A Predator pilot in California saw a “splash” on his video screen as a Hellfire missile slammed into a small group of people thought to be Taliban fighters. Almost immediately, he knew something was wrong. Other figures rushed into the field of view but showed no signs of fighting their way to the scene.
The morning of April 6, 2011, in Afghani-stan’s Sangin River Valley turned out to be a Predator pilot’s worst nightmare, and one that prompted intelligence analysts to gain a greater voice during the fast-paced discussions preceding the firing of weapons — something Air Force intelligence officers had been pushing for since 2009.
The Hellfire missile killed Marine Staff Sgt. Jeremy Smith and Navy Hospitalman Benjamin Rast, the mission’s medical specialist, during a mission to sweep a stretch of farmland and residential compounds along a key road in the Sangin Valley.
A 385-page investigative report of the Sangin fratricide incident — obtained by our sister publication Air Force Times under the U.S. Freedom of Information Act — paints a stark portrait of poor situational awareness during the mission. The report redacts the names of those involved and finds no one negligent. But it recommends changes in Predator processes that in January were blessed in a formal Joint Urgent Operational Need statement.
By the end of the year, Air Force intelligence analysts at each of the service’s intelligence fusion sites are to have the ability to converse by voice with Predator crews instead of relying on digital chat messages.
This account of the Sangin fratricide incident is based on transcripts of interviews conducted by Marine Corps investigators and the detailed timeline contained in their report. All times are local Afghanistan time.
The ground troops had been working together for about a month in an area known for daily improvised explosive device and small-arms attacks. The team left the patrol base in vehicles shortly after 7 a.m., arriving minutes later at their assigned stretch of Route 611. Part of the team stayed behind as a mounted element while the rest of the team struck out on foot.
In a fateful decision, the dismounted squad leader dispatched one part of his force to sweep the road and western shoulder, while another was deployed farther to the west to provide overwatch for the convoy.
Almost immediately, the overwatch team encountered enemy forces and gave chase, halting 200 to 250 meters to the west of the road. The mission had called for securing the route out to a distance of 100 meters, but the force had chased the enemy to an “unanticipated position.” The ground commander had lost track of his troops: “The on scene ground force commander believed that he had one dismounted [force] adjacent to the route when in fact he had two,” investigators said.
Complicating matters for the ground commander, he had to rely only on “communicated position reports” to know where his forces were. It was this loss of situational awareness that triggered a deadly chain of events, investigators said.
At the Predator ground station at March Air Reserve Base, Calif., the next 25 minutes would be punctuated by rapid preparations for the strike, conflicting demands for mapping software at the control site, and a confusing burst of online chat messages from analysts at the Air Force’s Distributed Ground System fusion site in Terre Haute, Ind. The analysts there were studying the video in real time, as a backstop for the Predator crew in California. They could send instant messages to the Predator crew, but could not hear or speak to them.
At 8:40 a.m., the Predator crew spotted the heat signatures of three people. At 8:41, a burst of muzzle flashes emanated from the group.
A critical question was the direction of the muzzle fire. If the direction was east, toward friendly lines, they were potential insurgents trying to kill U.S. troops. If it was west, away from friendly lines, they might be U.S. personnel or other allies.
The Predator crew and the forces in Afghanistan proceeded as if the muzzle fire were coming toward friendly positions. Only minutes earlier, ground forces had reported contact with the enemy over the Predator crew’s radio link. As the Predator circled over Sangin, its ARC-210 radio received transmissions from the ground and fed them into the drone’s satellite link, which relayed them to the Predator control station in California. Given the situation, the Predator crew was furiously scanning for targets to strike with one or more of the plane’s Hellfire missiles.
At 8:41, the JTAC in Afghanistan sent the “9-line” message containing coordinates for a strike once approved by the ground commander.
The analysts at DGS Indiana (DGS-IN) also saw the targets in the video feed. Their leader chimed in with a series of online messages raising questions about their identification as enemy. At 8:40, DGS IN reported “3 friendlies in FOV,” then 40 seconds later, “pers are shooting W,” and almost immediately after that, “disregard,” “not friendlies,” “unable to discern who pers are.”
If the DGS-IN analysts meant that the figures were too grainy or shadowy to say for sure whether they were U.S. troops, but that the enemy would not be firing in that direction, they were unable to convey this information unambiguously, partly because they did not have a voice link to the Predator crew. Online chat was the only recourse, other than picking up a telephone and calling the control station, which was frowned on in such an intense moment.
After the strike, a member of the DGS team in Indiana told investigators about his team’s reservations about the strike. “At no time, to my knowledge, did any of the crew members feel comfortable firing upon the personnel,” according to written testimony from an Air Force staff sergeant stationed in Indiana, whose name was redacted.
