The radar signature of an object floating away from the shuttle Columbia on the second day of its mission matches up well with a wing leading edge carrier panel, members of the Columbia Accident Investigation Board revealed today. The tile-covered carrier panels form a smooth surface between the edges of U-shaped reinforced carbon carbon - RCC - panels making up the wing leading edge and heat-shield tiles permanently bonded to the wing's lower surface. If Columbia began its ill-fated re-entry with a missing carrier panel, enough heat could have entered the left wing to trigger the catastrophic chain of events that led to the shuttle's destruction.

But board member Roger Tetrault said other explanations are possible and that it's still not certain a carrier panel is what was separating from the orbiter on flight day two. Asked if one could conclude a missing carrier panel was the root cause of the disaster based on all the data to date, Tetrault said "your logic could lead to someone's death."

Even though photo analysis shows foam debris from Columbia's external fuel tank hit Columbia's left wing leading edge around RCC panel 6 - and that it could have hit the carrier panel immediately behind RCC panel 6 or an adjacent panel - that doesn't mean the breach that doomed Columbia began at that location.

At least not until engineers rule out other scenarios that are still on the table.

"You can't rule out all of these (other scenarios) and make this leap of faith that something you know happened between (RCC) panels 6 and 12 actually happened in this particular area," Tetrault said. "You just can't make that leap of faith."

But he agreed that the loss of a single carrier panel almost certainly would have allowed enough heat to enter the wing behind the leading edge to trigger the temperature increases seen early on. And "early on" is clearly the operative expression when it comes to Columbia's catastrophic descent.

Ongoing analysis of data recovered from a salvaged telemetry recorder shows the onset of heating in Columbia's left wing began 270 seconds after the shuttle entered the discernible atmosphere. That's three minutes and 26 seconds earlier than the previously known first indication of anything unusual - an unexpected yawing motion - and three minutes and 38 seconds before the previous first indication of elevated temperatures in a brake line in the left main landing gear wheel well.

As it turns out, that earlier brake line reading, a 1.5-degree Fahrenheit increase detected at 8:52:17 a.m., may have been normal. Tetrault said NASA investigators have identified at least 13 previous shuttle flights with so-called 1.5-degree "bit flips" in the first eight minutes and 20 seconds after atmospheric entry, or EI+500 seconds. In 11 of those missions, bit flips occurred earlier than the one seen aboard Columbia.

"We are finding brake line temperature D (sensor), which established the first sensor previously, is probably not off nominal at the time that's on the timeline," Tetrault said. "It's probably later than that. That's an important point because sensor D goes off first and it is in the aft of the wheel well and it is high in the wheel well, which tends to indicate heat is coming from the aft part of the wheel well moving forward.

"If you really look at it, it is much more likely that the heat is coming from the forward side of the wheel well going aft," he said. "Temperature D is probably nominal for a much longer period of time."

In any case, more than a hundred engineers working over the weekend have managed to extract key data from the recovered payload experiments recorder, or OEX recorder, that was found by search crews near Hemphill, Texas, on March 19. Tetrault briefed reporters today on the performance of four specific sensors in the left wing.

Columbia fell into the discernible atmosphere - entry interface, or EI - 400,000 feet above the Pacific Ocean northwest of Hawaii at 8:44:09 a.m. Sensor G9921, a strain gauge located in the interior of the shuttle's left wing roughly in line with RCC panel 9, began to show signs of unusual stress at 8:48:39 a.m., or EI+270 seconds. That was 206 seconds earlier than the first signs of unusual yaw moments at EI+476 seconds.

Twenty seconds after sensor G9921 went "off nominal," a temperature sensor mounted on a support brace on the outboard side of RCC panel 9 began sensing unusual temperatures. This sensor was located under insulation in a cavity behind the U-shaped RCC panels just in front of the wing spar. It began sensing high heat at 8:48:59 a.m. and then failed at 8:52:19 a.m., after reading 50 degrees, presumably because the hot air rushing into the wing cut through its wiring.

The next sensor to show unusual readings was located on the front side of Columbia's left orbital maneuvering system rocket pod. At 8:49:53 a.m., sensor 9220 started seeing unusually low temperatures. Nine minutes later, the sensor showed a rapid increase in heating, climbing as high as 1,200 degrees Fahrenheit. The normal reading is 600 degrees.

"First, we think there is a change in the mass flow, which moves the mass flow probably below the OMS pod, and then there is something else that happens that brings the temperature rapidly up, which may be burning or burning aluminum or a number of other effects we have to go look at," Tetrault said.

At 8:51:14 a.m., or EI+425 seconds, sensor 9895, located directly behind RCC panel 9 on the inner side of the wing spar, began recording an unusual temperature rise. One minute and 35 seconds later, the sensor failed after reaching 450 degrees.

Analysis of the OEX data is continuing and the investigation board plans to wrap that information into NASA's existing re-entry timeline, which currently stands at revision 15.

