Mr. Chairman, distinguished members of the Committee, it is an honor to appear once again before this body and to have the opportunity to share some of my views on ballistic missile defense with you. My name is Peter Zimmerman. In 1969 I earned a Ph.D. in physics from Stanford University, and since then I have been active in many fields of pure science such as nuclear and particle physics, but also in more applied fields, including orbital mechanics and image analysis, which bear fairly directly on the business of this hearing. Since 1984 I have devoted myself, substantially full time, to analysis of problems where the hard logic of quantitative science butts against the less precise -- but no less important -- field of defense and security policy.
When I last appeared before you, I indicated my belief that the United States was not now faced with a ballistic missile threat which made the deployment of a nation-wide defense useful. And I inferred from the evidence I had gathered that no such threat was likely to emerge -- whether from the Third World, the former Soviet Union, or from our friends in Europe -- within the next decade or two. I want to emphasize today that nothing I have learned since last autumn has changed those views in the smallest amount. I testified against deployment of a ballistic missile defense system in November, and I am taking this opportunity to restate that position most emphatically so that neither you nor any of my colleagues can have any doubt about the motivation behind the analysis I will share with you on this occasion.
I first became professionally interested in the controversy surrounding the performance of the Patriot missile as an ATBM during the Gulf War last December and January when I read the exchange of columns published in Defense News by two scientists I have known for many years and whose work I have always respected: Charles Zraket and Theodore Postol, with whom I share this panel. It seemed to me then that the issues surrounding the Patriot were "decidable questions" of science and not political in nature. The fact that we are all here today proves that I was somewhat off the mark in that respect. Nonetheless, I believe that some of the questions can, indeed, be answered objectively. And I conclude, again from our presence here, that they must be so answered. At once.
Let me begin by saying that the impressions created by initial reports such as the one which appeared to claim "41 out of 42" scuds had been intercepted were not credible (1). No missile system is that good, even after long combat experience, and certainly not the first time out. The army did itself a disservice by allowing such obviously exaggerated reports to emerge and by allowing the president to go on record supporting such numbers -- even with the reservation that "intercept" might not really mean that the Scud warhead was destroyed. Even the later claims of an overall success rate in the 70% - 80% region also seemed to me to be optimistic. I would have come to that conclusion even had the Iraqi up-rated version of the Scud been the threat against which the PAC-2 Patriot had been designed, and even if the stresses of reentry had not caused most of the missiles to break up, unintentionally shedding parts which then became reasonably effective decoys. As an aside I note that break-up on reentry is not unique to the Al-Hussein missile; the German V-2 also frequently broke up on reentry, a problem which was not licked until very nearly the end of the V-2 assaults on London, Antwerp, and other targets.
I have received classified briefings on the Patriot's record, and so I have asked people from Raytheon and the Army to glance through my written statement to make certain that I have not inadvertently released classified information. Since little of the information which is still considered Secret has very much bearing on my analysis and conclusions, I feel comfortable in assuring you that, if I could use the classified data today, it would not change my conclusions in any significant way. By that I mean the following: if an unclassified analysis says that the Patriot successfully intercepted 45 % of the Scuds which were engaged in Israel, and a classified discussion indicated that the correct number were really 40%, or 50%, I would argue that the figures are identical for the purpose of determining if the Patriot "worked". The small size of the statistical sample, and the relatively poor data collection procedures followed during the war when people's lives were at stake and little thought was given to hard science, reinforce my decision to treat almost all numbers in this analysis as "fuzzy."
I would like to discuss several analytic questions with you today. The first problem is to determine the validity or invalidity of "photographic" evidence extracted from commercial videotape; the next is to look at the statistical reason why so many film clips of Patriots missing their targets compared to missiles destroying Scuds have shown up; the third will demonstrate that, high-speed impacts of missiles without warheads (or with dud warheads) produce bright flashes of light which sometimes look amazingly like warhead explosions on film or video tape. A fourth section of my statement focusses on estimates of interceptor success and Scud-inflicted damage during engagements over Israel, where I think the best data are available; finally, I will
share some thoughts with you about the lessons Patriot's success has for the likely success of strategic defenses against nuclear ballistic missiles and, hence, the advisability of embarking on a full-scale SDI development and deployment program today.
I believe we will all agree that commercial videotape from news cameras is not a precision product, useful as an input to serious scientific study. This is particularly true for the study of night engagements between Patriots and Scuds when the cameras were operated at the extreme lower limit of their sensitivity, staring at a dark sky with only a few points of very intense light in their fields of view. That is why we saw so much snow in the pictures we watched every evening. I will show, however, that commercial video is not merely unsuited to scientific study, but guaranteed to be misleading.
