STRAY WEAPONS AND THE LOSS OF INNOCENTS

Pressures to avoid civilian casualties and unintended damage to nonmilitary structures were greater in Allied Force than in any previous campaign involving U.S. forces. Nevertheless, despite rules of engagement characterized by USAF Major General Charles Wald as being “as strict as I’ve seen in my 27 years in the military,” there were more than 30 reported instances throughout the air war of unintended damage caused by errant NATO munitions or mistakes in targeting, including a dozen highly publicized incidents in which civilians were accidentally killed.[306]The first serious loss of civilian lives occurred on April 12, when an electro‑optically guided AGM‑130 released by an F‑15E struck a targeted rail bridge over the Jusna Morava river in Kosovo on the Belgrade‑Skopje line 300 km southeast of Belgrade just as a passenger train full of noncombatants, in a tragic moment of fateful timing, happened to be crossing it.[307]Belgrade later reported that more than 55 civilians had been killed in that incident. Two days later, in the worst case of collateral damage to have occurred at any time throughout the operation, attacks against presumed enemy military vehicles at two sites in southwestern Kosovo near the town of Djakovica were said to have killed numerous ethnic Albanian refugees when USAF F‑16 pilots mistook civilian vehicles for a convoy.[308]

These and similar possible target identification errors resulted, in at least a few instances, from constraints imposed by the requirement that NATO aircrews remain above 15,000 ft to avoid the most lethal enemy infrared SAM and AAA threat envelopes, which made visual discrimination between military and civilian traffic difficult at best. Discriminate attacks against moving military vehicles amid a virtual sea of civilian refugees typically bordered on being an impossible mission when pilots orbiting at medium altitudes could not determine for sure whether a convoy consisted of military trucks, military vehicles carrying refugees, or civilian vehicles. General Wald, the deputy director of strategic planning on the Joint Staff and the commander at Aviano during Operation Deliberate Force in 1995, conceded that “the job is about as hard as it’s going to get for targeting.”[309]

Another contributing factor was the occasional tendency of allied aircrews to maneuver their aircraft in such a way as to put clouds within the targeting pod’s field of view between the aircraft and the target, thus blocking the laser beam illuminating the target and depriving the weapon of guidance. On April 6, near the end of the second week, the first LGB went astray in that manner, hitting an apartment building in the small town of Aleksinac 100 miles southeast of Belgrade and reportedly killing at least seven civilians and injuring dozens more. The intended target had been an artillery brigade headquarters, but the bomb’s steering toward its desired mean point of impact was disrupted by clouds that deflected the laser beam after weapon release.

In the case of the Djakovica incident noted above, there were initial reports that Yugoslav aircraft had intentionally attacked the civilian tractors and wagons near Prizren. Those reports ultimately proved groundless, although Pentagon officials did confirm that the Yugoslav air force was still operating low‑flying Galeb ground‑attack jets and attack helicopters.[310]In all events, the alleged occurrence of an inadvertent bombing attack on noncombatant civilians took place at midday, despite the greatest operational discipline on the part of the involved USAF pilots. The F‑16 strike force leader, who was operating as an airborne forward air controller (FAC‑A), determined the initial convoy to be made up of uniformly sized, colored, and spaced military vehicles whose occupants seemed engaged in systematic house‑burning. Extensive radio discussion then ensued between the FAC‑A and the ABCCC stressing the need to avoid inadvertently harming any Kosovar refugees. The ABCCC, backstopped by an orbiting UAV, confirmed the convoy to be a valid military target and marshaled as many fighters against it as were available in the immediate target area.