The OEX tape contains two types of data. One set ends at 9:00:13.4 a.m. and the other stops at 9:00:19.4 a.m. The latter is 15 seconds beyond the point where telemetry that was downlinked in realtime came to a sudden stop on Feb. 1. During a final two-second burst of downlinked telemetry, the shuttle was seen to be yawing rapidly to one side. Main vehicle breakup, based on video analysis, began around 9:00:21 a.m.

The OEX data likely will help engineers pinpoint how Columbia began breaking apart and what major components tore off first. In addition, the OEX data will shed light on what was going on during a 25-second dropout in the downlinked telemetry that occurred between 8:59:37 a.m. and 9:00:02 a.m.

At the front end of the timeline, the OEX data "will show us that when we start looking at where the electrical lines were cut - a lot of them cut 540 seconds after EI - that it will begin to help us to localize where in the front edge the problem is," Tetrault said. "Up until now, if I took collectively all the data we had, I could in my mind locate this event somewhere between RCC panel no. 5 and no. 12. I think this will help shrink that down to some much narrower (focus).

"When you begin to put that where the photos have shown where the foam hit, I think you can start making some assessments of what are the probabilities of that being an initiating event, if you will. I wouldn't say necessarily the cause, but there may be an initiating event that leads to a whole series of other things that winds up at the accident."

One possible initiating event is the loss of a carrier panel on flight day two. NASA sent 29 leading edge components and tiles to Wright Patterson Air Force Base for testing to match up their radar signatures with the debris seen leaving Columbia. As of today, a carrier panel is the only item that has not been ruled out.

"With 3,100-plus observations of Columbia by the DOD, we've got a lot of radar cross section feedback," said board member Maj. Gen. John Barry. "There have been 29 various materials examined at Wright Patterson Air Force Base and we've concluded that right now, that only the carrier panel remains a viable candidate for the day two object. ... We think with the (attachment hardware) and the carrier panel, it gives us one of the best candidates we've had so far after reviewing all the testing. But we still have some more testing to do."

The carrier panels are mounted in place by bolts at each end. The bolts run through the heat-shield tiles on their outer surface and the bolt holes are filled with ceramic plugs. The bolt holes are relative weak spots and tests are planned to determine what sort of damage might have been caused by an impact on one of those tiles during launch.

Tetrault said video of Columbia's launch and the subsequent foam impact had been enhanced by the National Imagery and Mapping Agency, or NIMA, and that investigators now believe the shuttle was struck just once, by a piece of foam measuring 24 inches by 15 inches by 5 inches, with an uncertainty of a few inches in any dimension. The debris, thought to have weighed about two pounds, slammed into the left wing at 640 feet per second, or 436 mph.

The foam hit the left wing in a two-foot-wide footprint centered on the lower surface of RCC panel 6. The footprint includes parts of two carrier panels.

Tests are on tap at the Southwest Research Institute in San Antonio, Texas, to fire foam debris into leading edge panels, carrier panels and other hardware to determine damage scenarios. Those tests originally were scheduled to begin early next week, but they will be delayed a week or so to give engineers more time to develop procedures incorporating foam samples closer to the size the new NIMA analysis indicates.

On another front, Barry said engineers have cut into the insulating foam of an external tank similar to Columbia's and found numerous voids where the material is bonded to the tank structure. The area in question is the so-called bipod ramp, an aerodynamic slope in the insulation that protects struts holding the nose of the shuttle to the tank from aerodynamic stress.

Engineers looking under the foam in the bipod ramp area of tank 120 found 14 voids, or open spaces, under bipod ramp foam on the right side of the tank and 18 on the left. Such voids are potential trouble spots because extreme low temperatures in the tank during launch can cause air to liquefy in the voids and then explosively evaporate during ascent, blowing foam off in the process. That mechanism is one possible explanation for the foam that came off during Columbia's launch.

To double-check their findings to date, engineers plan to cut into the foam of a tank that is virtually identical to Columbia's - tank 94 - to find out if similar voids are present.

On yet another front, investigators are looking into the formation of pinhole defects in the RCC panels themselves. Barry said today any given RCC panel has between 20 and 40 such age-related pinholes. Any such defects greater than 0.04 inches across are repaired or refurbished. Virtually all of Columbia's leading edge panels had experienced such repairs or refurbishment and investigators are looking into how that process is carried out.

As it turns out, the pinholes may be caused by material from launch pad primer that causes oxidation on the RCC panels. But that remains to be seen.

Finally, Tetrault said the OEX data showed no clear evidence any kind of response from the foam impact on the left wing. But sources familiar with the OEX analysis say a pressure sensor located on the lower surface of the wing, roughly in line with RCC panel 9, shows a small "spike" at 84 seconds that could be the result of post-impact cloud of foam debris sweeping past the detector. No other sensors detected anything unusual.