In order to use videotapes to indicate whether a given intercept resulted in a kill or a miss one must make precision measurements of the pictures themselves to locate the two missiles in space. Only then can the analyst seek evidence of a kill on the videotape itself. Making the precision measurements is called photogrammetry, and it cannot be done using ordinary television pictures.
Most videotape which has been used in public (and in private) shows a single view from a single camera. Without stereoscopic information, it is impossible to locate either the Patriot or Scud missile in three-dimensional space; it is even impossible to locate the two missiles relative to one another. No commercial videotape of intercepts shows anything in the view field at approximately the distance of the target from which one can judge scale, distance, or relative distance, and so, it is impossible to determine the distance from Scud to Patriot by measuring the apparent distance from one object to the other as captured on tape. This remains true even if the focal length, the magnifying power, of the camera lens is available, which it usually is not because most TV cameras have continuously adjustable zoom lenses -- and few cameramen were taking careful notes while they were under fire.
Single view imagery is not entirely useless. While two objects may, in fact, be farther apart than they appear without information from the third dimension, they cannot be closer together. And so, if one saw Scud and Patriot miss by a considerable distance, one ought to be certain that a miss had occurred. In fact, appearances can deceive, even in this simple circumstance.
If precision high frame-rate scientific stereo film were available, one could calculate the trajectories of both missiles and then determine if the Patriot ever got close enough to the Scud to wound it mortally. However, no precision films appear to exist. It has been claimed that precision video tapes existed in Israel. In fact, the Israeli video tapes had a data acquisition rate of 50 fields per second (25 full frames delivered as 50 interlaced fields) and so were even less
suitable for precision analysis than some U.S. news tapes. The Israelis did try to use high speed tracking cameras from their test ranges in conjunction with Patriot fire units, but were unable to lock onto and track the Patriot/Scud engagements. Hence they were unable to acquire useful data.
It is impossible to determine the relative positions of interceptor and target from videotape running at standard speed. Consider that U.S. standard television displays 30 complete frames each second produced as 60 interlaced fields or "half frames." Thus 16.7 milliseconds elapse between sequential scans of the same point, while the closing velocity of target and interceptor is approximately 9,400 miles an hour. The range between target and interceptor changes by about 230 feet between sequential half frames. Most action takes place between the scans of the two missiles. Most modern TV cameras use charge coupled devices, CCDS, which take 60 "snapshots" each second. To be compatible with the U.S. television standards these single frames are transmitted as 60 interlaced fields per second to provide the equivalent of 30 complete frames per second. This introduces an additional source of error.
For some modern CCD cameras, the actual time between independent frames is 1/30th of a second, 33.3 milliseconds, and so all of the distances I gave above must be doubled. The relative positions of the two missiles changes by 460 feet between pictures.
Locating the point in space at which the Patriot warhead actually explodes and launches its fragments is impossible with this spatial and temporal resolution. In fact, a correctly timed burst of the Patriot's warhead will always appear to take place behind the Scud. The actual explosion only lasts 0.2 millisecond, and cannot be captured on tape.
The first television field recorded after the Patriot detonates will come an average of 8.33 ms after the detonation. This frame will show the position of the explosion with reasonable accuracy because the visible cloud is incandescent air and does not move along with the Patriot's velocity, but it will show the Scud warhead at its position after the explosion of the interceptor. Since the Scud is still moving at about 5,600 miles an hour when it is hit, the Patriot's explosion fireball must always be behind the target. Unless the Scud's warhead is detonated high-order by the Patriot, there will be no evidence in this video field that a successful intercept has taken place.
The next several frames are unlikely to show any evidence that the Scud has been neutralized either. The Scud warhead weighs several hundred pounds, even if it has completely separated from the airframe of the missile, and so its trajectory is not likely to be altered directly by collision with a handful of Patriot warhead fragments, each weighing only 45 grams, an ounce and a half. It is inescapable that the videotapes will show a Scud continuing, seemingly unperturbed, along its original trajectory. It will be many tenths of a second, perhaps even several seconds, before aerodynamic forces acting on the shredded skin of the missile begin to shift its course by an amount visible to a TV camera, and even then their effect is likely to be most manifest by the shedding of luminous pieces of red hot metal.
