An asteroid is heading for Earth. With just days to go before the collision a beefed-up space shuttle is sent to intercept it. A brave team of astronauts and oil-rig workers drills deep into the space rock, plants a nuclear bomb and blows it in two. The two halves fly apart and miss the Earth.

Dream on!

The idea of blowing up an asteroid makes for good movie scripts, but is not the way to do it in the real universe. Many of the fragments would remain on a collision course and like the blast from a shotgun; the fragments can do up to ten times as much damage as the original, intact object.

In any case, Erik Asphaug from the University of Southern California has modeled "rubble-pile" asteroids and finds that blowing them up with bombs may be much more difficult than with asteroids made of solid rock. It is a bit like the difference between hitting a sandbag and a solid sandstone block with a sledgehammer -- the sandbag absorbs the impact with little disruption but the sandstone block shatters.

"Stand-off" nuclear explosions are favored by some scientists (see below) and might work with both solid and rubble-pile objects.

A nuclear bomb is detonated several hundred yards away from the object. Surprisingly, it is the intense radiation generated by the explosion that does the job. In one scenario, the radiation grills one half of the asteroid and causes a very thin surface layer to vaporize and fly off into space.

"Yippee-Ki-Yay …Hmmm, maybe not." Bruce Willis meant well in "Armageddon" but he could have doomed us all

Tens of tons of material blasting off the asteroid at high speed would be sufficient to jolt the asteroid in the opposite direction. The effect is like the recoil of a rifle -- a small bullet moving at high speed causes the heavier rifle to recoil at low speed.

One thing most scientists agree on is there is no need to maintain an arsenal of nuclear weapons in space ready to intercept rogue asteroids. They also point out that there are ways to deflect asteroids that don't require nuclear explosions and we should be looking at these methods more closely.

In theory, an asteroid that is found to be on a collision course with our planet can be deflected to avoid an impact.

The deflection involves changing the asteroid's course with a sideways push or, preferably, changing its orbital speed so that it arrives before or after, rather than when Earth crosses its path. In either case the deflection is far more effective if it can be carried out years or decades ahead of the predicted collision.

For example, after twenty years, a nudge of just 1 m.p.h. (1.6 kilometers per hour) would change an asteroid's location in space by about 170,000 miles (273,500 kilometers). That is more than halfway to the moon.

Recent discoveries suggest that deflection of some Earth-threatening asteroids may be easier than first thought. Most schemes for nudging asteroids into a safer orbit assumed a single catastrophic encounter with Earth. This meant changing the course of the object by at least 4,000 miles (6,300 kilometers) -- the radius of Earth.

Nuclear Deflection: A safer, more effective procedure

Alan Harris, from NASA's Jet Propulsion Laboratory, explains that scientists now realize an asteroid will usually make several close passes by the Earth before a collision occurs.

The recently discovered 1000-yard (1-kilometer) wide asteroid designated 1999 AN 10 provides an instructive example. It will make a close pass of Earth every few decades. During each pass the asteroid is deflected slightly by the Earth's gravity.

Astronomers in Italy have calculated that a critical deflection could occur in 2027. This would involve the asteroid passing through an imaginary hoop in space they call a "keyhole". If the asteroid were to pass through this keyhole, which is only about 60 miles (100 kilometers) across, then it would collide with the Earth on its return in 2039.

When the initial calculations were made, astronomers didn't know the orbit well enough to determine if it might pass through the keyhole. After important follow-up observations were made they have now pinned down the orbit enough to be sure that it will not pass through any keyhole in 2027 and there is no chance that it will collide with Earth in the next century or so.

If, however, they had determined instead that there was a chance it would pass through a keyhole in 2027, then a mission to place a transponder, like a radio homing device, on the asteroid would have been wise so that its orbit could be determined precisely.

Harris explains that such a high level of precision would likely be required to determine for sure if the asteroid were on a course through a keyhole and, if it came to be, to measure the success of any deflection efforts. In this case a deflection of just a few hundred miles prior to the 2027 keyhole event would be all that was needed to avoid the 2039 collision.

Deflection of dangerous asteroids that are not in a "keyhole" orbit is more difficult because a larger change in course is required. The task is still feasible provided that sufficient warning time is given.

If a serious global effort is made to discover most large near-Earth asteroids within the next decade, then we should have decades, or even centuries of warning before a devastating impact. With such lead times only a relatively small nudge is required to change an asteroid's course so that, decades later, it will miss Earth.