The Alaska Test Bed Fallacy: Missile Defense Deployment Goes Stealth

Lisbeth Gronlund and David Wright

The Bush administration has proposed building a set of missile defense facilities in Alaska by 2004, including five silos for ground-based interceptor missiles at Fort Greely, and in early September Congress will begin considering whether to fund them. According to the Ballistic Missile Defense Organization (BMDO), these facilities are needed to allow more realistic testing of the ballistic missile defense system currently under development. BMDO has also stated that the so-called Alaska test bed could be pressed into use as an “emergency defense” should the United States be attacked by a small number of long-range missiles from northeast Asia before a fully developed and tested system had been deployed.

The Anti-Ballistic Missile (ABM) Treaty allows the United States and Russia two test facilities each for launching interceptor missiles, and according to a 1978 agreed statement to the treaty, either the United States or Russia can add a new test facility simply by declaring the site within 30 days of starting construction. The new site must be “consistent with the objectives and provisions of the Treaty,” and the total number of interceptor silos at all test ranges must not exceed 15. Thus, the United States could legally build the new test range as long as the facilities and activities were intended for testing and could not be used for a defense of the United States or to provide the base for such a defense.

However, careful analysis shows that some of the key facilities being planned—particularly the silos at Fort Greely—would have no utility in a testing program but rather appear designed specifically as a near-term deployment of a rudimentary missile defense system. As such, these facilities would violate the ABM Treaty, and Russia has stated that it would consider the Fort Greely site to be a violation.

The administration insists that its proposed facilities are part of a testing, not a deployment, program; and it is seeking funding for all these facilities under the research, development, testing, and evaluation (RDT&E) category of the 2002 Pentagon budget. At the same time, President George W. Bush has indicated that the United States will withdraw from the ABM Treaty “at a time convenient to America,” and the Pentagon has already begun clearing trees at Fort Greely with plans to begin silo construction next spring or summer. If the facilities constitute an initial deployment capability, the United States would have to provide the required six months’ notice of withdrawal sometime this fall or winter.

Unfortunately, the disingenuousness of this strategy is compounded by the fact that the immaturity of the interceptor technology and the lack of an ABM radar in Alaska means that any emergency deployment would not be effective against an actual attack. In short, if the United States withdraws from the ABM Treaty in order to deploy a contingency system in Alaska, it will incur the political costs of withdrawing from the treaty while reaping none of the potential security benefits from a working defense.

Existing Flight-Test Facilities

The only site where the United States currently launches interceptors against long-range target missiles is the Kwajalein Test Range, located on Kwajalein Atoll in the Marshall Islands.1 In addition to an interceptor launch site, this facility includes a prototype X-band radar, which is a one-third-scale version of the battle-management radar that is the key sensor to be used in the missile defense system. The facility also includes $4 billion of advanced sensors and computers that are used to observe and collect information on the tests.

In the four intercept tests of the midcourse system conducted to date, a target missile carrying a mock warhead was fired toward the Kwajalein test facility from Vandenberg Air Force Base in California, some 7,500 kilometers away. The missile then released the “target cluster,” which included the mock warhead and a spherical balloon decoy, along with the bus that carried them, and any debris generated in this process.

The initial notification of the target launch was provided by U.S. early-warning satellites. In an operational system, data from the satellites would be provided to an early-warning radar, which would track the missile after its boost phase. However, there is currently no early-warning radar that can participate in the tests, so in the most recent tests, the warhead carried a beacon that allowed a C-band radar on Hawaii to track it. The information from this radar was used to determine the rough location where the intercept would occur and then to launch the interceptor toward that location.2 The information from the C-band radar was also used to cue the X-band radar on Kwajalein, which then tracked the objects in the target cluster, refining the estimated intercept location and sending that information to the interceptor so that it could adjust its course.3

Once the kill vehicle was released by the interceptor booster and was close enough to the target cluster, the kill vehicle attempted to acquire the mock warhead with its on-board sensors and then maneuver to collide with it. In the tests conducted so far, the appearance of the warhead has been significantly different from the other objects and the defense sensors have been told what to look for, so the ability of the defense to realistically discriminate the warhead from other objects in the target cluster has not been tested.

