ARTICLES The U.S. Government seems to be committing itself again to a large scale program of defense against missiles - battlefield, theater, and strategic. This is a national policy arena in which the intersection of politics, science,, society, and technology is paramount, an area in which Forum members have a responsibility in informing and guiding the necessary national discussion of this vital topic - impacting, as it does, on national, and world, economic and military security. Hence this issue of P&S is devoted to the subject. Because of the importance of a well-informed debate, length restrictions on articles have been waived for the general technical background piece by Richard Garwin. This is followed by more specific, advocacy articles by Gregory Canavan and Lisbeth Gronlund. Because of space constraints in this paper edition, the first two articles have been considerably shortened by the editors. Complete versions, as submitted by the authors, including footnotes, references, and figures, will be found in the Web edition of this issue. Technical Aspects of Ballistic Missile Defense Richard L. Garwin Presented at Arms Control and National Security Session, APS, Atlanta, March 1999 ABSTRACT Hundreds or thousands of ICBMs armed with nuclear warheads could destroy the United States; even a few could kill millions of people. Blocking or deterring the acquisition or use of such capabilities is highly desirable. This talk, however, deals with the technical problems and prospects for effective defense against ICBMs-- in boost phase, mid-course, and during and after reentry. Topics include detection, tracking, guidance, and the ability to destroy the target or render it harmless. For interceptor missiles: multi-stage propulsion, sensors, and guidance. For directed-energy weapons: generation, propagation, and kill. The revolution in microelectronics and computing, and the experience of decades render some of these problems trivial (in principle), but effective countermeasures vitiate many approaches. Three durable countermeasures against non-nuclear intercept include release of submunitions on ascent, large enclosing balloons, and anti-simulation. INTRODUCTION. The guided ballistic missile was first used in warfare September 8, 1944-- the German V-2, bombarding English cities from the continent. The predecessor to the V-2 was the V-1, an early cruise missile with a pulse-jet engine (the "buzz bomb") that British defense forces became quite effective in intercepting. Toward the end of the V-1 campaign, British fighter aircraft would fly next to the buzz bomb and with the wing tip of the fighter would tip the wing of the buzz bomb in order to divert it into less populous areas. It was fortunate that the V-1 did not have a simple contact sensor to cause it to explode when touched in flight. There was no defense against the V-2, except to attempt to destroy it before launch; an effort out of all proportion to the damage caused by the V-2 was expended to this end. On a flat, non-rotating Earth, without an atmosphere, a ballistic missile attains maximum range for a given launch speed with an elevation angle of 45 deg, and the range increases as the square of the initial velocity, as is evident from high school physics, R = V^2/g; for R in km, V in km/s, g is 0.0098 km/s^2 For missiles of intercontinental range, the fact that the Earth is almost spherical is important, since an initial velocity of 8 km/s that would correspond to a range of 6530 km on a flat Earth now corresponds to infinite range, since it is the velocity required to enter low-Earth orbit (LEO). Using conservation of momentum in the frame of the accelerating rocket (with velocity V(t) and mass M(t)) leads to the rocket equation dV/dM = -V_e/M --> dV/V_e = -dM/M --> \Delta V/V_e = -\Delta ln M --> M_o/M_f = e^{\Delta V/V_e} (Eq. 1) where M_o and M_f are the initial and final masses, and \Delta V is the "velocity gain" of the rocket. If Eq. 1 is used in the approximation of continuous staging, the Ve that appears in the formula is reduced from the actual Ve to compensate for the fact that staging is discrete and for the structure discarded at each stage. The limiting speed of a single-stage rocket is simply the rocket nozzle exhaust speed V_e times ln (M_o/M_f). For a rocket with no payload, the ratio M_f/M_o is the "structure factor". In the rocket literature, propellants are characterized by their "specific impulse"-- Isp, measured in seconds, which is simply V_e/g. It should be observed that even though the mass that can be propelled by a rocket to a given high velocity is exponentially smaller as this velocity increases, according to Table 1 the efficiency of conversion of the thermal energy of the rocket fuel to kinetic energy of the payload remains high-- to velocities well beyond those of interest in ballistic missiles. The first row is the final velocity V_f; the second \alpha. is the ratio of final velocity to rocket exhaust velocity V_e. The third row is the mass ratio m from the rocket equation, while the fourth row is the energy efficiency in converting to kinetic energy of payload the internal energy of a mass of propellant equal to the initial mass of the rocket less the final payload. Eq. 2 shows the relevant formula, \epsilon == (K.E.)