Silencing the Bomb Read online

  Carl Romney was the lead seismologist working for the Air Force Technical Center (AFTAC) of the Department of Defense. AFTAC operates the U.S. classified (secret) monitoring system and the data analysis center for detecting and identifying nuclear tests by other countries. (In later chapters I discuss my involvement and controversy on several occasions with Romney about monitoring nuclear explosions and determining yields of Soviet explosions.) Although it is not very evident in his 2009 book Detecting the Bomb: The Role of Seismology in the Cold War, Romney was very conservative and incorrect in his assessments of monitoring the Soviet Union for many decades. He and Teller were extremely distrustful of arms control agreements with the Soviet Union and of the effectiveness of nuclear monitoring.

  Many U.S. public officials in the 1960s were left with the misguided impression that it was possible to construct a cavity suitable to fully decouple an explosion of 300 kilotons and that such a nuclear explosion would not be detectable even with much improved seismic networks. Unfortunately, the issue continues to be contentious today.

  Within only about a year of his proposal in 1959, Latter and a few others presented the decoupling concept as a well-developed and well-tested hypothesis at a joint U.S.-USSR-UK conference in Geneva and in hearings by the Joint Committee on Atomic Energy of the U.S. Congress. Nevertheless, publication of the decoupling theory and data from the Cowboy chemical experiments in salt to test it did not occur until 1961.

  Most federal agencies did not possess the scientific or technical expertise to evaluate the decoupling hypothesis, even with appropriate security clearances and the need to know. Most of those with purported expertise on decoupling were in the weapons labs, mainly Livermore.

  The calculations and inferences about decoupling by Latter and his associates from 1959 through the 1960s were incorrect and quite misleading for a number of reasons. They overestimated the amount of muffling by a large factor. They made extrapolations of explosions to different depths in the Earth and to other rock types without any basis in actual nuclear testing. They also used simplified assumptions about the containment of radioactive products. They assumed incorrectly that rock materials were uniform and without cracks, joints, and other imperfections that were familiar to geologists and geophysicists. Latter stated that muffling factors as large as two to three thousand could also be achieved by lining a large underground cavity with an absorbing material like carbon, even though U.S. tests incorporating that concept with two small nuclear explosions proved otherwise.

  Based on no experimental observations, Latter stated in 1959 that the propagation of seismic waves from underground explosions in salt should be poorer than those from Rainier, which was conducted in tuff, a light, porous rock formed by the consolidation of volcanic ash. When the United States resumed testing in late 1961, it conducted its first peaceful nuclear explosion, Gnome, of 3 kilotons underground in a thick salt deposit in southeastern New Mexico.

  Both the United States and the Soviet Union conducted peaceful explosions to form cavities in salt for storage of various products, breaking tight rocks for petroleum extraction and other purposes described later. Gnome was fully coupled, not muffled, and its seismic waves, in fact, were large, not small compared to Rainier’s as Latter had wrongly deduced. Seismic amplitudes were especially large for paths to stations to the east of Gnome in the United States. Seismic waves recorded for the 5.3-kiloton Salmon explosion set off later in salt in Mississippi on October 22, 1964, also were large at seismic stations in eastern and central North America.

  These areas acted as a good comparison for the more easily detected efficient propagation of seismic waves in similar old and strong rocks of the crust and uppermost mantle of the Earth beneath most of Russia. Similarly, the seismic amplitudes and magnitudes of many later USSR nuclear explosions in salt were large and easily detected compared to explosions of similar yields in softer rock in Nevada.

  President Eisenhower chose James Killian, the president of MIT, as his science adviser in late 1957 and looked to him for scientific guidance about arms control proposals. Killian was aided by scientists appointed to the new President’s Science Advisory Committee (PSAC), many of whom stressed the value of a test ban in preventing fallout from atmospheric explosions as a first step in controlling the arms race.

