NASA was warned nine years ago that the space shuttle could fail catastrophically if debris hit the vulnerable underside of its wings during liftoff -- the very scenario that may have brought down Columbia.

After receiving the warning, NASA made changes in materials and flight rules to lessen the risk of debris breaking loose, Paul Fischbeck, an engineering professor at Carnegie Mellon University who conducted the 1994 analysis, said Tuesday.

"There are very important tiles under there. If you lose the tiles on those stretches ... it can cause the shuttle to be lost," he said.

The underside of the left wing is where NASA is focusing its investigation into Saturday's disaster, which killed seven astronauts.

A spokesman at NASA's Johnson Space Center, David Drachlis, said Tuesday night that no one was available to comment on the report.

Also Tuesday, the space agency sent teams to check out reports of shuttle debris found as far west as California and Arizona -- material that could shed light on the earliest stages of Columbia's breakup. And investigators also are getting new military images of the spacecraft's fiery, dying streak across the country from an Apache helicopter.

Fischbeck and a colleague at Stanford University studied the damage caused by debris during the first 50 shuttle launches and concluded that on average, 25 thermal tiles per flight sustained damage of at least one inch.

He said his risk analysis showed that the most vulnerable spots on the shuttle were the undersides of both wings close to the fuselage, and right under the crew compartment. To reach that conclusion, he weighed three factors: which tiles were most likely to be hit by debris, which tiles endured a lot of heat on atmospheric re-entry, and which tiles had critical systems underneath them.

It is not clear exactly where under the wing Columbia was hit, but just before the shuttle break up, temperature spikes were detected around the left wheel well, which is close to the fuselage, and on the left side of the fuselage itself.

Fischbeck said NASA "took a lot of our advice to heart" and made improvements such as changing the foam insulation on the top of the booster rockets and allowing less ice to be present on the fuel tanks before liftoff. Ice forms on the tank because of the super-cold liquid hydrogen and oxygen inside.

The investigation into the Columbia disaster is focusing on the possibility that a 2-pound, 20-inch chunk of foam insulation from the shuttle's external fuel tank fell off during liftoff and hit the left underside of the wing, causing damage to the thermal tiles that doomed the ship.

Fischbeck stressed that no one knows with absolute certainty yet whether the debris brought down Columbia or where precisely that debris hit.

"There's risk associated with any engineered system, whether it be your school bus or a double-hull tanker or whether it be the space shuttle. There's always a combination of events that can lead to an accident," he said. "You cannot make things risk-free."

He added that in this case, the question is: "Did they do all they should have done?"

Fischbeck's report looked at damage to the tiles from the insulating foam on the external fuel tank and the booster rockets, as well as from ice. He found that there was a greater likelihood of the right side of the shuttle being hit than the left; he put the odds at about 60-40.

He said while the tiles under the right wing had a slightly higher likelihood of being struck, "if the ones under the left wing were hit, then they were equally damaging to the orbiter."

Fischbeck said during the first 50 shuttle flights, from 1981 through 1992, the damage per flight ranged from three tiles to 150 tiles.

The foam and the tiles have been a source of concern at NASA practically from the start.

Over the years, foam insulation often damaged the tiles. In fact, soon after NASA stopped using Freon in the foam, for environmental reasons, Columbia sustained significant tile damage during a 1997 liftoff because of flyaway foam, according to a report by NASA engineer Gregory Katnik.

Katnik raised the possibility at the time that the new foam concoction was not compatible with the severe conditions of takeoff. The foam recipe apparently was altered somewhat after the 1997 incident.

"The thing of this is, almost since Day One, the insulation has been a pain. Pieces break off," said Seymour Himmel, a retired NASA executive who served two decades on an aerospace safety panel and looked into the potential dangers of the foam.

While the foam is a lightweight, polyurethane, spray-on material that goes on like shaving cream, it hardens like styrofoam. And given the speed at which shuttles hurtle into space during takeoff, it can have a devastating effect. Moreover, the black, silica glass fiber tiles that cover the belly of the shuttle are famously fragile, so much so that even a bump or nudge can cause cracks or dings.

NASA has said that during Columbia's mission, its engineers carefully analyzed the potential damage done by the foam and concluded that the crew was in no danger.

On Monday, Michael Kostelnik, deputy associate administrator for spaceflight, said that the chunk that fell from the tank and hit Columbia's tiles was "probably the largest piece we've had" of any shuttle mission.

Kostelnik noted repeated attempts to fix the foam problem. "There's been a lot of activity and a lot of focus on the foam in particular and how to attach it, so this has been a continuing effort," he said.

Even some of the pros have trouble accepting the notion that foam could have caused the catastrophe.

"I am absolutely stunned," said astronaut Mike Mullane, who rode Atlantis into orbit in 1988. "I can see it scratching or even gouging a couple of tiles. But God, the idea that it could compromise the system. I don't know, but I just have a hard time believing that."

He recalled that a tip of one of his booster rockets broke off during launch and slammed into the belly of the shuttle, damaging a few hundred tiles. One came off entirely. Yet Atlantis and its crew of five still made it safely home.

Some of the broken-off pieces of insulation, ranging from flecks to chunks, hit the shuttle and usually do nothing more than ding or scratch the tiles. Some miss the shuttle altogether.

The foam has undergone "a continuing series of process changes, both in terms of how you apply it and where you apply it, under which conditions, as well as the materials," Himmel said.

Lockheed Martin Corp. makes the 154-foot fuel tanks, but a long list of subcontractors provide the foam and its ingredients. The silica tiles on the shuttle's belly were developed and manufactured by Lockheed.

The tiles protect shuttles against the searing heat of atmospheric re-entry. If some are missing in an especially vulnerable area, it could set off a chain reaction that could destroy the shuttle.

Shuttles almost always return to Earth with marred or missing tiles, the result of debris smacking into them during liftoff.

Himmel said he and other members of the Aerospace Safety Advisory Panel, created after the deadly 1967 Apollo spacecraft fire, never included in their reports any reservations about the vulnerability of the thousands of tiles that cover shuttles.

"It was so well-known, it would have been superfluous to say, `If you lose a lot of tiles, you've bought the farm,"' Himmel said.

Retired aerospace engineer Tom McKeever used to see the tiles damaged all the time. "The whole tile situation has been not only fragile in concept, but I believe somewhat fragile in design," he said.

During Columbia's launch last month, it is possible that the foam was coated with ice. That would have made it a more dangerous projectile. Shuttle program manager Ron Dittemore said he hopes to know about ice by the end of the week.

During the launch, the strip of foam peeled away from the tank 81 seconds into the flight. The shuttle was traveling at 2 times the speed of sound at the time, or just over 1,900 mph.