All four intercept tests that have been conducted have used the same intercept geometry—that is, the trajectories of the target and interceptor missiles and the planned intercept point have been the same in each test. In all these tests, a two-stage surrogate booster has been used in place of a three-stage booster that is under development and that will be used if the system is actually deployed. As a result, the speed of the kill vehicle has been much lower than would be expected in the operational system, so the range of the kill vehicle has been very short and the closing speed has been low. Moreover, these intercepts were designed to occur at a relatively low altitude—roughly 230 kilometers—to minimize the spread of debris, both on the ground and in orbit.

The Proposed Test Bed

BMDO’s planned “2004 RDT&E test bed” includes the following new facilities and upgrades, funding for which is in the Pentagon’s proposed budget for the coming fiscal year:

  • Five ground-based interceptors (GBIs) at Fort Greely, in eastern Alaska. The five interceptor silos would be constructed in spring or summer 2002. The interceptors would use the so-called Payload Launch Vehicle Plus as a booster, which uses the three stages of a Minuteman II missile.4 The kill vehicle deployed on the interceptors would be the current prototype used in the tests.
  • Upgrades to the existing Cobra Dane radar on Shemya Island at the far end of the Aleutian Islands in Alaska “to provide a surrogate for planned UEWR [upgraded early-warning radar] capability.”5
  • Two ground-based interceptor launch silos at the commercial Kodiak Launch Complex on Kodiak Island, Alaska. BMDO proposes to construct the silos in spring or summer 2003, assuming the favorable completion of the environmental impact statement required as part of the National Environmental Policy Act process.
  • Upgraded software for the existing early-warning radar at Beale Air Force Base in Marysville, California (north of Sacramento). According to BMDO, this would permit the Beale radar to be used during intercept tests instead of the C-band radar on Hawaii before a target launched from Vandenberg comes within range of the X-band radar on Kwajalein.
  • Battle Management Command and Control (BMC2) nodes at four sites: Fort Greely, Kwajalein, the Joint National Training Facility in Colorado, and Colorado Springs.
  • A total of seven In-Flight Interceptor Communication System (IFICS) data terminals at four sites: two at Fort Greely, two at Eareckson Air Station on Shemya Island, two at Kwajalein Missile Range, and one at the Kodiak Launch Facility. The IFICS terminals would be used to communicate between the BMC2 and the interceptors while they are in flight.
  • Fiber or satellite communications between all the facilities.
  • A second launch silo at both the Kwajalein and Vandenberg sites.
  • Long-range air-launched target missiles, which the United States is developing the ability to launch from a cargo plane.

In addition, BMDO plans to build an X-band radar in the mid-Pacific, possibly in Hawaii, by approximately 2006. Funding for this project is not in the fiscal year 2002 request, and this radar would not be part of the 2004 test bed.

It is also worth noting that two facilities that might have been built for or incorporated into this test bed have not been included in BMDO’s plans. First, under the Clinton administration, the United States was planning to build an X-band radar on Shemya Island as part of the initial deployment of the national missile defense (NMD) system. Plans for building this radar have been shelved—at least temporarily—by the Bush administration. Second, there is an early-warning radar at Clear Air Force Base in central Alaska that was reported in March 2001 to be operational after having been upgraded, but it has not been discussed as part of the test bed. It is unclear why these sensors have not been included in discussions of current or future plans.

More Realistic Testing?

BMDO officials have stated that the 2004 test bed was proposed to respond to criticisms of the testing program made by Philip Coyle, the Pentagon’s former director of operational test and evaluation, and others. In his August 2000 evaluation of the NMD test program, Coyle raised concerns about the missile defense test program and recommended a number of steps that should be taken to make the tests more challenging and realistic6

  • incorporate more realistic countermeasures than are currently planned;
  • include tests in which the defense has limited a priori knowledge of the target complex, target trajectory, and time of launch;
  • include tests involving multiple interceptors launched against multiple targets;
  • include more variation in the engagement conditions, including the time of day, the weather conditions,7 the intercept geometry, the interceptor flyout range, the altitude of intercept, and the closing speed between the kill vehicle and target.

Coyle also wrote that the operational realism of the current tests is limited by the location of the radars being used relative to the current missile trajectories. There are two problems. First, there is no early-warning radar that can observe the target early in its trajectory as it moves toward the radar. Second, the X-band radar being used is not located as far away from the interceptors as it would be in some operational scenarios. Instead, the X-band radar at Kwajalein is co-located with the interceptor missile.