/(P.E.) = (1/2 M_fV_f^2)/(1/2 V_e^2 (M_o-M _f)) (Eq. 2)  +----------------------------------------------------------+ | TABLE 1: For final velocity Vf achieved by rocket | | propulsion with exhaust velocity V_e = 3 km/s, the | | payload fraction is \mu and the fraction of fuel total | | energy present in the payload kinetic energy is | | &epsilon.. | +----------------------------------------------------------+ | | |V_f 3 6 9 12 15 18 km/s | | | |\a1pha 1 2 3 4 5 6 | | | |\m 37% 13.5% 5.0% 1.83% 0.67% 0.248% | | | |\eps 59% 62% 47% 30% 17% 9.1% | | | +----------------------------------------------------------+  with numerical values tabulated in the fourth row of the Table. Table 1 assumes practical "staging", without which there is a firm limit on the speed that can be achieved, even with zero payload. A "Scud" is a ubiquitous single-stage liquid-fueled missile manufactured in the Soviet Union or indigenously in many nations of the world. Hundreds of Scuds were exchanged between Iraq and Iran, and scores of "extended-range Scuds" armed with high explosive were launched by Iraq against Israel and Saudi Arabia. Iraq had lengthened the fuel tank of the Scud and thus increased its range in producing the al-Husayn missile. THEATER MISSILE DEFENSE IN ACTUAL WARFARE. (1) On January 18, 1991, a U.S. Patriot air defense system launched its interceptors against an al-Husayn attacking an air base in Saudi Arabia, but it soon became evident that the al-Husayn warhead was not an easy target. The modifications to the Scud had affected the stability of reentry (which normally takes place in these missiles with the warhead still attached to the spent missile), so that the warhead broke off and descended in a tight helix, greatly reducing the effectiveness of the intercept. The missiles usually broke up into three pieces-- the warhead, the fuel tanks, and the ---------------- 1 A useful reference is IEEE SPECTRUM, September 1997, "Ballistic Missile Defense: It's Back", with 5 articles on BMD. Also J. Pike, "Ballistic Missile Defense: Is the U.S. 'Rushing to Failure?', Arms Control Today, April 1998, Vol. 28, No. 3, p.9-13. (Available at http://www.armscontrol.org/ACT/april98/pikap98.html) rocket motor, but the Patriot radar and computer were usually able to distinguish the warhead from the other potential targets. The Patriot system uses a ground-based phased-array radar that continually scans in a flexible fashion a sector of the sky for approaching objects. It characterizes them and, either automatically or via manual intervention, launches one or two interceptors against the aircraft or missile. The interceptor itself is tracked by the radar and steering commands are automatically generated and transmitted in order to intercept the warhead within the atmosphere. The Patriot interceptor is endo-atmospheric, since it relies on aerodynamic forces to maneuver. The interceptor itself carries a receiver of the signal transmitted by the radar, so that in the vicinity of the offensive warhead the interceptor picks up the reflected radar return and relays it to the radar; "track via missile" approach greatly improves the performance of the interceptor against the targets for which it was designed. At the optimum time, high explosive in the interceptor warhead is detonated by the fuzing system, spraying the incoming warhead with steel pellets. It is entirely reasonable to have a defense against warheads armed with high explosive, just as it is worthwhile to defend against aircraft delivering such weapons.(2) U.S. defense against theater ballistic missiles is being improved with the substitution of the Patriot Advanced Capability-3 (PAC-3). This is a smaller interceptor that employs "hit-to-kill" technology and is to destroy the incoming warhead by the kinetic energy of its collision rather than by explosively driven fragments. A similar Navy system with endo-atmospheric capability is under development, with the first units being operational in late year 2000 and 2002 respectively. Enlargement of the area defended from a single interceptor site can be achieved with intercepts outside the atmosphere, and an Army theater high-altitude air-defense (THAAD) system and corresponding Navy theater-wide system are to be available in 2008 and 2010 respectively. Of course, radars ---------------- 2 I judge that no more than a small fraction of the offensive warheads were destroyed by Patriot intercept during the Gulf War, based on a review of the work published by T.A. Postol and his collaborators at MIT, as well as analyses published by Raytheon, the Patriot builder. See G.N. Lewis and T.A. Postol, "Video Evidence on the Effectiveness of Patriot during the 1991 Gulf War," Science and Global Security, Vol. 4, no. 1, 1993, pp. 1-63. must be able to detect and to discriminate the threatening warheads at greater range for these systems to be effective.