The Problems with the ‘Gorgon Stare’ Surveillance System
After Secretary of Defense Gates criticized the Air Force in 2008 for providing insufficient support for the Army in combat, Gorgon Stare development appeared in DOD budget documents as a response. It is a "wide-area airborne surveillance system" typically described in the media as a "revolutionary" intelligence, surveillance, and reconnaissance (ISR) upgrade for finding (and prosecuting) targets in, for example, Afghanistan. (The Gorgon Stare moniker comes from Greek mythology: with their unblinking stare, Medusa and her two sisters could turn those who looked at them into stone.)
According to DOD testers, Gorgon Stare is ineffective ("not operationally effective") and unreliable ("not operationally suitable"). As described below, it cannot readily find and identify targets (especially human targets), and it cannot reliably locate what it sees. (Moving targets of any size at any location present a different problem.)
Gorgon Stare is developed by the Sierra Nevada Corp in conjunction with the USAF’s 645th Aeronautical Systems Group, also known as "Big Safari." According to now retired Lt. Gen. David Deptula, the Air Force’s former deputy chief of staff for ISR, it offers "many orders of magnitude" improvement over existing sensors now on drones in Afghanistan.
Consisting of two pods for the MQ-9 Reaper, Gorgon Stare provides imagery from five electro-optical (EO) cameras and four infrared (IR) cameras in one pod through a separate processor pod for daytime and nighttime operations. In Deptula’s words, "Instead of looking at a truck or a house, you can look at an entire village or a small city" with the multiple cameras, simultaneously.
The imagery is provided in three modes: "chip out" imagery to remote video terminals for troops in the field, "near real time," "sub view" imagery to dedicated ground stations for analysis and interpretation, and post-flight downloads for further "forensic" analysis.
Pods are described in the media to cost $17.5 million per set, not including the cost of the ground station equipment unique to the system. DOD budget data shows an $82 million dollar request for four "sensor suites" in 2009 and $426 million spent on it and closely related USAF R&D activities in FY 2009-2011. A request of $152 million was programmed in 2011 for 2012. These cost data, however, may be incomplete. There are also plans for an increment II with more EO and IR cameras and perhaps additional sensors, such as synthetic aperture radar. The cost-whatever it may actually be-will clearly grow.
(For additional details describing the system and for the source of the quotes of Gen. Deptula above, see "Gorgon Stare Broadens UAV Surveillance" by Richard Whittle from Aviation Week)
On January 2, the Washington Post gave Gorgon Stare nation-wide fame in an article proclaiming "we can see everything." The Post asserted that the only real limitation was the ability to "sift through huge quantities of imagery quickly enough to convey useful data to troops in the field."
The Post was quite wrong. There are major issues that go well beyond the "problem" of bringing human operators up to the level of performance of the technology. The Post didn’t seem to look very hard for the other side of the story on Gorgon Stare. It took me one e-mail and one phone call. The attached test report was the result.
The 53rd Wing of the Air Combat Command at Eglin Air Force Base was tasked to test Gorgon Stare in an "operational utility evaluation." In a draft of its test report, which I understand is fundamentally unchanged in its final form, the testers -
found 13 "Category 1" (i.e. serious) deficiencies;
stated Gorgon Stare is "not operationally effective," and it is "not operationally suitable." (See p. 2.) That’s a flunk in OT&E terms: it’s both ineffective and unreliable, and
recommended that Gorgon Stare not be "fielded" (deployed) to Afghanistan, or anywhere else. (See pp. 1 & 6.)
The more you read, the worse it gets:
The imagery from Gorgon Stare is frequently marginal to poor, depending on the mode of use. Specifically, when it is used for "near real time" field or ground station use, the electro-optical (EO) imagery "can find and track [objects as small as] vehicles" but not "dismounts" (people). (See p. 3.) For contemporaneous users in the field, the Infrared (IR) imagery is worse: it is "marginally sufficient to track vehicles" and "not sufficient to track dismounts." (See p. 4.) "In general, IR imagery quality is poor, which yields marginal mission capability at night." (P. 3.)
