the predictive window of when quakes are likely to happen. They can also use these data to anticipate how severe the next tremor, whenever it comes, is likely to be. This can help drive policy decisions like improving building codes, reinforcing infrastructure and zoning some areas as unsuitable for development. That's hardly the same as precise predicting, but if your home or your life is the one that's spared, it's plenty good enough.