Giving analysts a literal voice during preparations for a missile strike was one of the key recommendations of the investigation. If Indiana “had a transmit capability to the [Predator crew], DGS-IN could have ensured that the entire BLESSE 13 [Predator] crew immediately received their conflicting assessment regarding the direction of fire away from friendly forces,” the investigators said.
With no voice link, DGS-IN kept up the chat. At 8:45, DGS-IN underscored its concern about the direction of fire in a private “whisper” message seen only by the Predator mission intelligence coordinator and his team: “pers are shooting W and the convoy is to the E.”
The Predator pilot, who would be in charge of releasing the missile, never knew about this whisper chat. Later, investigators explored why among the mission intelligence coordinator, the trainee coordinator and the senior mission intelligence coordinator, none told the pilot about the chats. The attitude among the coordinators, said the mission intelligence coordinator, was to “trust but verify” information, including such things as conflicting accounts about muzzle fire.
With the clock ticking down toward the missile launch, the mission intelligence coordinators scurried to assess the conflicting information about the direction of fire. They wanted to load up the local terrain and view it in the same orientation that the Predator’s camera was pointing in. They would replay the video of the muzzle fire to view the direction of the fire relative to Route 611, which was not visible in the soda-straw view from the Predator.
They tried to use the station’s FalconView software, but the pilot needed it. So they moved over to a station that had a projector for looking at Google Earth imagery. They were unable to get the view they desired before the missile was launched.
“We never made an assessment because we never had a chance to review the feed,” the mission intelligence coordinator said.
According to investigators, one or more members of the ground team realized they had made a targeting mistake “seconds before the delivery of fires.” With the Hellfire in flight for just 17 seconds, there was no way to do anything.
Moments later, the pilot saw reports of friendly casualties in the online chat. The pilot was stunned. He had been certain that the direction of muzzle fire was toward friendly forces. But when he reviewed the video in the briefing room, he saw that he was mistaken. He “could not maintain bearing” and left the building, where he was met by a chaplain.
What if the pilot had known about the concerns over the direction of fire relayed in the private chat? The pilot told investigators he was “90 percent likely” to have informed others and would have “considered putting a hold on the engagement.”
The investigators recommended that in the future, any intelligence or information that could prevent fratricide or a violation of the Law of Armed Conflict be posted to the crew’s main chat room so that the entire crew would see it. However, despite the pilot’s comments and their own recommendation, the investigators said they were doubtful that posting the conflicting information would have changed the outcome in Sangin on April 6.
It was “likely that the strike would have occurred anyway due to the ground force’s perception of friendly disposition and the source of enemy contact,” the investigators said.
In a January interview with Air Force Times, Air Force Maj. Gen. Robert Otto said the investigators’ conclusion “rang true” about the near-inevitability of the mistake, given the circumstances.
“You have to understand, the Sangin Valley is bad-guy territory,” Otto said. “If you don’t know where the friendlies are, it’s pretty difficult for you to know [what] to overturn based on what one of the supporting forces thought.”
Otto is commander of the Air Force Intelligence, Surveillance and Reconnaissance Agency.
Even though Otto isn’t sure a voice link would have saved the day, he said U.S. forces are taking seriously the investigator’s recommendation to install a transmit-and-receive voice capability at all DGS locations.
“We’ve been able to get broad agreement from all the people that would have input in this that it’s the right thing to do,” he said. “We just got an urgent operational need [statement] to do just that from the warfighter down range.”
Still to be worked out, Otto added, are the funding source and technical aspects to ensure security. Switching from listening mode to two-way conversations won’t be as simple as flipping a switch. “There are technical and procedural rules that need to be followed. Some of those rules require installation of additional hardware to enable the two-way communications,” Otto said.
The right balance also will need to be struck on the issue of voice communication procedures, he added. “There’s a point when a discussion’s over,” he said. “And while we always want somebody to have the ability to speak up when they fear a rule-of-engagement violation or wait a minute, there’s women and children, you have to balance fighting a war by committee with a ground-force commander who is presumed to have the situational awareness and has the authority to say, we need to strike this target.”
In the Predator control stations, mission intelligence coordinators are wary of compli- cating matters for pilots and sensor operators, a fact that is brought out in the report.
In the Sangin case, the mission intelligence coordinator hesitated to bring the controversial whisper chat to the attention of the pilot. The coordinator told investigators the crew was trained not to address whisper chats “during a dynamic situation.” And yet, those chats were the only communications mechanism for the DGS-IN analysts.
The message from Otto and the investigators is that there is no panacea for preventing fratricide once a ground commander loses track of his troops.
“Absent this clarity, no unit or person involved in the strike process could have confidently intervened to prevent this tragedy,” the investigators said.
This story appeared in the March edition of C4ISR Journal.