NEAs only survive in their orbits for a few million years. They are eventually eliminated by orbital decay and accretion by the Sun, collisions with the inner planets, or by being ejected from the solar system by near misses with the planets. With orbital lifetimes short compared to the age of the solar system, new asteroids must be constantly moved into near-Earth orbits to explain the observed asteroids. The accepted origin of these asteroids is that main belt asteroids are moved into the inner solar system through orbital resonances with Jupiter. The interaction with Jupiter through the resonance perturbs the asteroids orbit and it comes into the inner solar system. The asteroid belt has gaps, known as Kirkwood gaps, where these resonances occur as the asteroids in these resonances have been moved onto other orbits. New asteroids migrate into these resonances due to the Yarkovsky effect which provides a continuing supply of near-Earth asteroids.
A small fraction of near-Earth asteroids are extinct comets that have lost all their volatile constituents, and a few near-Earth asteroids still show faint comet-like tails. These near-Earth asteroids were probably derived from the Kuiper belt, a repository of comets residing beyond the orbit of Neptune. The rest of the near-Earth asteroids appear to be true asteroids, driven out of the asteroid belt by gravitational interactions with Jupiter.
There are three families of near-Earth asteroids:
Many Atens and all Apollos have orbits which cross that of the Earth, so they are a threat to impact the Earth on their current orbits. Amors do not cross the Earth's orbit and are not immediate impact threats, however their orbits may evolve into Earth-crossing orbits in the future.
Also sometimes used is the Arjuna asteroid classification for asteroids with extremely Earth-like orbits.
Asteroids with diameters of 5-10m impact the Earth's atmosphere approximately once per year, with as much energy as the atomic bomb dropped on Hiroshima, approximately 15 kilotonnes of TNT. These ordinarily explode in the upper atmosphere, and most or all of the solids are vaporized. Objects of diameters of order 50 meters strike the Earth approximately once every thousand years, producing explosions comparable to the one observed at Tunguska in 1908. Asteroids with a diameter of one kilometer hit the Earth an average of twice every million year interval. Large collisions with five kilometer objects happen approximately once every ten million years.
The general acceptance of the Alvarez hypothesis, explaining the Cretaceous–Tertiary extinction event as the result of a large asteroid or comet impact event, raised the awareness of the possibility of future Earth impacts with asteroids that cross the Earth's orbit.
On June 6, 2002 an object with an estimated diameter of 10 meters collided with Earth. The collision occurred over the Mediterranean Sea, between Greece and Libya, at approximately 34°N 21°E and the object exploded in mid-air. The energy released was estimated (from infrasound measurements) to be equivalent to 26 kilotons of TNT, comparable to a small nuclear weapon.
On 5 October 2008, scientists calculated that a little Near-Earth asteroid just sighted that night should impact the Earth on 6 October over Sudan, at 0246 UTC, 5:46 a.m. local time.. The asteroid arrived as predicted.. This is the first time that an asteroid impact on Earth has been predicted before it occurred.
On March 18, 2004, LINEAR announced a 30 meter asteroid 2004 FH which would pass the Earth that day at only 42,600 km (26,500 miles), about one-tenth the distance to the moon, and the closest miss ever noticed. They estimated that similar sized asteroids come as close about every two years.
Astronomers have been conducting surveys to locate the NEAs. One of the best-known is the LINEAR which began in 1996. By 2004 LINEAR was discovering tens of thousands of objects each year and accounting for 65% of all new asteroid detections. LINEAR uses two one-meter telescopes and one half-meter telescope based in New Mexico.
Spacewatch, which uses a 90 centimeter telescope sited at the Kitt Peak Observatory in Arizona, updated with automatic pointing, imaging, and analysis equipment to search the skies for intruders, was set up in 1980 by Tom Gehrels and Dr. Robert S. McMillan of the Lunar and Planetary Laboratory of the University of Arizona in Tucson, and is now being operated by Dr. McMillan. The Spacewatch project has acquired a 1.8 meter telescope, also at Kitt Peak, to hunt for NEAs, and has provided the old 90 centimeter telescope with an improved electronic imaging system with much greater resolution, improving its search capability.
Other near-earth asteroid tracking programs include Near-Earth Asteroid Tracking (NEAT), Lowell Observatory Near-Earth-Object Search (LONEOS), Catalina Sky Survey, Campo Imperatore Near-Earth Objects Survey (CINEOS), Japanese Spaceguard Association, and Asiago-DLR Asteroid Survey.
"Spaceguard" is the name for these loosely affiliated programs, some of which receive NASA funding to meet a U.S. Congressional requirement to detect 90% of near-earth asteroids over 1 km diameter by 2008. A 2003 NASA study of a follow-on program suggests spending US$250-450 million to detect 90% of all near-earth asteroids 140 meters and larger by 2028.
Asteroid impact predictions often make the news. The next few observations show an increasing chance of impact, but then further observations rule out any impact. The reason for this pattern is shown in the diagram at the right. The ellipses in this diagram show the likely asteroid position at closest earth approach. At first, with only a few asteroid observations, the error ellipse is very large and includes the Earth. This leads to a small, but non-zero, impact probability. Further observations shrink the error ellipse, but it still includes the Earth. This raises the impact probability, since the Earth now covers a larger fraction of the error region. Finally, yet more observations (often radar observations, or discovery of a previous sighting of the same asteroid on archival images) shrink the ellipse still further. Now the earth is outside the error region, and the impact probability returns to near zero.