(3) And they must also perform technically against targets that may have features that reduce the effectiveness of the defense-- i.e., countermeasures. DEFENSE AGAINST ICBMS. A typical intercontinental ballistic missile (ICBM) will have a range of 8000-10,000 km and a speed of around 7 km/s. Minimum propulsion for a given range is achieved with a reentry angle closer to 22 deg than to 45 deg because the range is not small compared with the radius of the Earth. Of course, the atmosphere of depth 10 tons per square meter exerts a substantial drag on the ballistic missile as it rises, so that a practical design does not achieve intercontinental speed within the dense atmosphere; this is accomplished by limiting the thrust and hence the acceleration of the missile. This also limits the "dynamic pressure" and the skin heating on ascent. But too low an acceleration implies excessive "gravity loss", since it is only the excess of vertical component of thrust over the force of gravity that actually results in acceleration of the missile. Even so, a typical large liquid-fueled missile may have a thrust of 1.3 times its gross weight, so that initially only 0.3/1.3 of the thrust accelerates the missile just after liftoff. The kinetic energy of the RV, which is a good fraction of the total energy of the propellant of the entire ICBM, must be dissipated on reentry. ICBM warheads must be protected by a reentry vehicle (RV) against the heat and deceleration of the atmosphere. A typical peak deceleration is on the order of 60 g. However, only a small fraction of this energy need be absorbed by the material of the RV-- the rest being carried off by the wake of the reentry. Although initial ICBM RVs used a heat-sink approach, this was soon superseded by a much lighter protection system that uses ablative material that gradually sacrifices its heated surface layer and erodes in a controlled fashion on reentry. The RV shape approximates a sharp cone with a small nose radius. An ICBM of concern to the United States is usually protected in a silo or concealed inside a mountain. It might need to be removed via horizontal tunnel and erected before launch, or it might be stored vertically, and launched in that position after opening a protective door. The missile can ---------------- 3 See footnote re MIT charts. be ignited in the silo, or it can be ejected more gently and ignited as it emerges. The first-stage booster, when fuel is exhausted, is then separated by explosive bolts, as are eventually the second stage and the third stage, so that the warhead travels on its own to the reentry point. In order not to prejudice accuracy because of the uncertain orientation of the warhead on reentry, the warhead can be fitted with monopropellant rocket jets to force it to pitch over to assume the appropriate orientation for reentering along its longitudinal axis. It is then often "spun-up" by additional jets, in order to maintain that orientation for the remaining 20 minutes or so of flight. Alternatively, the alignment with velocity might be delayed until the beginning of reentry, and the RV spun up at that time. The United States and other countries with multiple warheads or "penetration aids" often use a "bus" (more formally a PBV, for Post-Boost deployment Vehicle) which carries the guidance unit for the missile and which has the job of accelerating each of the warheads (or decoys or penetration aids) sequentially to the proper velocity so that it will fall to its particular target. Bussing may take five to ten minutes in order to distribute the warheads and appropriate decoys to accompany them. A ballistic missile defense system could interfere with the ICBM either before launch, or in boost phase (during the operation of the first, second, or third-stage engines) during the bus activity, or it could counter a warhead either above the atmosphere or on reentry, until it achieved its detonation altitude. The ICBM presents different vulnerabilities and opportunities for intercept in its various stages: PRE-BOOST-PHASE INTERCEPT. The United States tries to learn where all potentially threatening missiles are based. It could destroy them preemptively in the case of hostilities. However, some ICBMs are mobile, and if they are out of garrison and not otherwise observed, they are not vulnerable to such attack. Even if the location of the mobile launcher were