Let me outline for you a simple, but still technically accurate, example. Consider a series of television frames shot by an older "vidicon" or tube type TV camera. Vidicons record 60 fields per second to make up the 30 frames of normal TV, so they are actually better than a modern camera at capturing fast action. The sequence I will describe begins when the two missiles are 150 meters (458 feet) apart and on an almost perfect collision course. The first field transmitted by the camera will show a Patriot and a Scud separated by 458 feet. One sixtieth of a second later the TV camera snaps another field [field 2]; the range between the two missiles has closed to 80 meters, 244 feet. The TV image accurately shows this situation. One sixtieth of a second later still the camera snaps another picture [field 3]; the two missiles are now 10 meters apart, 31 feet. The Patriot is capable of measuring the distance between itself and its target, and selects an appropriate distance at which to detonate its warhead. That distance is classified, but is certainly less than 30 feet. Therefore, this field certainly cannot show the detonation of the Patriot warhead.
To make things simple, and unclassified, assume that the Patriot detonates when it is exactly even with the Scud. It cannot really do this, because it must lead its target slightly, as any duck hunter understands, but the error produced in the rest of my analysis is negligible. The Patriot's warhead detonates between field 3 and field 4, one sixtieth of a second later still. What does the TV image show us on field 4, the first picture taken after the Patriot explodes?
Using unclassified numbers, the closing velocity of the Scud and Patriot is 9,400 miles an hour (4,200 meters/second). Since the fireball from the Patriot's warhead is heated air mixed with the gases from the explosives themselves, it does not continue to move ahead with the full speed of the Patriot. Let me assume that it stops dead and marks exactly the point in space at which the Patriot detonates -- this is certainly what Professor Postol suggested in his television news interviews and in the videotape which he showed at the Chicago meeting of the American Association for the Advancement of Science. In fact, several milliseconds elapse between the time of the explosion and the time the hot gases are braked to a stop along the trajectory of the interceptor, so the real situation is far more favorable to my line of reasoning than is this assumption.
Field 4 is snapped 16-2/3 milliseconds after field 3, but more importantly 14 milliseconds after the Patriot's warhead launched its fragments at their target. In that 14 milliseconds the Scud travelled 35 meters, 108feet, past the point of closest approach.
On the television screen, field 4 appears to show a late detonation, a complete miss by more than 100 feet. We know that this is not the case; the intercept was successful, and the Scud warhead which continues more or less along its original trajectory probably resemblesa lump of Swiss cheese. The moment of intercept could only have been captured on a field recorded between fields three and four, a missing field, a field which no TV camera could have registered.
To be sure, the intercept which I have described occurred early in the interval between fields 3 and 4. With slightly different assumptions the detonation could have taken place a
moment before field 4; with other assumptions the intercept might have occurred half way between the two recorded fields. One thing, however, is certain: the first field on which the explosion of the Patriot warhead can be recorded will take place after the explosion. And the Scud warhead will have moved a significant distance from where it was when the intercept took place.
If this engagement had been recorded with a modern video camera which takes 30 independent full resolution frames per second, frames which are electronically broken down into sixty fields per second, the Patriot warhead would appear to have detonated even later, making the miss distance appear to be about 200 feet.
Video images will always show a late detonation; it is in the nature of the recording process. Television will always show the Patriot detonating behind the Scud. As a result, videotape will nearly always mislead the unwary analyst into perceiving a clean intercept as a miss with the Patriot warhead detonating too late, behind the Scud warhead which is its target.
Many videotapes which actually do show intercepts have been called failures because a large, glowing, fragment continues along roughly the trajectory of the incoming Scud. While some of these may be misses, one should remember that non-nuclear warheads do not annihilate their targets. One or more large pieces of Scud debris must emerge from the intercept, and a badly damaged warhead or airframe is likely to have glowing fragments torn off by the air screaming by at about 5,600 miles an hour.
Some Patriot warheads probably did explode late, and many missed their targets completely. Indeed at least two out of three Patriots launched missed their targets since three interceptors were launched at each Scud, but six out of ten Scuds were hit by one of the interceptors launched at them. If the first interceptor hit its target, the others detonated on surviving fragments, or in the air; if the second interceptor hit the Scud, it is obvious that the first must have missed. The reason so many Patriot missiles were fired was to improve the chances of getting one lethal hit on each incoming tactical ballstic missile.
But many of the "misses" depicted in the available commercial videotape archives were good intercepts, mistakenly interpreted.
One can find many videotapes of Scuds landing and apparently exploding, and even more of Patriots which really do miss their targets. One might infer that if 80% of the Scuds engaged over Saudi Arabia and 50% of those over Israel had been successfully intercepted, a large number of hits would turn up in the videotape archives and a smaller number of misses. The facts are otherwise.