Clearly, the Alaska facilities would not address Coyle’s first, and most fundamental, criticisms of the current testing program: the lack of realistic countermeasures and the information given to the defense in advance of a test. Even if BMDO is able to develop the ability to conduct reliable hit-to-kill on the test range, it must show that the system can work in realistic tests against countermeasures of the type that would be expected in a real attack, without the defense having full a priori information about the target suite. The realism of the tests can be increased in this way with the current test facilities. Moreover, tests that vary the time of day or include heavy rain and clouds can all be conducted by launching interceptors from Kwajalein and target missiles from Vandenberg. The proposed facilities in Alaska are not needed for any of these purposes.

The one proposed addition that would be genuinely useful to the test program would be the second launch silos at Kwajalein and Vandenberg because they would permit multiple interceptors to be launched against multiple targets, thereby addressing Coyle’s third point.

What about the remaining issues? Would the proposed test bed allow a variation in the other engagement conditions listed above—the intercept geometry and altitude, the interceptor flyout range, and the closing speed—that would not be attainable with existing facilities? Would the test bed allow a test configuration where the radars are more realistically located with respect to the target and interceptor missile trajectories? BMDO officials have not publicly specified how the proposed test bed would address these specific concerns. BMDO briefing slides simply state that the proposed facilities would add “intercept areas” and assert that this would increase realism of the testing program. Below we examine the role that the major individual elements of the test bed—the five interceptor silos at Fort Greely, the upgrades to the Cobra Dane radar, the IFICS data terminals, and the two interceptor launch silos at Kodiak—would play in an improved flight-test program.

Fort Greely

Although the five proposed interceptors and silos at Fort Greely are included as a major part of the proposed 2004 test bed, these silos would not be used to launch either interceptor or target missiles. Thus, these facilities would not allow a variation in engagement parameters or a more realistic radar location and therefore would not enhance the test program.

For safety reasons, the United States does not launch long-range missiles from an inland site. Indeed, BMDO has stated that test launches cannot be conducted from Fort Greely because it is too near populated areas. According to Senate testimony given July 31 by Patricia Sanders, deputy director of BMDO for test, simulation, and evaluation, “At present, BMDO does not intend to launch any GBI from Fort Greely during the testing process because these missiles would fly over land in violation of current flight-test safety restrictions.”

Moreover, BMDO apparently plans to build the five silos on the basis of an environmental impact statement completed by the Clinton administration in anticipation of a decision to begin deployment of its national missile defense. That statement considered the deployment of up to 100 interceptor silos at one of five potential sites, with Fort Greely identified as the “preferred alternative.”8 The final NMD environmental impact statement says, “Launches would occur only in defense of the United States from a ballistic missile attack. There would be no flight testing of the missiles at the NMD deployment site.”9

Thus, building missile silos or storing interceptor missiles at Fort Greely would in no way be useful to a flight-test program.

In response to questions about the testing utility of building silos that could not be used to test launch interceptors, BMDO has stated that it might try to change the regulations so interceptors could be launched from Fort Greely. There is no compelling reason, however, to launch interceptor missiles from Fort Greely rather than Kodiak for a research and development (R&D) test program.

BMDO also argues that building these facilities at Fort Greely would be useful—even without flight testing—because doing so would allow it to conduct other kinds of testing. According to Sanders’ testimony, “The GBIs at Fort Greely will allow BMDO to prove out the design and siting of a GBI field that would be required to fire in a salvo without having the GBI interfering with each other GBI, to test the communication between all component parts, and to test for fuels degradation in the arctic environment, as well as to develop and rehearse maintenance and upkeep processes and procedures.”

These activities have no place in an R&D testing program for a system about which no deployment decision has been made; rather they are the kinds of activities needed to get a deployment site up and running.

Even if such assessments were needed or justified at this stage in the testing process, BMDO would not need to build five interceptor silos to test how fuel degrades in the arctic or to develop maintenance procedures. In fact, if there is a potential problem with fuel degradation at Fort Greely, it would be premature and potentially wasteful to build five silos there before the issue was adequately understood.