  This helped to counter the negative views about nuclear arms control of Admiral Lewis Strauss of the Atomic Energy Commission; Edward Teller; Admiral Arthur Radford, chairman of the Joint Chiefs of Staff; and many others in the Department of Defense. As an example of some of the rhetoric during this era, Radford stated in May 1957, “We cannot trust the Russians on this [nuclear testing] or anything.” Those who still take that view today, sixty years later, are unlikely to change their minds about a Comprehensive Nuclear Test Ban Treaty even in the face of huge improvements in verification.

  In his 1977 book Sputnik, Scientists, and Eisenhower, Killian, a pragmatist who would have been classified politically as a moderate or Rockefeller Republican, discussed the “big hole” or decoupling concept of evasion in his role as Eisenhower’s science adviser. He states, “Teller wished to make a dramatic demonstration of the possibilities of cheating, and this [decoupling] was it.” Killian goes on to note that “the Berkner panel heard Latter’s theories about the big hole, and in its report concluded that decoupling techniques existed which could reduce the seismic signal by a factor of ten or more…. The big-hole technique proved to be much more difficult than expected by its advocates…. It was a bizarre concept, contrived as part of a campaign to oppose any test ban.”

  Killian continued, “I was asked by the State Department to lead an American technical delegation to London to give the British the information about the ‘big hole’ and other methods of concealing nuclear tests…. While we were in London, Dr. Latter said to me in casual conversation that whatever advances might be made in detection technology, the West Coast group led by Teller would find a technical way to circumvent or discredit them.”

  In discussing the process of scientific review and advice to the government, Killian said, “As Henry R. Myers has written, this is true even today. There seems to be a widely held obsession with the possibility of violations rather than with their probability, or their significance…. Opponents of limitations on nuclear testing have exploited this obsession by encouraging fears that have little basis in fact. We should have strengthened the campaign for a test ban by making clear when an apparent technical question is not really technical…. We who spoke for science never succeeded in making clear the difference between probability and possibility.”

  The hypothesis of decoupling was confirmed in general during the Cowboy experiments of 1959, in which chemical explosions of up to one ton (0.9 metric tons) were set off in small cavities excavated in salt in Louisiana. Decoupling or muffling factors of one hundred were calculated for salt soon afterward, using the Cowboy data and theoretical calculations. Nearly twenty years after the Cowboy experiments, however, it became known that the decoupling effectiveness for those chemical shots had been overestimated. The amounts of decoupling for Cowboy were revised downward by 30 percent to account for new information on the energy release associated with the detonation of the unconfined high explosive, Pelletol, which was used in the cavity explosions for Cowboy, as compared to the energy release of the same explosives when they were confined.

  This revision to a factor of seventy brought the muffling factor from the Cowboy experiments into agreement with the decoupling factor of seventy obtained for the U.S. Sterling underground nuclear explosion of 0.38 kilotons (380 tons) in 1966. Sterling was detonated in the cavity in Mississippi produced by the 1964 Salmon nuclear explosion of 5.3 kilotons in salt.

  Since 1966, several people working on evasive testing have continued to extrapolate the Cowboy and Sterling data to yields as large as 10 to 100 kilotons. Clearly, given the political implications for monitoring a CTBT, much better peer review of the original Cowboy data and a similar experi
ment by independent groups should have been made. Peer review also would have been in order for the early estimates of the amounts of muffling claimed by Latter and others.

  I decided to work on decoupling around 1985, because it was so critical to the success of a future CTBT. I discussed muffled nuclear testing in a long paper I wrote in 1996, “Dealing with Decoupled Nuclear Explosions Under a Comprehensive Test Ban Treaty.” John Murphy of Science Applications International Corporation (SAIC), a proponent of the belief that other countries were able to conduct large decoupled explosions evasively, described my 1996 paper as “the other view.” Hans Bethe of Cornell University, one of the most important physicists of the twentieth century and an arms control advocate, however, was very complimentary about my paper, which greatly pleased me.

  Actual data on decoupled (muffled) nuclear explosions are very meager because little testing was undertaken and most information on it is thirty to forty-five years old. The 2012 National Academies Report states that in the era of nuclear testing prior to early 1996, nuclear experiments by the United States to test this scenario were not considered important enough to be given priority and financial support except at very small yields.