Consider the situation in Israel. On average four Patriots were launched at each incoming Scud which was engaged. This expenditure of interceptors was due to the standard firing
doctrine and the fact that, early in the war, some interceptors were fired against debris and false targets. According to Israel Defense Force (IDF) reports, somewhat fewer than one half of all attempted intercepts met with success -- the origin of the U.S. Army's figure of "almost" 50% success (2). Certainly no more than one Patriot from the quartet launched for each engaged Scud will intercept successfully (if the first hits, the Scud's trajectory is likely to be so perturbed that the second Patriot will not fuze close to the target, etc.).
And so, for every eight Patriots launched, there will be only one success. From a box of random unlabelled videotapes of intercepts over Israel, seven out of every eight will show misses, demonstrating that it is a lot easier to find video of misses than of hits.
For engagements above Saudi Arabia where the Army has reported a much higher rate of success than above Israel, one might guess that the situation would be a little more favorable for finding video of hits. There are reasonable estimates which suggest that "about 80%" of the intercepts were successfull.
In Saudi Arabia an average of three interceptors was launched at each Scud which was engaged, so one random film clip in three would show a hit if 100% of all engaged Scuds had been destroyed. That was not the case, so the fraction of videotapes showing successes would actually be less that one out of three or 27%. The correct result for Saudi events is that only about 27% of all random news videotapes would show successes, but 73% would show misses. The Saudi situation is not significantly different from the Israeli case, and in neither instance would one find very many successes.
Anybody who has ever dug a freshly-spent bullet out of a sandbox knows that kinetic energy can be swiftly transformed into a great deal of heat. The kinetic energy in a very rapidly moving missile is, per kilogram of mass, far higher than in a bullet. An M-16 rifle slug travels at about 1,000 meters per second (3,050 feet per second) according to Jane's Infantry Weapons; a Scud travels at 2,500 meters per second (8,200 feet per second) and so carries more than six times the kinetic energy per kilogram or per pound. Even after deceleration in the atmosphere a Scud warhead is likely to hit the ground at a kilometer per second. A rock travelling at 2800 meters per second has as much kinetic energy as the same mass of high explosive. It is no wonder that when a Scud warhead impacts, even if it has been dudded, a bright flash is often seen.
This is not an unusual phenomenon. I have observed it in high-speed film of unarmed anti-tank missiles striking their targets at speeds far lower than that of a Scud hitting the ground.
While the kinetic energy flash looks much like an explosion, it lasts for a much shorter time than does the fireball from a Scud warhead detonation. One must not use the presence of a bright flash coinciding with the impact of a Scud fragment or warhead on the ground as prima facie evidence of a warhead explosion; the only evidence of a high-order explosion would be the presence of a large crater or significant damage of a type which is characteristically produced by explosives.
Simple calculations indicate that the 500 to 600 pounds of high explosive in a Scud warhead would be sufficient to subject an area 25 to 50 yards in radius to 5 pounds per square inch blast overpressures. This is sufficient to demolish virtually all types of residential construction and much commercial construction. When Scuds explode in populated areas such devastation is unmistakable, and the lack of such devastation associated with the identified impact of a Scud warhead is conclusive evidence that the warhead did not explode on impact. An Israeli friend tells me that the damage from a real warhead detonation was visible for up to five blocks away. Duds and falling tank fragments do not produce such effects. The most a tank can do is to punch a hole in a roof or a building; it does not destroy structures tens of yards away, even if a small amount of ullage fuel should happen to ignite.
To arrive at correct conclusions one must compare apples with apples and not with Jaffa oranges. In assessing the ratio of damage done in Israel "before Patriot" was operational to that after Patriot fire units were fielded, one cannot simply compare the number of damaged apartments in the two time periods and the gross number of Scuds launched at Israel. One must count the number of Scuds which fell into the soon-to-be-defended areas of Tel Aviv and Haifa "before Patriot" to the number of missiles which would have impacted in the defended areas after the fire units were deployed. Patriots would not have been launched against missiles which were expected to land in the Mediterranean or in the desert; neither would such a missile have been likely to cause any damage.
The exact number of Scuds which were predicted to impact within inhabited areas of Tel Aviv and Haifa before and after the deployment of Patriot fire batteries remains classified, but the ratio between those two figures is not. About three times as many Scuds were expected to hit the defended areas after Patriot became operational as actually hit before the interceptor system came on line. So if the number of apartments destroyed in the defended period was 50% less than the number of similar dwellings destroyed before, as the data in Professor Postol's International Security article indicate, then one must credit the missile defense with reducing the number of apartments destroyed by a factor greater than four. This is not the characteristic of a system which "failed."