known at the time of launch of ICBM, if it were in motion it would be safe from destruction by the nuclear warhead on a ballistic missile. But if the mobile launcher could be tracked continuously, then updates could be sent to a maneuvering warhead; the amount of divert fuel required against a moving ground target is negligible since the /Delta V is rigorously(4) less than twice the maximum velocity of the ground vehicle. ---------------- 4 F.J. Dyson & W. Press, "Transverse Boost Requirements for ICBM Targeting of a Maneuvering Aircraft," Appendix A from JASON report JSR-79-03 (1979). BOOST-PHASE INTERCEPT. The United States maintains in high-Earth orbit a set of Defense Support Program satellites (DSP) which for decades have reported in real time every ballistic missile launch of significant size. It was revealed officially that the U.S. observed in this fashion every Scud launched during the Gulf War. With a 6000-element linear infrared sensor that rotates once every ten seconds, DSP can determine the launch point with an accuracy on the order of a kilometer. Since a typical ICBM burns for about 250s, multiple observations are possible and pretty good trajectory information can be obtained in this way. In the early seconds of boost, an ICBM is vulnerable to a command-detonated mine adjacent to the site or to a rocket-propelled grenade. Even short-burning Scuds could be destroyed by small homing interceptors launched by radio from as much as 50 km distance from the launch site. Normal ICBMs would be vulnerable in boost phase to ground-based interceptors (GBI) (or sea-based interceptors) from anywhere within a region of about 1000 km of the launch site. Such an interceptor would be launched by command on the basis of DSP data, without there ever having been a radar detection of the ICBM. Fitted with a sensor capable of detecting the missile flame, it could direct its limited field of view in the direction commanded according to the data from DSP, and accelerate toward a predicted intercept point. The prediction would need continued refinement, by observation from the interceptor of the current position of the ICBM booster. But the interceptor would have to be launched from a site sufficiently close and have sufficiently high performance in order to reach the missile while it was still burning. Furthermore, the interceptor could not simply home on the flame but in the late stages of intercept would need to look "ahead" of the flame, in order to strike the solid missile and not sail harmlessly through the tenuous flame. This could be done either by blind reckoning because of the known shape of the flame, or by actual detection of the solid missile with a proper design of the interceptor seeker. Because of the ocean area east and north of North Korea, North Korean ICBMs aimed at the United States are an ideal target for ground- or sea-based boost-phase intercept. Specifically, it should be possible to use an interceptor of the same gross launch weight as the GBI of the NMD program (about 14 tons, with 12.5 tons of solid fuel) to boost the kill vehicle (of perhaps 60 kg mass and containing some 15 kg of liquid fuel) to a speed similar to that of the ICBM-- 7 km/s, but with larger engines relative to the mass, so it will reach its final speed more rapidly. A simple calculation shows that the sea-based interceptor could be deployed as much as 2100 km downrange from the launch site and still be able to catch the ICBM while it is still burning. We assume a burn time of 250 s to ICBM speed of 7 km/s (an acceleration of three times that of gravity-- "3 g") while the interceptor acquires 7 km/s in 100 seconds--an average acceleration of 7 g. Because the interceptor must rise vertically in the lower atmosphere, it probably moves only about 250 km toward its target while it is burning, and then in the remaining (250-100) seconds moves some 1050 km. So in the burn time of the ICBM, the interceptor can reach out a total of 1300 km from its launch site. The ICBM at an average speed of 7/2 = 3.5 km/s in 250 s moves no more than 875 km from its launch site. The interceptor could be deployed 1100 km east or west of the ICBM trajectory, about 800-1000 km downrange. So there is plenty of room for U.S. navy ships to carry these interceptors. The ships need have no missile-tracking radars. Such a sea-based boost-phase intercept system is not compliant with the 1972 ABM Treaty; but Russia and the three other parties to the Treaty might well agree to a specific exception, especially if this were combined with progress on lower missile levels in Russia and the United States. My own judgment is that the ABM treaty plays a valuable role in U.S. national security and in the reduction of Russian nuclear weapons, and that it should not be abandoned lightly. Alternatively, Russia and the U.S. might each deploy 15 