Finally, the argument that the United States needs to build five silos at Fort Greely to “prove out” the design and siting of an interceptor site where GBIs fired in a salvo would not interfere with each other makes no sense. BMDO is not proposing to test launch five GBIs in a salvo to demonstrate that they do not interfere with each other, but rather to prove that it can build a site with five silos of the proper design. If the United States knows how to design and build silos so missiles can be salvo-launched, there is no need to prove this capability at Fort Greely as part of an R&D program. If, on the other hand, the United States does not know how to design and build such silo sites, then building the silos at Fort Greely is not the way to demonstrate this capability because the interceptors could not be fired from these silos. Moreover, BMDO is already planning to build multiple interceptor silos at Kwajalein and Kodiak to test the system with simultaneous interceptor launches.

For all of these reasons, Congress should treat a decision on whether to fund the proposed construction at Fort Greely as a deployment issue and not a testing issue.

Cobra Dane Radar

The strongest motivation for upgrading the Cobra Dane radar appears to be to have a radar on Shemya Island that could be used as part of an emergency deployment system, to substitute for the X-band radar that has been shelved.

This radar has a frequency between those of the early-warning radars and the X-band radar. It was placed on Shemya in the mid-1970s to observe Soviet missile tests and has a fixed orientation appropriate to this mission, looking in a fan northwest from Shemya over Russia. As a result, it cannot observe any missiles in the Kodiak-Kwajalein-Vandenberg test area. (See Figure 1.) It could, however, observe missiles on trajectories from North Korea to many parts of the United States.

At a July 11 Pentagon briefing, BMDO officials stated that this radar is “in the right orientation for the test of air-launched targets.” But they have not stated why this orientation is desirable or what the purpose of such tests would be.

The only conceivable advantage of using Cobra Dane as an early-warning radar in tests appears to be that the radar would be in a different position relative to the target trajectory than is possible in current tests. Air-launched target missiles could be launched from the Bering Sea toward the Cobra Dane radar, whereas target missiles launched from Vandenberg fly away from the early-warning radar at Beale, California. (The C-band radar currently substitutes for the early-warning radar, but the Pentagon intends to eventually use the Beale radar because it would make the tests somewhat more realistic.)

Using Cobra Dane in this way would not correspond to a realistic operational configuration. The location of the radar places several serious limitations on its use in tests against air-launched target missiles. First, using the Beale radar to observe target launches from Vandenberg has been criticized because the radar would see the missile during boost phase and therefore would get a large signal pinpointing the location of the warhead that it would not receive in a real attack. But using Cobra Dane with air-launched targets would have the same limitation. Because of the proximity of Cobra Dane to Russia’s Kamchatka Peninsula, test missiles could only be launched from a very restricted area around the Bering Sea, at a distance of about 800 kilometers or less from the radar. As a result, the boost phase of the missile would again be visible to the radar and would give a large signal that would not be present in a real attack. In a real attack, the early-warning radars would search to detect the target cluster based on data provided by US early-warning satellites.

Second, and again because of its location, the Cobra Dane radar would only be able to track such a target for a short time. An early-warning radar must track a missile after boost phase to determine its trajectory once it is no longer under powered flight. Because a long-range missile booster will travel several hundred kilometers before it burns out, the radar would only be able to track the warhead for roughly a minute after boost phase before it passed out of the radar’s field of view. The accuracy with which a radar can determine an object’s trajectory depends on the length of time it tracks the object, and in an operational situation one would typically expect a much longer tracking time.

Thus, using the Cobra Dane radar in tests against air-launched missiles would add little realism to the testing program. Using Cobra Dane in such a flight test might serve mainly to test the performance of the upgraded radar itself, in preparation for its use as part of an emergency deployment system.

IFICS Data Terminals

The IFICS data terminals are a component of the battle management and command and control system, and would be used to communicate between the BMC2 and the interceptors while they are in flight. According to the environmental impact statement prepared for the Clinton administration’s planned national missile defense, a deployed system would require two data terminals at each site hosting IFICS “to meet NMD reliability requirements,” and four or more data terminals should be located with the interceptors “to meet reliability and communication requirements.”

For a deployed system that is required to work at all times, it is standard practice to build redundant components to compensate for the potential unreliability of the individual components. However, this issue of reliability is much less of a concern for components intended to be used for R&D testing, in which the time of a test is known well in advance and a test can be postponed if a component does not check out before the test. The current plans for the Alaska test bed call for building one IFICS terminal at Kodiak, and two each at Fort Greely, Eareckson Air Station on Shemya Island, and Kwajalein missile range. This raises the question of whether a second IFICS terminal at the three sites is needed for the test program or is only justified by the “emergency defense” rationale.

