The Patriot antimissile system, hailed by U.S. officials as one of the high-tech success stories of the Iraq war, also inflicted some of the most damaging "friendly fire" of the conflict.
The Defense Department has acknowledged that the antimissile system was involved in the downing of two allied warplanes, resulting in the deaths of three airmen. The two aircraft -- one American and one British -- are the only confirmed cases of planes being shot down during the war. Another plane narrowly escaped becoming the third victim of the Patriot system.
"The Patriot should have been stood down until they figured out why it was shooting down planes," said Joseph Cirincione, who directed a congressional assessment of the Patriot after the 1991 Persian Gulf War. "It's bad enough that it happened once. It's unconscionable that it happened again."
The sophisticated ground-based antimissile system rocketed to prominence in the 1991 Gulf War, gaining the nickname "Scudbuster" as it seemed to obliterate Iraqi long-range ballistic missiles heading for Tel Aviv and Riyadh. But after the war, congressional and independent analyses concluded that the Patriot may actually have missed every Scud it targeted.
Now, the friendly-fire incidents in the current campaign have raised concerns that the long-troubled system is still too complex and poorly tested to be deployed.
The military maintains that the system is vastly improved since 1991. It has spent $3 billion on the Patriot's missile and tracking technologies, and, during the current campaign, U.S. Central Command in Qatar said the system destroyed all nine Iraqi missiles it targeted.
"We are well along in our goal of demonstrating its reliability," Lt. Gen. Ronald Kadish, director of the Pentagon's Missile Defense Agency, told a Senate Appropriations Committee panel April 9. He called the Patriot's overall performance "very good."
But at the same hearing Kadish acknowledged there could be lingering problems. When asked whether the downing of the two allied jets and the near-shooting of another were caused by human or mechanical error, Kadish said, "I think it may be both."
He added, "There are investigations underway into each of the three incidents. I think we should wait until they're complete before we begin jumping to conclusions as to where the fault lies."
The Patriot attacks on allied planes were particularly puzzling given that throughout the war, no Iraqi aircraft were aloft.
"Why were the Patriots even shooting at aircraft?" asked Philip Coyle, former assistant secretary of Defense and director of operational testing and evaluation for the Pentagon. "We ruled the skies in Iraq, so almost by definition any aircraft up there was either ours or British."
The Patriot system, designed in the 1970s to shoot down enemy aircraft, should have been able to distinguish between relatively slow-moving planes and speedy rockets fired by Iraqi forces, Coyle said.
"If they can't tell the difference between a missile and an airplane, then they need much more restrictive rules of engagement," he said.
The first friendly-fire incident occurred March 23, when a British Tornado GR4 fighter-bomber was shot down by a Patriot battery protecting an airfield in northern Kuwait. Flight Lts. Kevin Barry Main and David Rhys Williams were killed.
A week and a half later, according to Central Command spokesman Lt. Herb Josey, a U.S. Navy F/A-18C Hornet was shot down by a Patriot missile over central Iraq, killing the pilot, Lt. Nathan D. White.
On March 24, a Patriot battery locked its radar on a U.S. Air Force F-16 Falcon in preparation for firing a missile. The plane's pilot, flying about 30 miles south of the Iraqi city of Najaf, averted disaster by firing first, disabling the battery with a radar-seeking missile. No U.S. soldiers were injured in the incident.
"The F-16 pilot probably knew he was being painted by a ground surface-to-air missile radar, but he may not have known it was a Patriot. He may have thought it was an Iraqi SAM," Coyle said. "The F-16 pilot did the right thing."
Experts said that such friendly-fire episodes stem from the long-range, split-second nature of modern war. A compounding factor is the enormous complexity of the Patriot system, which makes it vulnerable to mechanical, computer and human error.
The Pentagon will look closely at the Patriot's automated engagement mode, in which the system's radar may lock onto a target without operator intervention, as a possible culprit, Coyle said.
At the Senate hearing, Kadish said that possible system defects or a breakdown of the "identification friend or foe" system -- a primary way to identify aircraft in battle -- also will be scrutinized.
All aircraft are equipped with IFF transponders that, when polled by a friendly ground station, will respond with codes to avoid being targeted as an enemy. Something as simple as a broken or inadvertently inactivated transponder could trigger a friendly-fire incident. But given that there were three separate cases, such errors seem unlikely.