test interceptors at a new joint ABM test range south of Vladivostok. The Air Force is developing an airborne laser (ABL) for boost-phase intercept. It is a chemical oxygen-iodine laser mounted in a Boeing 747 aircraft that has the task of focussing through the turbulent atmosphere (further disturbed by the passage of the laser beam itself) in order to weaken or melt the structure on a missile during boost phase. The ABL laser operates at 1.317 \mu m, perhaps(5) at a power level of 3 MW. The ABL will operate at an altitude of 13 km-- most of the time above the clouds and, assuming that the laser works as planned and that laser beam propagation is as assumed, Forden assesses the range for "decisive engagement" from 320 km for the al-Husayn to 185 km for the North Korean Nodong 1300-km-range missile. This assumes that a substantial arc of the missile skin must be softened so that the missile collapses. For a less catastrophic criterion, Forden estimates the range limit from the missile launch point to be 320 km, and 1000 km or more for an ICBM (assuming the ABL downrange from the ICBM launch and so can attack at closer range). He notes that ---------------- 5 G.E. Forden, IEEE SPECTRUM, Sept. 1997, "The Airborne Laser", pp. 40-49. ABL could be stationed west, south, and east of North Korea for use against Nodong missiles and ICBMs. BUSSING. This phase has no particular vulnerabilities and will not be further discussed. If one catches the bus toward the end of its maneuvers, one can counter in this way only a fraction of the RVs. Catching it at the beginning seems less likely than doing the boost-phase intercept. MID-COURSE AND EXO-ATMOSPHERIC INTERCEPT. RVs falling through space are on a highly predictable trajectory, so that repeated radar observation, for instance, can refine that trajectory and contribute to the intercept capability. Intercept of a single RV is simpler with a nuclear-armed interceptor, with an effective kill range measured in kilometers. Several long-range weapons effects can be important in this regard(6) -- x-ray-induced blowoff of the external surface drives a shock into the material of the RV and can damage the warhead by "spall" or by deformation of the structure; the fissionable material can be melted by neutrons from the interceptor thermonuclear warhead. Large and powerful radars are required to see a reentry vehicle at ranges of several thousand km, and such have long been deployed in the ballistic missile early warning system (BMEWS) in Alaska, Canada, Greenland, and Britain. The criterion is simply that the radar direct enough energy on the RV within the required search time for the reflected energy to the radar to exceed thermal noise. This is embodied in the radar equation: E_s = E_o \sigma/R^2 d\Omega, (Eq. 3) where the solid angle d\Omega = \lambda^2/A into which the radar energy is emitted is related to the antenna area A and the wavelength \lambda, on the assumption that the antenna is diffraction limited. The energy Es scattered by the target is proportional to the effective back-scatter cross section &sigma.. The energy received from the radar antenna E_r = E_s A/4\pi R^2 = E_o A^2 \sigma/(4\pi R^4 \lambda^2)=NkT (Eq. 4) where R is the range to the target, kT the thermal energy, and N the margin required for losses and to make thermal excursions above the signal threshold sufficiently rare. Radar and detection theory are highly evolved and can take into account the fact that reentry vehicles do not suddenly appear in space. The RVs have an easy ride through the vacuum of space, and an ICBM launch for which the RVs do not wish to be seen or identified can provide a vast amount of clutter or chaff ---------------- 6 Bethe, H.A. and Garwin, R.L. "Anti-Ballistic-Missile Systems," Scientific American 218, 3, pp. 21-31, March 1968. that can mimic the radar return from an RV. Early-warning radars usually operate in the VHF/UHF range (typically 400 MHz), so that a half-wave dipole is 37 cm long. Vast numbers of such dipoles can be formed of metal, glass, or carbon fibers with a metal coat, and can thus provide a substantial radar return that masks real RVs. Similar radar clutter can be provided by inflated balloons of metal-coated plastic, and, for good measure, one can put a balloon around the RV itself-- a simple form of "antisimulation" to enable the RV to simulate a decoy that is easy to make. The other approach to wide-field detection of RVs in space is via their thermal (infrared) radiation. An RV at the temperature of the Earth (with a black-body surface) radiates about 400 W/sq m, and in a 1-\mu m band centered at 10-\mu m about 40 W. In a typical example,(7) a 10-cm diameter telescope at a range of 1000 km from an RV with a black-body radiating area of 1 m^2 would collect about 1000 photons in a millisecond dwell time, for a scanning line of 5000 