The Kodiak Launch Facility

Of the major elements of the proposed test bed, the Kodiak launch facility is the only one that addresses any of the concerns raised by Coyle. But even the Kodiak facility is not needed to address the most important testing limitations listed above, and its main utility appears to be in one particular scenario.

Radar locations. Launching interceptors from Kodiak would not allow for more realistic radar locations in the tests. The only X-band radar that would be part of the 2004 test bed is the prototype radar at Kwajalein and that radar would be in a less realistic location with respect to interceptors launched from Kodiak than in the current test configuration with interceptors launched from Kwajalein. Launching interceptors from Kodiak would not significantly change the relative location of the early-warning radar, because target missiles would still be launched from Vandenberg.

Similarly, the radar location is not improved if one launches target missiles from Kodiak with interceptors launched from Kwajalein.

As noted above, the United States may eventually build an X-band radar on Hawaii or somewhere else in the mid-Pacific, but it is not planned as part of the 2004 test bed. A radar on Hawaii would give a more realistic radar location for tests carried out with the existing test site since it would be forward-deployed relative to interceptors launched from Kwajalein against target missiles launched from Vandenberg. It would not help with the radar location for launches from Kodiak.

Interceptor flyout range. As noted above, the trajectories of the target and interceptor missiles and the planned intercept point have been the same in each of the four intercept tests conducted so far. In all these tests, a two-stage surrogate booster has been used in place of the planned three-stage interceptor booster. Consequently, the kill vehicle has been much slower than would be expected in the operational system, and the distance it traveled to the intercept location has been very short, about 700 kilometers. It is important to conduct tests with a variety of interceptor flyout ranges, including long interceptor ranges.

Once the prototype booster is available, the booster itself will no longer impose constraints on the interceptor flyout range. Two additional factors, however, will constrain the interceptor range.

First, to conduct realistic tests of the midcourse system, it is important that the flight tests include an X-band radar. These X-band radars are intended to be an integral part of the midcourse defense system: they would be used to track incoming warheads and are a key sensor in BMDO’s plans to discriminate warheads from decoys and debris and to conduct kill assessment. Thus, a realistic test should be conducted so that the target is in the field of view of the X-band radar at least until the kill vehicle begins to acquire the target suite with its onboard sensors. Preferably, the intercept itself should take place within the field of view of the X-band radar, so the radar’s ability to make the kill assessment can be tested.

Second, BMDO wants intercepts to take place at low altitudes to minimize the spread of debris in space and on the ground. To minimize the debris problem, BMDO has said that it intends to conduct most of its tests so that the intercept takes place at altitudes of less than 320 kilometers, allowing an X-band radar to observe these intercepts out to a ground distance of roughly 1,800 kilometers. The fact that the Earth is curved means that a radar could observe intercepts at a greater distance only if they occurred at higher altitudes. (See Figure 2.)

The only X-band radar that would be part of the 2004 test bed is the prototype at Kwajalein, which was built to provide a prototype component for missile defense testing. Thus, if we assume that the intercept itself takes place within the field of view of the X-band radar on Kwajalein, the intercepts would be restricted to within roughly 1,800 kilometers of the atoll. The interceptor flyout range for interceptors launched from Kwajalein is not yet bumping up against this constraint and could be increased to roughly 1,800 kilometers from the current 700 kilometers.

Achieving intercepts with longer interceptor flyouts from Kwajalein would require either conducting the intercepts at higher altitudes to allow them to be observed by the X-band radar at Kwajalein, or conducting intercepts outside the field of view of the X-band radar (and other sensors at Kwajalein) and instead using a surrogate radar. The latter option not only reduces the realism of the test but also does not allow the sensors at Kwajalein to collect data on the intercept.

Because some of the potential intercept locations that BMDO has shown on briefing slides lie considerably farther than 1,800 kilometers from Kwajalein, then BMDO must be planning either to intercept at higher altitudes or to use a surrogate for the X-band radar. Once it is planning to do so, however, it can conduct tests with long interceptor flyouts using the interceptors launched from Kwajalein.