Theodore Postol, a physicist at the Massachusetts Institute of Technology and co-author of an influential study criticizing the Patriot performance in the 1991 Gulf War, suggests that the problem may lie with the way the Patriot system interpreted IFF signals. With the Iraqi air force grounded, he said, the return signal "is never going to be a 'foe,' so there's no reason ever to fire at an airplane," he said. "These systems apparently didn't work."
Soldiers operate the Patriot battery -- which includes a command station, radar unit and launcher assembly that can hold as many as 16 missiles -- using computer displays that show digital maps of the sky. Icons indicating friend, foe or unknown move across the screen, much like in an air-traffic control system. When a target such as a jet, robotic drone or missile is confirmed, one or more intercept missiles are launched.
Tim Carey, vice president of the Patriot product line at Raytheon Corp., the prime contractor, said the system is designed to distill an accurate picture from complex and confusing information. For example, each battery relies on two operators to handle the high flow of battlefield data. "When things are hot" -- if a battle is raging -- "both soldiers have exactly the same display up, and they will work together to confirm the targets," Carey said.
The Army has deployed several versions of the Patriot system in Iraq and Kuwait. All the batteries now employ an enhanced radar system developed by Raytheon. Some of the batteries use a new missile -- the PAC-3 -- that vaporizes an incoming missile warhead by striking it with tremendous force, rather than with an explosive charge. A collar of 180 independently fired rocket motors guides the missile to its target.
Each Patriot missile costs as much as $2 million; a battery can cost $225 million. About 50 batteries were deployed during the Iraq war.
Cirincione and other independent missile experts believe that the changes made during the 12-year, $3-billion program to upgrade the Patriot system have never been fully tested.
Only four "operational" tests of the PAC-3 -- simulating some actual battle conditions -- have been conducted. Two of seven PAC-3 missiles destroyed their targets, one hit a glancing blow, and four others missed or failed to launch.
In the same tests, two out of three PAC-2 missiles scored hits. But in each of those cases, they destroyed a drone aircraft -- the slowest, most vulnerable target.
None of the tests involved a Scud, a target that is particularly difficult to hit because of its high speed and erratic flight path.
"The PAC-3 missile has been retrofitted into a system that is basically '70s technology," said Cirincione, now an analyst with the Carnegie Endowment for International Peace.
"We may need a brand-new system. But we'll never know that until you realistically test the Patriot."
Military planners argue that until airborne laser weapons are deployed -- years from now, at best -- the Patriot represents the best missile-defense option for general combat.
The Army's enthusiasm for the Patriot has been bolstered by the system's seemingly perfect record of nine life-saving missile kills in the Iraq war.
But skepticism of such claims is running high because of the Patriot's history of exaggerated accomplishments.
In the 1991 Gulf War, the system also seemed to perform almost flawlessly -- at first.
The Army said that Patriots intercepted 45 of the 47 Scuds targeted. But MIT's Postol analyzed video footage of 28 of those cases and found that not a single Scud warhead was destroyed. In fact, the Patriots often missed the target by hundreds of meters. Such misses were apparently claimed as hits, he said, if the crudely targeted Scuds struck uninhabited areas.
"It's as if you were sitting at home viewing a Scud attack on television, clapped your hands, and then if there wasn't major damage, you claimed to have intercepted it," Postol said.
His results were confirmed by a House Governmental Operations Committee study led by Cirincione in 1992.
"When we scrubbed the data, it turned out that the best evidence supported successful intercepts of between zero and four Scuds," Cirincione said.
Ultimately, the Army reduced its claims from 45 to 24 warhead kills.
So far, the only evidence that a Patriot shot down an Iraqi missile in the current conflict is a photo of an Al Fatah short-range ballistic missile recovered in the Kuwaiti desert.
The photo alone cannot verify that the missile's warhead was destroyed, experts said. Nor can it prove the missile had been intercepted rather than malfunctioning on its own, experts said.
Military officials declined to release details about the nine reported intercepts.
Even if the claim of a perfect Patriot record in the Iraq war can be verified, Postol noted that all of its kills were of short-range weapons such as Al Fatah, Al Samoud or Ababil-100 missiles that experts say are far easier to hit than Scuds.
"Given the history of lying [about the Patriot], the only way I would accept their claims is to see their data," Postol said.