A large defunct NASA satellite is expected to fall to Earth late Friday afternoon (eastern time), September 23, or early Saturday morning. As it makes the plunge, it will catch fire and break apart, but not all of the 6.5-ton spacecraft will burn up. Debris is expected to reach the surface, roughly 26 large pieces, but the exact location and time of re-entry are still unknown. NASA officials say predictions will become more defined within 24 to 36 hours, however, they do know that the satellite will not be passing over North America.

Debris from the bus-sized Upper Atmosphere Research Satellite (UARS) should fall across an area roughly 500 miles long, and has a 1-in-3,200 chance of hitting a person, which is considerably remote, says agency officials. The satellite’s most likely landing spot is the ocean, which covers almost 75 percent of Earth. NASA estimates any “surviving components” of UARS will land within a zone between 57 degrees north latitude and 57 degrees south latitude–ranging from northern Canada to southern South America.

UARS was deployed in 1991 on a mission to study Earth’s atmosphere, particularly the ozone layer, and was decommissioned in 2005. It has been falling faster than anticipated due to increased solar activity, which can cause Earth’s atmosphere to heat and expand, increasing drag on low-flying spacecraft, according to Space.com. Due to it’s unpredictable nature–it is essentially tumbling out-of-control–scientists won’t be able to pinpoint the satellite’s point of impact until about two hours before re-entry.

The U.S. Strategic Command at Vandenberg Air Force Base in California and NASA are closely watching the plummeting UARS using sophisticated modeling software. I previously wrote about these systems, which are also used to anticipate collisions between spacecraft and space junk, for Technology Review,

To foresee the paths of space junk so that collisions can be avoided, NASA developed one of the world’s most sophisticated predictive models. Called Legend (for “low-Earth to geosynchronous environment debris”), the three-dimensional model simulates the routes of all trackable space objects and even factors in new debris from future crashes. To take uncertainty and randomness into account, hundreds of scenarios are generated using the Monte Carlo method, a set of algorithms that can calculate risk factors in a complex environment. With Legend, NASA scientists use the average of multiple simulations to estimate the number, size, and type of objects that will collide—and approximately how often. Unlike models used by the U.S. Strategic Command Joint Space Operations Center, which detects and tracks large objects and screens active satellites daily for possible collisions within 72 hours, Legend includes smaller fragments and looks far into the future.

In place since 2004, the NASA model is constantly fed with data gathered from the results of ground tests and spacecraft that have broken up in orbit; from telescopes and radars viewing the sky; and from analysis of crater-marked spacecraft surfaces that have returned to Earth. That means new simulations must be run continually. Legend enables scientists to calculate the consequences of a particular breakup or collision and helps them alert managers at the space station that a piece of debris could be in its path. The model also advises soon-to-launch satellites of areas to avoid and will guide scientists as they attempt to develop and launch debris removal technology for the first time.

While the falling satellite is a concern, it is not the first or the biggest spacecraft to come crashing to Earth. Other notable plummets include, NASA’s Skylab in 1979, Space Shuttle Columbia in 2003, and one of the largest ever, Russia’s Mir space station in 2001.

Updates from NASA:

As of 10:30 a.m. EDT on Sept. 23, 2011, the orbit of UARS was 100 miles by 105 miles (160 km by 170 km). Re-entry is expected late Friday, Sept. 23, or early Saturday, Sept. 24, Eastern Daylight Time. Solar activity is no longer the major factor in the satellite’s rate of descent. The satellite’s orientation or configuration apparently has changed, and that is now slowing its descent. There is a low probability any debris that survives re-entry will land in the United States, but the possibility cannot be discounted because of this changing rate of descent. It is still too early to predict the time and location of re-entry with any certainty, but predictions will become more refined in the next 12 to 18 hours.