Launching interceptors from Kodiak would be useful for one specific intercept geometry suggested by Coyle: one with a long interceptor flyout but a low closing speed between the target and interceptor (so that the interceptor was approaching the target somewhat from behind it).10 It appears that if a target were launched from Vandenberg and the interceptor from Kodiak, this intercept geometry could be achieved while having the intercept occur near enough to Kwajalein that it would be within the field of view of the Kwajalein radars.

Intercept geometry. As noted above, in all four intercept tests that have been conducted, the trajectories of the target and interceptor missiles and the planned intercept point have been the same. As part of the test program, it would be useful to vary the intercept geometry so that the warhead and interceptor approach each other from varying directions.

The United States could achieve very different intercept geometries simply by having the interceptor fly out from Kwajalein in different directions and launching the target missile from Vandenberg to a different point within 1,800 kilometers of Kwajalein. (See Figure 3.) Although there would be safety restrictions on where such intercepts could occur, this would still allow tests with a variety of intercept geometries.

Furthermore, if BMDO’s plan to use air-launched target missiles proves practical, it would allow additional intercept geometries using interceptors launched from Kwajalein, beyond those already available by launching targets from Vandenberg. Launching interceptors or target missiles from Kodiak would not appear to add any variations in geometry beyond what could be achieved by launching interceptors from Kwajalein and targets from airplanes.

Intercept altitude. As discussed above, to restrict the spread of debris, BMDO plans to conduct most intercepts below roughly 320 kilometers. However, realistic testing would eventually require tests involving intercepts at considerably higher altitudes. In operational use, the defense would not wait until the end of the warhead’s trajectory unless it had to but rather would try to intercept the warhead in the middle of its trajectory, when it was at an altitude of 1,000 kilometers or more.

Launching interceptors from Alaska rather than from Kwajalein would provide no advantage to testing intercepts at high altitudes. Indeed, as noted above, intercepting at higher altitudes would permit longer interceptor flyout ranges from Kwajalein while still allowing the intercept to be observed by sensors at Kwajalein.

Closing speed at intercept. In past intercept tests, the closing speed at intercept has been low compared to what could realistically be expected: 7.4 kilometers per second, compared to roughly 12 kilometers per second for the operational system with the same intercept geometry. The barrier to achieving higher and more realistic closing speeds is not the interceptor launch location but the burnout speed of the booster. The previous tests have used a surrogate interceptor booster that reached a speed of about 2.2 kilometers per second; the booster currently under development for the midcourse system is intended to accelerate the interceptor to 7 or 8 kilometers per second. The closing speed could also be increased by having the target and interceptor meet more nearly head-on than in past tests. Launching interceptors from Kodiak is not needed to address either of these issues.

An “Emergency Defense”

The other argument for a crash program to build the proposed missile defense facilities in Alaska is to create a near-term “emergency defense” that might offer some capability in the event of a missile attack on the United States. However, such an “emergency defense” system would have very limited utility for two reasons.

First, the United States will have little basis by 2004 to be confident in the performance of the midcourse interceptors deployed at Fort Greely. The interceptors will almost certainly not have begun operational testing by that time—under the Clinton schedule, operational testing was to begin in FY 2005, but this schedule has already slipped by more than a year.

Second, the proposed system would provide very little protection should an attack from North Korea occur. Even if the interceptor and kill vehicle technology worked perfectly by that time, the system performance would be severely limited by its sensors. BMDO has stated that the X-band radar is a key sensor needed to discriminate the warhead from other objects, including debris or simple decoys, but there would be no such radar that could see trajectories from North Korea fired toward the United States.

According to Senate testimony by Deputy Secretary of Defense Paul Wolfowitz on July 17, an “upgraded Cobra Dane radar will provide enhanced early warning and may have some ABM radar capability.” Wolfowitz noted, however, “In any operational system, we anticipate that the X-band radar at Shemya would be required to provide needed discrimination, even with all possible upgrades to Cobra Dane.”


This analysis leads to several important conclusions. First, building the facilities at Fort Greely cannot be justified as part of the proposed 2004 test bed and should be debated as the deployment issue it is. Moreover, building these facilities would be a clear violation of the ABM Treaty and is opposed by Russia. Because the emergency deployment system would in reality have little capability against an enemy attack, the result of proceeding with construction at Fort Greely would be that the United States would incur large political costs up front in exchange for little actual benefit.