  The present nuclear decoupling database includes the following:

  1. Only one fully decoupled nuclear explosion, 0.38-kiloton Sterling, detonated by the United States in Mississippi in 1966

  2. A partially decoupled explosion, Azgir 3-2, by the Soviet Union at Azgir in western Kazakhstan in 1976, 8 to 10 kilotons

  3. Mill Yard in 1985, about 0.02 kilotons (20 tons), and three other very small nuclear explosions at the Nevada Test Site (NTS)

  The Sterling and Azgir 3-2 nuclear explosions were conducted in cavities in large salt domes created some time earlier by much larger, fully coupled nuclear explosions—Salmon of 5.3 kilotons and Azgir 3-1 of 64 kilotons. Salt domes are large bulbous structures of nearly homogeneous salt formed by the instability of less dense salt rising upward through more dense sedimentary rocks.

  A huge cavity in salt is needed to fully decouple a nuclear explosion of 5 kilotons (figure 4.1). The depth range for containment of fully decoupled explosions of various yields in salt is illustrated in figure 4.2. At shallow depths, nuclear explosions are not contained and either blow out to the surface, form large craters, or leak detectable radioactive materials (upper blue area). The minimum depth for containment increases with yield.

  FIGURE 4.1

  To fully decouple a 5-kiloton nuclear explosion in salt requires a spherical cavity of diameter 282 feet (86 m) at a depth of about 3000 feet (900 m). It would be larger than the Statue of Liberty and its pedestal.

  Source: Office of Technology Assessment, 1988.

  FIGURE 4.2

  Depths (vertical axis) for conducting decoupled nuclear explosions in an air-filled cavity in salt. Yield of the explosion is on the horizontal axis at top; the cavity radius is at bottom.

  Modified from Davis and Sykes, 1999.

  Salt, however, is unlike most common rocks in that it deforms readily at low confining pressures and shallow depths in the Earth’s crust. A stable air-filled cavity in salt becomes unstable and deforms severely or collapses at depths greater than about 3000 feet (900 m). Thus, the depth range in the Earth for conducting a fully decoupled nuclear explosion in salt is quite limited. The 1966 Sterling and the 1976 Azgir 3-2 explosions were detonated at nearly optimum depth, about 3000 feet, to insure containment and cavity stability. If those explosions had been conducted at shallower depths, larger cavities of greater volume would have been needed to obtain the same amount of decoupling.

  The USSR’s 1976 Azgir test is very important for estimating the detection and identification capabilities of decoupled nuclear explosions larger than one kiloton because its yield was about twenty-three times larger than that of the U.S. Sterling. This 1976 explosion, however, was about three times larger than that required for full decoupling to occur. Hence, it is not surprising that it was only partially decoupled, with seismic waves muffled by twelve to fifteen times, as I reported in my 1996 paper, not the seventy times associated with full decoupling. The cavity in salt in which the 1976 Azgir explosion was detonated was huge, with a mean diameter of 243 feet (74 m). It was large enough to contain the Statue of Liberty and its pedestal. Its seismic magnitude mb of 4.06 was well recorded by stations as far away as Canada.

  Nevertheless, on May 10, 1976, Newsweek reported that the 1976 Soviet Azgir explosion had been promptly identified as a fully decoupled nuclear test. The yields mentioned in that article on it and the 1971 nuclear test that created its cavity, however, were not corrected for seismic magnitude bias (described earlier). Newsweek’s yields were much too large—210 kilotons for 1971 and 40 to 50 kilotons for the 1976 test, not 64 kilotons as published by the USSR for 1971 and 8 to 10 kilotons that I calculated for 1976 by including my estimate of seismic bias.

  As I stated in my 1996 paper, “The Newsweek article conveys the impression that the USSR was capable of conducting a decoupled test of 40 to 50 kt, not as we now know, only a partially decoupled test of 8 to 10 kt.” News media don’t just make up a story with details like that. Some person or agency must have provided that information. Again, the U.S. overestimation of Soviet yields had major implications for national security policy, in this case a large overestimate of what the Soviet Union could achieve by evasive nuclear tests.

