The sun has long been known to have "weather" -- regular cycles of activity and inactivity. Every day, the sun exhibits hundreds of sunspots, relatively cool areas that appear as dark blotches. In addition, there are numerous flares and other fluctuations. Predicting the details of such activity is a difficult task. But it may be about to get easier.

NASA scientists David Hathaway, Robert Wilson and Ed Reichmann recently looked at more than a dozen techniques for predicting sunspots. They tested these methods to see how accurate they are, and then combined the two best techniques into what they believe will be an ever better model for predicting solar activity.

Such predictions are useful because solar activity affects Earth in a variety of ways, for instance disrupting radio communications and altering local weather.

The scientists, who work at NASA's Marshall Space Flight Center, found that the best predictions result from measuring the sun's impact on Earth's magnetic field. "Explosions from the sun travel through space and hit the Earth, causing the magnetic field to wobble and shake," says Hathaway.

By determining the impact of previous solar cycles, it is possible to estimate the intensity of the next solar cycle. Hathaway's team does so by combining techniques developed, respectively, by Joan Feynman of NASA's Jet Propulsion Laboratory and Australian astronomer Richard Thompson.

Sunspots come and go over an 11-year cycle. The Hathaway team's model predicts that the current solar cycle will reach a peak of 154 sunspots in the year 2000, with an uncertainty of plus or minus 20. This prediction has a narrower range of error than a previous widely accepted prediction, which placed the sunspot maximum at 160 with an uncertainty of 30.

Amateur astronomer Heinreich Schwabe discovered the 11-year cycle in 1843. Since then, people have occasionally tried to predict the stock market based on sunspot activity, but there is no evidence that such a correlation exists.