Second, the most important measures needed to address shortcomings in the existing testing program and to make the tests more realistic can be implemented using the current test range, especially once additional silos are built at Vandenberg and Kwajalein. Adding a launch site for interceptors and targets at Kodiak would allow some additional testing capability, but it is not the highest priority and does not appear to be time urgent. However, adding a testing facility at Kodiak need not lead to large political costs because the United States could declare it to be an allowed test site under the ABM Treaty.

Third, a key consideration in conducting realistic intercept tests is that the tests include an X-band radar that is used for discrimination and to guide the interceptor. Because the only X-band radar that would be included in the 2004 test bed would be the prototype at Kwajalein, this constrains what test geometries would be useful. In addition, conducting tests within sight of the sensor complex at Kwajalein has the very significant advantage that it allows the United States to collect information that makes the tests as useful as possible.

Fourth, upgrading the Cobra Dane radar adds very little to the test program. The primary reason for upgrading Cobra Dane appears to be to use it as part of an emergency deployment system.

Fifth, the Bush administration has not laid out a clear or convincing rationale for the new facilities it is requesting. It implies that having several sites to launch interceptor missiles and target missiles would greatly increase its testing capabilities and allow it to conduct a more realistic testing program, but has not explained how it would address the specific testing shortcomings that have been raised. Congress should require BMDO to articulate more clearly what tests it plans to do using the test bed facilities and what capabilities it would gain that it could not achieve in other ways.

Finally, this analysis sheds some light on the Bush timetable for withdrawal from the ABM Treaty. The Bush administration appears eager to give its six months’ notice of intent to withdraw from the treaty this fall so it can legally begin building five interceptor silos at Fort Greely next spring. But, as we have shown above, the silos would have no role in a flight-test program. President Bush wants to pull out of the treaty to start building an unproven missile defense system, not to test it.

Figure1: The proposed test bed, showing the coverage of the Cobra Dane radar on Shemya.

Figure 2: The variation of radar horizon with distance from the radar.

Figure 3: The angle and speed at which the kill vehicle intercepts the target warhead can be varied simply by using different intercept locations around Kwajalein.Three different intercept geometries using a single target launch site and a single interceptor launch site are illustrated here.

1. This site was recently renamed the Ronald Reagan Ballistic Missile Defense Test Site, or Reagan Test Site (RTS) for short.
2. Press briefing by Major General Willie Nance, August 9, 2001. The use of the C-band radar has been controversial because the limited capability of the radar requires the transponder “beacon” to be attached to the warhead to amplify the signal and send it back to the radar. Although apparently intended to compensate for the limitations of the C-band radar, questions remain about the quality of the information from this radar and how it is used.
3. Radars are typically classified by the frequency of the radar waves they use. US early-warning radars operate in the ultrahigh frequency (UHF) band and have a corresponding wavelength of roughly 70 centimeters. A C-band radar has a wavelength of roughly 5 centimeters, and an X-band radar has a wavelength of roughly 3 centimeters. The smaller the wavelength of the radar waves, the greater the accuracy with which the radar can determine the location of an object, and the greater its ability to detect details on an object and thus to distinguish between different objects. That is why an X-band radar would be used for the demanding task of guiding the kill vehicle and attempting to discriminate between the warhead and decoys.
4. This is neither the surrogate two-stage booster used in the tests so far nor the three-stage prototype booster under development.
5. “Midcourse Defense Segment PE 0603882C” BMDO FY02 budget description, p. 4. The full BMDO budget request can be found at
6. Director, Operational Test and Evaluation, Report in Support of National Missile Defense Deployment Readiness Review, August 10, 2001. The report is available at
7. Because the performance of an X-band radar is degraded by heavy rain and dense clouds, Coyle has advocated testing under such weather conditions.
8. The other four sites are Clear Air Force Base, Alaska; the Yukon Training Area (Fort Wainwright)/Eielson AFB, Alaska; Grand Forks AFB, North Dakota; and the Stanley R. Mickelsen Safeguard Complex, North Dakota.
9. Executive summary, p. 6. The entire statement is available at
10. See Report in Support of National Missile Defense Deployment Readiness Review, p. 22, Figure IV-4.

Lisbeth Gronlund and David Wright are both senior staff scientists in the Global Security Program at the Union of Concerned Scientists and research fellows at MIT’s Security Studies Program.