A more statistically meaningful number may be the ratio of apartments destroyed or severely damaged before and after since severe damage was far more widespread than total destruction. "Severe" destruction would also be less susceptible to small random fluctuations because, as noted above, a Scud would have to detonate within about 50 yards of an apartment
building in order to level it. An explosion 75 or 100 yards away would do severe damage, but not level the building. According to the data presented in Professor Postol's International Security article, twice as much severe damage was produced in the defended period than in the undefended period. Since there were slightly less than three times as many tactical ballistic missiles (TBMS) fired in that same time, then it would appear that Patriot resulted in a factor of 1.5 attrition of the Iraqi attack, as reckoned at these damage levels. Widespread superficial damage might actually increase during Patriot missions since the interceptors were supersonic at low altitudes and, because of an apparent attempt to intercept Scuds at low altitudes, were kept closer to the ground in Israel than in Saudi Arabia. Hence, sonic booms could be responsible for such damage as broken windows and cracked roof tiles.
If the Patriot successes and failures claimed on a Scud-by-Scud basis by the U.S. Army were available, it would be interesting to look at the damage reported night-by-night to see whether more damage was produced on nights where the Army and IDF agree that the Scuds were not successfully intercepted than on the nights when the defenders claim to have been successful. Public availability of this information awaits the declassification of the relevant missile performance data. As an analyst who has examined some of this kind of data, I can tell you that I consider its tabulation to be the most convincing evidence I have seen that Patriots caused Scuds to fail to explode on a regular basis.
One problem which surfaced early-on in the discussion of whether or not the Patriot "worked" in the Gulf was the quality of the supporting evidence and the way in which it was used. In effect the scorecard listed "hits" and "misses," lumping together all potential successes into one category. But the evidence was never that clean. A more appropriate way to tabulate the data might have been to identify the few incidents -- three or four, I have been told -- in which the Scud warhead detonated in flight and call them "certain" or "confirmed" kills. Another group would have included those engagements for which a kill was judged to have been probable, but this category spans a spectrum from "almost certain" down to "possible but not highly likely." This spectrum should have been recognized, and broken down into two or three subcategories with separate tabulations.
A few engagements fall into the category of "unknown outcome" where we can never be certain whether or not the Scud was interdicted due to a lack of evidence. And finally the number of misses where the Scud penetrated the defense to explode on the ground should have been tabulated. I believe that such a pragmatic analysis of Patriot's real successes in the Gulf would have been defensible and credible, and would have eliminated the reasons for the present Congressional investigation.
The Patriot missile was an astounding success, even if it only cut in half the damage which could have been produced by the Scuds. Patriot was used to defend against a threat beyond the outer edge of its original design envelope, but it frequently succeeded. For the first time in history ballistic missiles launched in combat were countered by defending interceptors, and as often as
not the good guys won. How many high-tech weapon systems, weapon systems operating in an entirely new arena of warfare, do you know of with that good a record in their first combat trials? Nevertheless, it is easy to exaggerate the value of that success as an indicator of the future success of other weapons.
It is tempting to project from the present-day success of the Patriot in intercepting slow (compared to intercontinental ballistic missiles), short range TBMs armed with a few hundred kilos of high explosive and equipped with only "accidental" countermeasures to the probable future success of interceptors intended to hit modern, fast, long range, strategic ballistic missiles armed with weapons of mass destruction. Indeed, Ambassador Henry Cooper, Director of the Strategic Defense Initiative Organization, has frequently made such a connection. This is logically false.
The technical problems of intercepting Scuds are orders of magnitude different from those of intercepting nuclear-armed strategic ballistic missiles:
But today we hear talk of deployment of a system of limited strategic defenses, GPALS or "Global Protection Against Limited Strikes." By definition a limited strike is going to be limited in time, probably to one set of missiles launched over a period of minutes or hours. An inadvertent, accidental, or unauthorized strike will be a unique event, with missiles launched over a period of seconds or minutes. There will be no time whatsoever to learn from our mistakes and to understand unexpected events or surprising behavior on the part of the attacker's missiles. Once before in testimony I suggested that a
strategic defense system will probably "get it right" by the second or third nuclear war. I stand by that judgement today.
"If Patriot, therefore SDI" is false logic. I urge you to reject the proposition.