During the course of the precision attacks with 500‑lb GBU‑12 LGBs that then ensued, it was reported as “possible” that some of the vehicles may have been civilian tractors, at which point the FAC‑A immediately called all fighters off “high and dry” (clear of the target area with their armament switches deselected), and the ABCCC, in turn, requested reverification of the targets as hostile. At that point, nearby OA‑10s were called in so that their pilots might reconnoiter the situation and provide such reverification with onboard nine‑power space‑stabilized binoculars. One OA‑10 pilot reported observing definite military vehicles but also multicolored and possibly civilian vehicles, whereupon the FAC‑A terminated all further attacks. Afterward, Serb news reports claimed that 80 civilians had been killed, although the persistent ambiguities were such that NATO only conceded that it “may have attacked” civilian vehicles. Some reports suggested that the civilians involved had been machine‑gunned rather than bombed, and eyewitnesses on the ground reported the use of human shields in the convoys and nearby Serb mortar fire at the same time the convoy was being attacked by the F‑16s. The commander of the 31st Air Expeditionary Wing whose F‑16s were involved in the tragedy, Brigadier General Leaf, later told reporters that the incident involved “a very complicated scenario, and we will never be able to establish all the details.” He further stated that he could not explain the bodies of the civilians that had been shown on Serbian television and conceded only, in light of the ambiguous evidence, that there “may have been” unintended civilian fatalities.[311]

The extraordinary media attention that was given to events like these attested to what can happen when incurring zero noncombatant casualties becomes not just the goal of strategy but also the expectation. Thanks to unrealistic efforts to treat the normal friction of war as avoidable human error, every occurrence of unintended collateral damage became overinflated as front‑page news and treated as a blemish on air power’s presumed ability to be consistently precise. Indeed, the added constraints imposed on NATO aircrews as a result of such occasional tragic occurrences indicated the degree to which modern air power has become a victim of its own success. During the Gulf War, cockpit video images of LGBs homing with seemingly unerring accuracy down the air shafts of enemy bunkers were spellbinding to most observers. Yet because of that same seemingly unerring accuracy, such performance has since come to be expected by both political leaders and the public alike. Once zero collateral damage becomes accepted as a measure of strategy success, not only air power but all forms of force employment get set up to be judged by all but unreachably high standards. Inevitably, any collateral damage then caused during the course of a campaign becomes grist for domestic critics and the enemy’s propaganda mill. Anthony Cordesman rightly noted how characterizations of modern precision bombing as “surgical” overlook the fact that patients still die on the operating table from time to time.[312]Nevertheless, a nontrivial number of proposed sorties in Operation Allied Force were either canceled outright or aborted at the last minute before any weapons were released because their targets (wryly characterized by some USAFE staffers as “morally hardened”) could not be positively identified or because of the perceived risk of causing collateral damage. At best, that made for a necessarily constrained and therefore inescapably inefficient air operation compared to the standard set earlier in Desert Storm.

A bevy of criticism arose from some quarters after the bombing ended alleging that many of the 500 or more Yugoslav and Kosovar Albanian civilians who lost their lives to collateral damage incidents had died needlessly as a direct result of NATO attack aircraft having been kept above 15,000 ft in the interest of minimizing the likelihood of losing friendly lives. Critics further charged that operating at that altitude had somehow been risk‑free, cowardly, and even immoral on the part of NATO’s aircrews.[313]In league with other detractors of the way the air war was conducted, strategist Edward Luttwak, for example, characterized 15,000 ft as a “not‑optimal” but “ultra‑safe” altitude from which allied pilots might carry out “perfectly safe bombing.”[314]

In point of fact, 15,000–20,000 ft was precisely the “optimal” altitude block from which to conduct LGB attacks–not only to keep the attacking aircraft clear of short‑range air defenses in the immediate target area but, more important, to give the LGB time to acquire the target and assume a stabilized glide. Contrary to the suggestions of critics, operating at medium altitude provides no protection whatever against radar‑guided SAMs. It merely puts attacking aircraft outside the lethal envelope of “trash fire” threats (small arms, AAA, and infrared SAMs). These threats are impossible to detect in a timely way and offer little or no warning of imminent danger; as a result, they cannot be countered very effectively. Indeed, operating at medium altitude actually increases the risk of being engaged by unnegated enemy radar‑guided SAMs because the aircraft can no longer take advantage of terrain‑masking opportunities. The more important point, however, is that when medium‑altitude attack tactics are employed, the timeline for target acquisition and weapons guidance is substantially longer, thus improving the chance of achieving a hit.