detectors. The focal plane could be cooled to liquid hydrogen temperature to reduce the self-generated "noise" from thermal radiation in the detector itself. In the modern era, one can use a "staring" array so that longer integration times are possible, although the motion of the light source across the visual field (because of relative motion of the RV and the observer) limits the integration time that can be used in this simple fashion. Such telescopes are planned to be mounted eventually on the space-based infrared system in low-Earth orbit (SBIRS-low) which would observe in the thermal ir looking at warheads against the black background of space. (Of course many stars will be observed, but they are readily discriminated from warheads because they do not move.) Detection of an RV from two or more satellites will fix it in space at the intersection of the two lines of sight, so that an interceptor can be directed accurately toward the RV which is moving in a Keplerian orbit. As the interceptor approaches, even a relatively crude ir telescope in the interceptor will be able to detect the RV. In the late stages of intercept, as the RV and interceptor coast into collision at a relative velocity typically of 11 km/s, small adjustments must be made in order to collide with the center of the RV, and not have a brushing collision or a total ---------------- 7 S. Weiner,"Systems and Technology" in "Ballistic Missile Defense", A.B. Carter and D.N. Schwartz, Eds. (The Brookings Institution, 1984). This volume is an accessible compilation of technical and strategic aspects of BMD. miss. The attached charts(8) show some of the elements involved in the modeling of the performance of an exoatmospheric kill vehicle (EKV) as a function of the assumed initial offset of the RV (the "impact parameter"). ENDO-ATMOSPHERIC INTERCEPT. Technically, intercept within the atmosphere is easier for the defense because the ICBM warheads are highly visible to radar and to optical sensors, because of the very hot "wake" produced by the Mach-23 RV as it enters the atmosphere. Balloons and light chaff(9) are no longer effective against sensors, because they will be retarded or destroyed on reentry. Within the atmosphere it is more difficult to make survivable and effective decoys that match the deceleration of the RV containing a nuclear warhead. And the interceptor can undertake much more aggressive maneuvers by aerodynamic force than it could conveniently with rocket propulsion in space. On the other hand, the RV is decelerating rapidly rather than existing in a well-defined orbit; it may also be maneuvering violently, whether intentionally or not. Sensors on the interceptor are much more difficult, since its high speed through the atmosphere requires heat resistant windows and adds greatly to the background in detecting infrared from the RV. Radars must be more closely spaced to see RVs down to altitudes of reentry, and interceptors cannot drive out hundreds or thousands of km through the atmosphere. So while endo-atmospheric intercept is important for defense against missiles of theater range, it is of little interest in the context of a national missile defense of the U.S. THE PROPOSED U.S. NATIONAL MISSILE DEFENSE. The NMD system under development by the Defense Department, according to Lieutenant General Lester L. Lyles, USAF, Director of the Ballistic Missile Defense Organization (02/24/99) "would have as its primary mission the defense of all 50 states against a small number of intercontinental-range ballistic missiles launched by a rogue nation." But General Lyles goes on "Such a system would also provide some residual capability against a small accidental or unauthorized launch of strategic ballistic missiles from China or Russia. It would not be capable of defending against a large-scale, deliberate attack." ---------------- 8 Yingbo He, "THAAD Interceptor and ABM Demarcation...", Presented at 6th ISODARCO-Beijing Symposium on Arms Control, Shanghai, October 1998 (used by permission of Yingbo He). 9 Illustrations in talk are from the MIT Defense and Arms Control Studies Program, Technology Working Group, used by permission of G.N. Lewis and T.A. Postol. As described by General Lyles, the NMD system is intended to use a ground-based interceptor launched from a site within the United States (North Dakota or Alaska) to strike reentry vehicles above the atmosphere. With North Korea as an example, in order to be specific, these ICBMs would have been launched toward the North in order to fall on the United States. The rocket launch flame will be detected by the Defense Support Program (DSP) satellites in geosynchronous orbit within considerably less than a minute after launch, and an approximate location of the launch site and a direction of the missile is established in that way. The upgraded ground-based