In fact, Gorgon Stare may be a step backwards. The multi-camera aspect of the design seems to have created problems. Some of the imagery is "subject to gaps between stitching areas [where the camera images meet], which manifests itself as a large black triangle moving throughout the image." (P. 3.) "Contrast differences between the four IR cameras degrade the ability to track targets across the image seams." (P. 4.) And, "dropped [image] frames . from a few seconds to several minutes-[make] it impossible to track moving targets over that period." (P. 4.)
Beyond the "seams" between images, the image quality is degraded from what users in the field have come to expect: "image quality does not support mission sets commonly used by RVT [remote video terminal] users" (p. 3). In plain English, the image quality is worse than that now provided by Predator and Reaper drones without GS.
There is a serious time delay problem. Transmissions to the ground, at the rate of two frames per second, arrives 12 to 18 seconds late for the "subview" ground station, and it arrives 2 seconds late to the "real time" users in the field. This "limits," if not eliminates, the ability to track and prosecute "dynamic" (i.e. moving) targets. Of course, when the target moves to the edge ("seam") of a camera frame, this problem becomes worse.
The better quality imagery that is obtained from the computer pod after flight takes too long to download "to conduct timely forensic analysis." (P. 4.)
There is another serious problem regarding the accuracy of location coordinates: "an unpredictable [i.e. random] software error generates a faulty coordinate grid" rendering location information "inaccurate and inconsistent." In other words, if Gorgon Stare is ever able to find and identify a target, it might generate a false location, rendering an attempt to attack it ineffective-and hitting an unintended location which may contain innocents or friendly forces. (See p. 3.) A tester unofficially remarked that means it cannot be used for sensor or weapons cueing, a primary reason for Gorgon Stare’s existence.
Limited bandwidth is the reason for the slow data receipt; a "work around" was established, but that reduces the quality of images even more.
Despite "full contractor logistics," Gorgon Stare performed poorly on measures such as "average failures per sortie; meantime between failures; and troubleshooting time following a failure." (P. 5.) Overall, one tester commented that it is "about 55 to 65 percent reliable."
In sum, the complexity of the system impedes both the imagery and the ability of operators on the ground to receive it – let alone figure it out. It cannot reliably find and track human targets; it has additional problems for moving targets, and the random location inaccuracy makes the system virtually unusable for prosecuting even stationary targets.
The testers ultimate view was to recommend against fielding Gorgon Stare until seven serious issues were fully addressed. (See p. 6.)
The program advocates ("Big Safari" at Air Force Material Command’s 645th Aeronautical Systems Group) claimed that the tests were unfair as they probed performance areas that were beyond the specifications for the system-just as operational testers are supposed to. Big Safari even protested that Gorgon Stare "was designed to operate in a different environment from which is currently envisioned-relatively flat earth with a greater number of vehicles." That would clearly exclude Afghanistan. And, therefore, they argue, it should be deployed immediately to Afghanistan!
Sadly, that logic may have won the day. With apparent JCS approval, Gorgon Stare will be deployed to Afghanistan without delay. The next unnatural act sure to occur is another article in a national newspaper heaping praise on this great boon to the war and the troops in the field.
The saddest irony (completely ignored in the press’ and technologists’ advocacy) is that even if Gorgon Stare were to be effective and reliable, it still would not help us win the war in Afghanistan, or anywhere else. As my friend and colleague Chuck Spinney argues at his website, Gorgon Stare does not enhance our fingerspitzengefühl (that is, the "fingertip feel" for the subtleties of any combat situation and the enemy that successful combat leaders have always found to be essential for success). In fact, the test evidence and Spinney’s explanation make it clear that Gorgon Stare impedes our acquiring fingerspitzengefühl. Find the informative explanation at http://chuckspinney.blogspot.com/.
Winslow Wheeler is director of the Center for Defense Information’s Straus Military Reform Project.