Even assuming the absence of undetectable “trash fire” threats, it is by no means a foregone conclusion that had allied aircrews routinely descended to lower altitudes in an effort to identify their targets more positively, the incidence of unintended collateral damage occurrences would have been that much lower. To begin with, VJ and MUP troops were highly accomplished at camouflage and hiding, and they made frequent use of the civilian populace as human shields. Moreover, in Kosovo, where most of the inadvertent civilian fatalities occurred, the mandated altitude floor was not invariably 15,000 ft as the critics implied. On the contrary, once operations against dispersed and hidden VJ forces in Kosovo began in earnest in mid‑April, FACs were cleared down to 5,000 ft as necessary to make positive target identifications, and strike aircraft could descend to as low as 8,000 ft for a nonprecision dive‑bomb delivery.

Even at those lower altitudes, however, positive identifications tended to be difficult, although in one case, as noted above, USAF OA‑10 pilots using nine‑power space‑stabilized binoculars managed to observe civilians intermingled with a VJ truck convoy after one vehicle had already been hit, as a result of which the ongoing attack was instantly terminated.[315]As a rule, however, routinely going lower and accepting the increased risk of losing an aircraft in the hope of doing better target discrimination would not, in all likelihood, have produced the desired result. True enough, flying even as low as 100 ft above ground level might have enabled NATO pilots to distinguish civilian from military traffic in a few fleeting moments, if that traffic happened to lie almost directly underneath the aircraft’s flight path. Yet operating that close to the ground at normal fighter airspeeds (500 nautical miles per hour or more) in defended airspace would have offered zero perspective and zero precision‑attack capability. It also would have increased the chance of NATO aircraft losses to enemy “trash fire” and just possibly brought about the overall failure rather than success of Allied Force as a result. Moreover, hidden targets would still have remained hidden.

The point of the foregoing is that for the kinds of circumstances that repeatedly occasioned the accidental loss of civilian life in Allied Force, the United States, to say nothing of its NATO allies, has yet to develop fail‑safe target discrimination capabilities and tactics for use either above or below 15,000 ft. As a result, it has had little choice but to rely on draconian rules of engagement (ROE), which are designed to hedge on the side of caution yet are anything but foolproof. In one case during Allied Force in which the ROE worked as intended, a USAF pilot was directed not to attack a confirmed SA‑6 launcher because it was parked immediately adjacent to a civilian structure in a village. There were other reported instances in which precision munitions in the process of guiding were deliberately steered away from targets at the last minute to avoid harming civilians who had not been seen in the target area until after weapon release.[316]In the most egregious instance in which the ROE regime appears to have failed, however, namely, the tragedy involving the convoy along the Djakovica road in Kosovo, the FAC who was coordinating the attack had been given a positive identification by the ABCCC that was completely consistent with the prevailing ROE. Upon observing that the vehicles were uniformly colored and evenly spaced, the FAC declared the convoy to be a valid target. He had also been given ABCCC approval to clear the fighters under his control to drop at will after one F‑16 orbiting overhead had drawn fire from one of the convoy’s vehicles.[317]

The solution to such challenges lies not in more relaxed operating restrictions but rather in the development of better tactics, techniques, procedures, and equipment–perhaps beginning with a more aggressive and effective use of offboard platforms like UAVs to perform combat identification and to provide cueing for engaged shooters.[318]Unfortunately, the sensor‑to‑shooter links that have been refined to a high art over the years for the air‑to‑air arena, such as the E‑3 AWACS and the joint tactical information distribution system (JTIDS) carried by some F‑15s, remain far less developed for ground‑attack operations when it comes to situation characterization and target identification.[319]In the absence of such capabilities, flying lower in Allied Force not only would not have solved the target identification problem, it would also have rendered weapons deliveries less accurate and, as a result, probably compounded the collateral damage problem rather than ameliorating it. As matters stood, although regrettable tragedies did occur because of occasional misdirected weapons, the munitions and tactics used by NATO in Operation Allied Force made the air effort a record‑setter when it came to achieving its declared goals with a minimum of collateral damage for an operation of that magnitude. Indeed, given the high volume of ordnance that was expended over the course of the 78‑day air war, it is most remarkable–even astonishing–that the incidence of unintended civilian fatalities was not higher.