A team of solar scientists says it has improved on approaches that predict the eruption of solar flares, violent bursts of energy that can damage satellites, endanger astronauts in orbit and even threaten the power grid on the ground. Space agencies, airlines, satellite operators and power utilities would like to have access to better forecasts of all kinds of space weather—the charged particles and streams of radiation spewed out in irregular burps and blasts by the sun.

To that end, National Oceanic and Atmospheric Administration (NOAA) research scientist Alysha Reinard and her colleagues made use of data from the Global Oscillation Network Group (GONG), a set of six telescopes around the globe that together keeps a continuous watch on the sun. GONG takes helioseismology measurements, tracking oscillations on the sun's surface that point to its convective inner workings.

GONG data since 2001 featured more than 1,000 magnetically active regions on the sun's surface, where solar flares originate. The swirling plasma beneath those active regions appeared to wind up to a critical point—the twisting flows, which presumably drag magnetic field lines along with them, would ramp up in a telltale way before a flare.

"We were seeing this twisting that seems to occur beneath the surface of the sun," Reinard says. "And this twisting seems to start really fast and then slow down to almost nothing right as this flare occurs." In a paper set to be published in The Astrophysical Journal Letters, Reinard and her colleagues used the analogy of a twisted rubber band—it becomes harder and harder to twist until it finally snaps. By tracking the twisting of magnetic fields in the sun's interior through helioseismology as well as the strength of the magnetic field at the surface, space weather forecasters should be able to predict solar flares two or three days in advance with unprecedented accuracy, the researchers claim. By one metric, their method roughly doubles the predictive skill of the forecast approaches surveyed in a December 2008 study.

Gordon Holman, an astrophysicist at the NASA Goddard Space Flight Center in Greenbelt, Md., who did not contribute to the new research, agrees that this appears to be the most reliable method of predicting flares at present. He says that peering inside the sun with helioseismology is a logical approach, as flares are ultimately thought to release built-up energy from magnetic fields twisted or sheared by flows in the solar interior. "Since the twisting occurs below the solar surface, this is where we really need to look to be able to confidently predict flares," Holman says.

But both Reinard and Holman say that the method could be developed further. Holman notes that only about half, and sometimes fewer, of the solar flares in the data set were correctly predicted by the new method, and the false alarm rate was greater than 50 percent. "So, it is still far from ideal," he says. "Nevertheless, this is excellent work and generally, I think, the best approach to flare prediction."

Reinard agrees that her approach is "still missing some events, and we still have some false positives. We definitely would like to improve what we have." She and her colleagues hope to have honed a more viable method by the time the next solar maximum rolls around, in three years or so. (The sun's activity waxes and wanes in an 11-year cycle; the next maximum in that cycle is projected for 2013.) And NASA's Solar Dynamics Observatory, set to launch in February, should provide solar scientists with higher-resolution data for forecasting.

Some concerned parties—notably airlines and the military—want almost perfect forecast accuracy, Reinard adds. "I don't know when that's going to happen," she says. Holman's benchmark is more realistic: "Being able to predict flares at the level of confidence that local weather several days from now is predicted would be a significant achievement," he says.