  Decoupled testing requires an air-filled or an evacuated cavity. Very large cavities have been constructed in salt by conventional and solution mining, mainly for storage of oil, gas, compressed air, and waste products. Solution mining is the least expensive method of forming a large cavity at depth in salt. It involves using a pipe to pump fresh water from the surface deep into salt. A mixture of salt and water called brine is then pumped back out through another pipe.

  Disposal of brine is a major problem, requiring about seven times the volume of water to extract one volume of brine using solution mining. A brine-filled cavity is not suitable for conducting a decoupled explosion because an explosion in brine or water is well coupled, producing large seismic waves. Hence, brine must be pumped out of a cavity and disposed of in secret if it is to be used for decoupled testing.

  Solution-mined cavities are also very fragile. Very large brine-filled cavities rarely have been pumped out, leaving them empty. The U.S. Strategic Petroleum Reserve is stored in large cavities in salt created by solution mining at Bryan Mound, Louisiana. Brine in those cavities is typically replaced by oil. When oil is removed later, the cavity is filled back up with brine. Part of their support comes from the brine or oil in them, each of which, of course, is much denser than air. Emptying them and not replacing oil by brine or water is strictly discouraged because these cavities are expensive, fragile structures that may collapse when emptied. The technical literature describes a number of large cavities in salt that have fully or partially collapsed.

  Relatively little is known about the strength of salt surrounding a cavity produced by solution mining. The Sterling decoupled and the Azgir partially decoupled explosions were conducted in cavities in salt formed by previous, much larger nuclear explosions. Much of the energy in a fully decoupled nuclear explosion goes into creating very high pressure in the cavity, about two hundred times atmospheric pressure at a depth of about 3000 feet (900 m). Less energy goes into forming seismic waves, which is why such an explosion is, in fact, decoupled. The walls of the cavities created by the two nuclear explosions were subjected to strong heating. Hence, the properties of the salt at and near those cavity walls are likely to be very different from the salt surrounding a cavity produced by solution mining.

  A weaker solution-mined cavity may well collapse or deform significantly if a decoupled test is conducted in it, considerations that do not seem to have been taken into account by L. A. Glenn of Livermore, William Leith of the U.S. Geological Survey, Larry Turnbull of the CIA, and others who have long maintained that an air-filled, solution-mined cavity in salt coul
d be used for decoupled nuclear testing with yields up to 10 kilotons.

  The Pre-Caspian depression, which contains the world’s largest concentration of salt domes, including those at Azgir, is now located mostly in the independent country of Kazakhstan. That salt extends into the Russian Republic near the mouth of the Volga River. Bedded salt—salt deposits that have not been deformed naturally into salt domes—is found to the north of Lake Baikal in the Russian Republic. Bedded salt typically is not as favorable as salt domes for constructing large cavities because the salt layers usually are interbedded with other sedimentary rocks.

  Another large area of salt domes is found near the northern coast of the Gulf of Mexico in the United States. Widespread salt deposits are also found in China and Iran, limited deposits in Pakistan, very limited quantities in India, and no known salt deposits in North Korea, a country of very old crustal rocks. Large volumes of water are needed to produce a solution-mined cavity in salt, but many parts of the Middle East and North Africa from Iran through Libya receive very little rainfall per year. Obtaining enough water for significant solution mining in those areas is highly problematic. Iran contains many large salt structures, but about half the country consists of deserts that receive less than 10 inches (25 cm) of rain per year. Higher rainfall is found in the Zagros Mountains of western Iran, but they are well monitored by seismic stations in adjacent countries.

  U.S. NUCLEAR EXPLOSIONS IN VERY SMALL CAVITIES

  Mill Yard, set off in 1986, and three other very small U.S. nuclear explosions were detonated in hemispherical cavities in soft rock at the Nevada Test Site with radii of about 36 feet (11 m). One publication estimated Mill Yard’s yield as about 20 tons (0.02 kt). It may have been decoupled but by an unknown amount. Purging the air in the tunnel used for Mill Yard released 5.9 curies of radioactive materials into the atmosphere, which would be easily detected today.