early-warning radars operating typically in the frequency band of 420-450 MHz would some minutes later detect the threat missile and on the basis of these data confirming the DSP information, interceptors would be launched. While the interceptor is in flight, a ground-based X-band (10,000 MHz) radar with better resolution will track the reentry vehicles and to some extent discriminate them from other objects put into space by the missile (perhaps intentional decoys, certainly other parts of the missile) to guide the interceptor close enough to the target missile for the interceptor's sensor to acquire the warhead and "to discriminate the warheads from potential decoys." Several interceptors would need to be launched at each warhead in order to achieve the NMD requirement to have high confidence in no ICBM warheads impacting on U.S. soil. In fact, there is no specific design for the NMD, as it is still evolving. Against the simplest threat of a few "rogue nation" ICBMs without countermeasures, it is expected that 20 GBI would be deployed, either at the Grand Forks (North Dakota) site the U.S. specified under the 1972 ABM Treaty, or at a site to be selected in Alaska. Against ICBMs launched from North Korea, the Alaskan site would give better protection to dwellers of Alaska and Hawaii, about 0.7% of the population of the U.S.. The three-stage GBI would deliver the exo-atmospheric kill vehicle (EKV) on a near-collision course with the target. The EKV is to have a multi- (possibly 4-) band ir telescope as well as a focal plane for visible light and is to be capable of transverse ("divert") accelerations of 3 g or more. The task of colliding with the RV is daunting-- especially at closing speeds of 11 km/s, but the real difficulty would arise from lack of cooperation by the RV. In July 1998, the nine-member Rumsfeld Commission to Assess the Ballistic Missile Threat to the United States (on which I had the privilege to serve) issued its unanimous report, judging that North Korea could have a true (but unreliable and inaccurate) ICBM within a couple of years-- specifically within five years of a decision to move forward with a program, assuming that it is thoroughly funded with a high priority. The Rumsfeld Commission also advised that there were other and earlier threats from missiles of shorter range launched from ships, and observed also that BW or CW agents could be packaged in the form of bomblets released early in flight, that would fly separately to the target region. I elevate this last conclusion to the status of a likelihood. It is far more effective militarily for an ICBM payload of biological warfare agents to be arriving in the form of individual reentry vehicles (bomblets) spread over an area 10 or 20 kilometers in extent, rather than to be delivered as 100-500 kg of BW agent at a single point in the target area. Under the latter condition, a very narrow plume will be produced by wind-born BW, threatening people within the narrow plume. But if the same payload were dispersed in the form of bomblets, a large number of such narrow plumes, each equally lethal within its interior, would threaten people in the target area. Given this undisputed increase in military effectiveness, any nation with the capability to make an ICBM and reentry vehicles would almost surely arrange to package the BW in the form of bomblets, released just as soon as the ICBM reached its final velocity on ascent. Placing the bomblets at predetermined positions in a rack within a spinning final stage, the release of the bomblets would then allow them to spread during their 20-minute or more flight to reentry, with the initial rotation rate determining precisely the spread, and the pattern being that in which the bomblets were stored in the missile.(10) This threat of BW bomblets released on ascent is to be expected whether or not a defense is deployed, but the proposed NMD would have strictly zero capability against these bomblets. First, there would be so many of them (with a loading of perhaps 1 kg of agent per bomblet) that it would exhaust any planned number of interceptors. If North Korea obtains fissile material either from its own reactors or from abroad, so as to make a nuclear weapon that could be carried to intercontinental range by an ICBM, it would initially have what is probably an unreliable warhead on an unreliable missile. The warhead would be likely to miss its city target entirely. But would a defense make any ---------------- 10 Those bomblets ejected to the side will land at ICBM range almost as if the displacement grew linearly with time, but those dispensed in the plane of the minimum-energy trajectory will land at the same point as the RV, but at an earlier or later time. To obtain a circular pattern from bomblets dispensed in an expanding circle requires the axis of rotation first to be