Can radon gas leaks predict earthquakes?

The earthquake that devastated L 'Aquila, Italy, on Monday casts a spotlight on the challenges scientists face as they try to improve earthquake forecasts.

Specifically, it raises the question of whether the release of radon gases from the ground can accurately predict the arrival of a temblor.

The short answer? Maybe.

Giampaolo Giuliani, a researcher at Italy's Gran Sasso laboratory, alerted authorities in the region of Abruzzo that a quake was imminent – and was condemned for raising a false alarm. Mr. Giuliani declined requests to be interviewed, and according to his wife, who was also reached by phone Tuesday, is busy working at his lab.

While the details about the timing and location of Giuliani's warning remain fuzzy, he reportedly based his warning on radon emissions that instruments at his lab were picking up.

Abruzzo is not an easy place to test such an approach, according to US scientists. It's a mountainous, quake-prone patch of central Italy that covers nearly 4,200 square miles.

It's about as complex a place geologically as one can imagine, notes Felix Waldhauser, a researcher at the Lamont-Doherty Earth Observatory in Palisades, N.Y.

It's under pressure from one of the world's large crustal plates, the African plate, which is rotating. It's also influenced by smaller "microplates" in the region. As a result, Abruzzo hosts several broad fault types. But while they fall into a few broad classifications, each fault carries its own unique traits.

Over the years, scientists have explored a number of techniques for trying to determine with some precision when and where a quake will strike. Radon, a naturally occurring radioactive gas, has been listed as one of several possible indicators of an impending temblor.

Indeed, Giuliani said during a March 23 video interview (click here to view it in Italian) posted on a local Abruzzo website, "Our instrument ... allows us to see continuously the seismic precursors within this element [radon], which manifests itself between six and 24 hours ahead of a quake."

Yet even he reportedly had the quake mistimed by at least a week.

That's one reason why the American scientist who first proposed studying radon emissions – and other potential signals – remains modest about their applications.

For all the study that potential precursor signals (such as radon and the electrical properties of fluids moving through faults) still receive, no single indicator or combination of indicators has panned out as a reliable quake predictor, according to Christopher Scholz, a geophysicist at the Lamont-Doherty observatory. He and two colleagues first set out the proposition that such phenomena might serve as quake warning signals 36 years ago, based on lab experiments of how rocks and liquids behave as stress builds along a fault.

When asked about the value of these potential predictors today, he says that the theory behind them still hasn't been proved or disproved. "There are so few measurements made that there's not a good database to say, 'Oh, yeah, this works,' or 'it doesn't,' " says Dr. Scholz.

One instance of a claimed success, as in the Abruzzo case, doesn't represent a reliable, repeatable result. To confirm that it works (among other problems), "you have to know by some other means when and where an earthquake will strike," says Scholz, in order to set out the instruments that can take the needed measurements.

In his March 23 interview, Giuliani said that more detectors were needed to create a network that "would allow us to monitor carefully radon activity."

Radon appears to have successfully presaged a quake in some instances, but not in others.

Even in China, where seismologists successfully used groundwater changes, animal behavior, and swarms of foreshocks to predict a quake in 1975, they were unable to foretell another devastating quake a year later, Dr. Waldhauser says.

And for all that seismologists have learned about earthquake processes over the past several decades, faults often still confound them.

Scholz cites all the attention paid to the northern end of the section of the San Andreas fault near Parkfield, Calif. Scientists seeded the area with sensors and drilled a bore hole deep into the fault to monitor conditions there. The earthquake originally predicted for 1998 came in 2004 – at the southern end of the segment, far from where the instruments had been placed.

"That's the way it goes in this business," he says.

Scholz says he holds out hope that his theory will get a more rigorous test as major earthquake zones become more heavily instrumented.

Still, "predicting earthquakes is the Holy Grail of seismology," adds Waldhauser. He notes that the Greeks are working extensively with measurements of fluid electrical properties as warning indicators.

Researchers in India, Japan, and elsewhere are continuing to probe the radon approach.

Some scientists are even looking at satellite measurements of small disturbances in the ionosphere – found through minute changes in the properties of radio waves at very low and extremely low frequencies – for potential warning signs.

But these days, seismologists – including those who continue to test the approaches Scholz and his colleagues first proposed – are doing well to provide forecasts, rather than predictions.

The forecasts, based on combing the historical records as well as digging trenches across faults themselves to build a history of temblors on a fault segment, sound more like an area weather forecasts. And they read like this, from the latest assessment by the US Geological Survey:

The USGS and other scientists conclude that there is a 62 percent probability of at least one quake of magnitude-6.7 or greater, capable of causing widespread damage, striking the San Francisco Bay region before 2032.