Astronomers have long stated that a large number of meteoroids frequently impact the Earth. Such incidents are rarely reported despite the enormous amount of energy released. However, newly-declassified documents show that secret military surveillance systems have been detecting such events.

Military satellites have been watching huge meteoroids slam into Earth's atmosphere for nearly two decades.

IN A CELEBRATED 1983 paper, Caltech planetary astronomer Eugene M. Shoemaker calculated that every year, on average, a fragment of asteroid or comet self-destructs somewhere in Earth's atmosphere with the kinetic-energy equivalent of 20,000 tons of TNT, or about |10.sup.24~ ergs. (The devastating Tunguska event in 1908, by comparison, delivered the equivalent of about 10 million tons of TNT.) Such large meteoroids would presumably be some 10 meters across, weigh upward of a thousand tons, and arrive at between 15 and 20 kilometers per second.

Fortunately, nonmetallic objects of this size are annihilated at altitudes too high to harm anything on the ground (S&T: March 1993, page 15). Still, no matter how high, a 20-kiloton explosion is not easily overlooked. It would, for example, eclipse the yield of the fission bomb that destroyed Hiroshima in 1945. Furthermore, in recent years the 36-inch Spacewatch telescope in Arizona has discovered numerous house-size objects hurtling through space very near the Earth. Based on these close calls, former Spacewatch team member David L. Rabinowitz (Carnegie Institution of Washington) concludes that Shoemaker's estimates of encounter rates were too conservative by 10 to 100 times. In the Astrophysical Journal for April 10, 1993, Rabinowitz counters that Earth should endure a 20-kiloton blast roughly every month and receive hundreds of kiloton-yield jolts annually.

Surprisingly, these cosmic cannonades are rarely reported -- only a handful are known from the last few decades. One was a dazzling, nighttime fireball over south-central British Columbia in March 1965, another a widely observed, daytime bolide that grazed Earth's atmosphere over the Rocky Mountains in August 1972. The sparse sightings hardly jibe with predictions of weekly events. So what's going on?

Much of the answer lies in an unprecedented body of spaceborne observations declassified and released in October by the U.S. Department of Defense. The report details a 17-year record of sightings made from orbit by satellites under the control of the U.S. Air Force Space Command. It will appear as a chapter in the book Hazards Due to Comets and Asteroids, to be published next year by the University of Arizona Press.

According to Edward Tagliaferri (ET Space Systems), the report's principal author, infrared scanners on military satellites have recorded a total of 136 atmospheric explosions since 1975, an average of eight probable meteoroids per year, each with an energy of roughly 1 kiloton or more. That's just the kind of observational input cosmic oddsmakers were hoping for. "I've been aware of this data for a long time," says Shoemaker, "and I've been waiting for the time when this could be released." Much of the credit for getting the once-secret database declassified goes to Simon P. Worden, a former astronomer who until recently headed a division of the Ballistic Missile Defense Organization (the "Star Wars" group).

Although the report neither discloses the identity of the surveillance system nor reveals details about its sensors, other sources suggest that the data have come from early-warning satellites of the Defense Support Program. Operating at the geosynchronous altitude of 38,000 km, the DSP sentinels use sophisticated detectors to monitor the Earth's upper atmosphere for rocket plumes and nuclear explosions. One type of sensor "stares" continuously at the entire disk. The high-altitude flashes it looks for are both brief and bright, making them detectable even in daylight against the slowly varying background of Earth lower down. If the airburst is vivid enough, a companion, visible-wavelength device then pinpoints its location.

Another sensor system uses arrays of supercooled detectors tuned to 2.78 microns, an infrared wavelength absorbed virtually completely by water vapor in the lower atmosphere. Married to large Schmidt optics with a 3.6-meter focal length, the array sweeps over Earth's infrared-dark disk every 10 seconds with enough sensitivity and spatial detail to pick up the hot exhaust of even modest ballistic missiles.

The new accounting is still well shy of the hundreds of comparably powerful annual events predicted by Rabinowitz. For example, the brightest flash recorded from orbit since 1965 probably resulted from a kinetic-energy yield of roughly 5 kilotons. One reason for the shortfall is that many airbursts go unreported, even though satellites detect them, because the systems and their handlers are watching for hostile activity and often either overlook or ignore natural events. Furthermore, the meteoric flashes last only a second or two, so on average the scanning infrared arrays miss at least four events for every one they glimpse. Allowing for such gaps, Tagliaferri estimates that at least 80 kiloton-yield meteoroids probably strike the Earth every year, a bombardment frequency higher than that derived by Shoemaker but still below the rate extrapolated from the Spacewatch discoveries.

Notably, of the 136 infrared events, only three turned up in records from the visible-light burst detectors, which have been flown since the early 1980s. Coincidentally, all three caught the attention of two satellites simultaneously. One occurred some 30 km above the western Pacific Ocean on October 1, 1990 -- a 2-kiloton blast right in the midst of the Kuwait-Iraq conflict. An observer on the ground below would have seen part of the sky briefly blaze like the Sun and would have heard a loud, low rumble soon thereafter. "Had this occurred over Kuwait it could have been a sticky situation," Worden observes. "We could tell it was natural, but they could not." Impact specialists hope the release of the previously classified records represents the first step in a long-term partnership with the surveillance community. They are pushing to have spacecraft collect such pivotal data more reliably and have it distributed openly. But Tagliaferri acknowledges that the existing systems have national-security objectives that simply cannot be compromised. One possibility, currently under negotiation, is to include comparable detectors on civilian satellites like those in the Global Positioning System.

Meteor researcher Douglas O. ReVelle (Los Alamos National Laboratory) notes that this is not the first time a defense-related system has yielded clues about the extraterrestrial infall rate. A global network of sensors operating between 1960 and 1974 (some from the rooftops of selected U.S. embassies) detected low-frequency acoustic waves from a handful of powerful airbursts that were probably meteoric in nature. One was believed to result from a small asteroid perhaps 20 meters across that fell harmlessly between South Africa and Antarctica on August 3, 1963 -- even though it packed the explosive punch of a half million tons of TNT!

With so many cosmic bombs bursting overhead one still has to wonder why there are so few visual reports of very bright bolides. Peter Brown, director of the International Meteor Organization for North America, offers at least two possible explanations. First, it may be that much more of the fireballs' energy is released in the infrared (where they're being detected) than at visible wavelengths (where they're not). Also, only in the last few years have fireball reports been collected on a systematic basis. In the future, he adds, the IMO and other meteor networks will be working more closely with their military counterparts to create a more complete record.