tilted with respect to the velocity. difference? Yes, if the launching country cooperated, but not if it wished to prevent the intercept of its nuclear warhead. Because the NMD interceptors are all "hit-to-kill" so that they must collide with the warhead in order to destroy it, the attacker need not conceal the existence of the warhead but only its exact location. This is readily done by the use of an enclosing balloon made of aluminum-foil coated mylar that can be put together by anyone who buys this article of commerce and a hand-held tool for heat sealing the plastic to make a large balloon. Even a balloon ten meters in diameter, inflated after the RV separates from the missile, would render it unlikely that an interceptor would actually strike the warhead rather than plunging harmlessly through the balloon. The balloon would be inflated in space by a tiny charge of gas-generating compound like that found in every automotive air bag, but instead of deploying in a 100th of a second or less, the balloon could deploy in a second. Since the launch country might fear that the interceptor striking the balloon might cause sufficient disruption to expose the RV, several balloons in sequence could be shrunk down on the RV (and would occupy very little space with the air removed by an ordinary vacuum cleaner). So each would be ready for deployment to hide the RV once again in case the balloon was intercepted. Alternatively, the launching country could deploy ten or more such balloons over a region 10 km or more in extent, so that these would need to be attacked one at a time. Even the dynamics of a balloon bouncing around over an enclosed object could be simulated in the decoy balloons with an enclosed object that weighed extremely little in that case-- a heavier, small balloon just big enough to enclose the RV in the one balloon in which it exists. This is an example of the utility of "antisimulation", in which the warhead itself is modified to make it easier to simulate by a cheap and convenient decoy. Thus, it may be desirable to choose alternative approaches that make decoys easier to produce than to copy the "bus" concept. One approach is to have the warhead rotate only slowly in space, and then to align itself with its velocity as it begins to reenter the atmosphere. The interceptor would normally track the RV by means of its infrared (heat) emission and it could readily distinguish an empty balloon from a balloon containing the RV, simply because the empty balloon would be colder, while the RV would not have had time to cool off during its 30 minutes or less of flight. But highly reflective aluminum not only reflects light (and infrared) but (by reciprocity and Kirchoff's law) it correspondingly radiates a lot less-- about 30 times less than does an unprepared surface. Furthermore, multi-layer insulation is an article of commerce that can reduce the emitted heat by another factor 50 or more. Finally, if the decoy maker wished to have even greater confidence that sensors would not be able to discriminate the decoy balloon from the balloon containing the RV, a small commercial lithium-metal battery weighing about 0.3 kg could be used to mimic the 40 watts of heat that would be emitted by the reentry vehicle shrouded in multi-layer insulation within its own balloon. The attached charts show a different approach-- in which a shaped shroud is made less visible at long range to the ir-homing telescope on the interceptor itself. These achievements are easy relative to the scale and cost of the effort required to develop an ICBM, and if a country expects the United States to have this NMD at the time of its first ICBM, then I am confident that these countermeasures can and will be provided. CONCLUSION The field of ballistic missile defense is full of fascinating problems of physics and engineering, but to have an effective defense requires attention to what the other side can do to defeat the system-- countermeasures. Since it is a big effort to analyze every plausible option and to choose the best, in a small program it is often better to choose an approach that gets the job done and avoids the cost and delay of the universal analysis. The nature of the defense the U.S. might build and its effectiveness depend critically on the type of countermeasures that might be encountered, in conjunction with antisimulation that facilitates the countermeasures job. A realistic assessment is necessary before a decision is made to build a defensive system and before large expenditures lock in a system that might be ineffective, to the detriment of approaches that are less susceptible to countermeasures. ____________________________________________________________ Final Richard L. Garwin Thonas J. Watson Research Center (IBM) P.O.Box 218, Yorktown Heights, NY 10598-0218 (914)945-2555, RLG2@watson.ibm.com