Researchers have determined, with the assistance of US military satellites, that the risk of Earth being struck by a killer asteroid is less likely than previously believed.

If you've ever gazed up and spotted a shooting star, you engaged in a form of astronomy in which Earths atmosphere serves as a giant detector: Space debris screams through the air, which heats the stuff up and makes it visible.

By noting observations night after night, you could develop a record of how frequently certain sized objects most no larger than a pea meet their demise by running into our planet.

But if you want to know how often larger hunks of cosmic rubble arrive, your job is far more difficult. It could take hundreds or even thousands of years of continuous observations to arrive at a reliable estimate.

The prospect was not an option for Peter Brown of the University of Western Ontario. So he and some colleagues turned to a higher-tech version of the same method, and then applied some fancy extrapolations. The researchers studied observations by U.S. government satellite sentinels that watch for potential nuclear detonations around the globe, 24/7.

More than eight years of data reveal 300 in-air explosions of space rocks ranging in size from large televisions to studio apartments.

Objects in this size range rarely reach the ground. They disintegrate catastrophically but high up. While some of their shards can hit the planet the burned remains of space debris add tons of heft to Earth every day, astronomers say the damage potential is limited.

"There will often be a shower of small stony objects," Brown explained. "But theres essentially no significant amount of energy thats imparted to the ground."

Because most of these events occur over remote regions or oceans (Earth is about two-thirds water) the majority of them go unnoticed except by satellites operated by the Department of Defense and the Department of Energy. Brown and his colleagues used the power output of the explosions to estimate how big each rock was.

The researchers were interested in larger objects, however, those roughly as big as a football field that strike even less often but can get real nasty down here on the ground. These large boulders typically explode, too, but they do so closer to the surface. A shock wave could kill millions of people if one exploded over a populated area.

The last known event like this was in 1908, over the remote Tunguska region of Siberia. Mostly uninhabited forest was flattened for hundreds of miles in every direction.

Until recently, experts thought events like this might occur once per century. A perception had developed in some minds that Earth was due for another mini-cataclysm.

When Brown extrapolated his data on relatively small rocks upward to estimate Tunguska event frequency, he found they probably occur every 1,000 years or so. The results will be reported in tomorrows issue of the journal Nature.

He notes, however, that because his team had less than a decade of data to work with, its possible the actual rate of events is higher than observed. Other researchers have speculated that swarms or streams of asteroids might generate flurries of impacts now and then.

No such events have been recorded in modern times. However, ancient tales and drawings hint at the possibility.

The new analysis dovetails with another recent study that approached the question from the opposite direction. Alan Harris of the Space Science Institute in Boulder, CO used actual tallies of Tunguska-sized rocks discovered out in space to estimate the quantity of smaller rocks that probably exist. He, too, found that Tunguska events ought to occur only about once every 1,000 years.

"We can all worry a little less about the risk of the next hazardous impact," said Robert Jedicke, a University of Arizona researcher who was not involved in the new study but wrote an analysis of it for Nature.

However, unlike the recent Leonid meteor shower, which was well predicted, scientists dont yet have enough data on the populations of large meteoroids and asteroids to say when the next one is coming. They caution that statistics and odds cannot be converted to precise timetables. A giant asteroid might explode disastrously above San Francisco tomorrow, or none might arrive on the entire planet for millennia.

Nature does not deliver doom on any scientists schedule.