Silencing the Bomb Read online

  Tiny Tot, another test of very small yield in a hemispherical cavity, was detonated in granite at the Nevada Test Site (NTS) in 1965 to provide information on the effects on hard rock of a surface nuclear burst. The amount it was decoupled has not been published and may not be known. Data from such an early very small underground test are likely to have been sparse. In 2013 Australian physicist Christopher Wright reported that, starting fifteen minutes after the explosion, an uncontrolled release of predominantly noble gases, particularly xenon, emanated from the mouth of the Tiny Tot shaft.

  These very small explosions in hemispherical cavities apparently were not intended as tests of full decoupling, but they may well have been partially decoupled. These examples from Nevada indicate that the containment of bomb-produced radioactive products, especially noble gases, is problematic.

  CHEMICAL EXPLOSIONS IN CAVITIES IN SALT AND HARD ROCK

  Given the scarcity of data for decoupled nuclear explosions, a number of individuals and groups have attempted since 1959 to estimate decoupling factors using chemical explosions in cavities in salt and hard rock. Their explosive yields, however, ranged from less than a ton to about ten tons, much smaller than the 380-ton yield of the small U.S. Sterling decoupled nuclear explosion.

  Several of those experiments were poor for one or more of the following reasons: (1) chambers containing the explosive were not sealed and were open to the outside; (2) the explosives used were old munitions; (3) tamped (fully coupled) explosions were not included in the experiments as comparisons; and (4) the explosives were placed on the floor, not in the center of a cavity. Suspending several tons of explosive near the center of a cavity is not easy. The U.S. Cowboy chemical explosions in salt cavities in 1959 and Soviet chemical explosions in cavities in limestone in Kirghizia in 1960, with explosive yields up to six tons, are some of the better experiments. Salt is the only rock in which a decoupling factor as large as seventy has been obtained for chemical explosions in underground cavities.

  Hard rocks are much more common than salt. Constructing a large cavity at depth in hard rock, however, is much more difficult and expensive than one in salt. In 1995 F. A. Heuzé of the Livermore Lab and others examined hard rocks in terms of their suitability for decoupled testing. They found that a salient characteristic of hard rocks is that they are seldom massive, monolithic formations, but rather are penetrated by numerous cracks, faults, and other discontinuities, which may provide pathways for radioactive leakage from decoupled tests. To my knowledge, no one has either measured released xenon and other gases or determined if they can be contained for a decoupled nuclear explosion in hard rock. In addition, it is difficult to characterize how hard rock will respond to the strong shock and high pressure of a decoupled nuclear explosion based on traditional methods of laboratory tests on small rock samples.

  Heuzé and others stated that the igneous rock in which the Piledriver, Tiny Tot, and Hardhat nuclear explosions were conducted at the Nevada Test Site was not a granite of good quality. Joints in it are spaced about 8 inches (20 cm) apart. Joints spaced about 3 feet (1 m) characterize the hard rocks of the French test site in the Hoggar massif of southern Algeria. Hence, one or more joints could well leak radioactive products following a decoupled test in hard rock.

  In 1992 I attended a session on the containment of bomb-produced radioactive materials at an open meeting on decoupling. Most of the participants, many of whom were experts on containment, thought that a large cavity in hard rock would need to be extensively reinforced to prevent collapse when a nuclear explosion was detonated in it.

  CLAIMS OF EVASIVE DECOUPLED TESTING BY THE SOVIET UNION

  In 1995 Larry Turnbull of the CIA wrote that nuclear explosions had been conducted evasively by the Soviet Union in mines, one in 1972 on the Kola Peninsula and a second in the Ukraine on September 16, 1979. The first claim is clearly contradicted by published information, and the second is likely false.

  In his 1975 review of Soviet peaceful nuclear explosions (PNEs), Milo Nordyke of the Livermore Lab described a proposed ore-breaking project using a 1.8-kiloton PNE. A Soviet list contains a 2.1-kiloton explosion on September 4, 1972, on the Kola Peninsula in a well-known mining area. Forty-seven open stations recorded it with a magnitude of 4.6. All indications are that it was well coupled, not muffled. Soviet geophysicists did measure seismic amplitudes on either side of a slit cut into the rock, which may have led Turnbull to claim incorrectly that it was a decoupled test.

  In 1992 the New York Times reported a nuclear explosion of 1/3 kiloton at noon on September 16, 1979, in a Ukrainian mine. Sultanov and others list it as occurring in sandstone within a coal mine with a yield of 0.3 kiloton. Using that location, Frode Ringdal of Norway and Paul Richards of Lamont computed an origin time at noon Moscow time and a magnitude of 3.3 using signals received at the Norsar seismic array near Oslo. It would be even better recorded and located today. Its somewhat smaller magnitude for its yield is reasonably attributed to the explosion’s being conducted in soft rock, not to decoupling in hard rock.

  U.S. MEETINGS ON DECOUPLING IN 1996 AND 2001

  In 1996 I was invited to attend a classified meeting on clandestine nuclear testing organized by the U.S. Arms Control and Disarmament Agency (ACDA) and the Defense Special Weapons Agency (DSWA). DSWA was formerly the Defense Nuclear Agency and later became the Defense Threat Reduction Agency. ACDA was subsequently merged with the State Department.

  Because ACDA did not have enough funding for the 1996 meeting, DSWA provided funding, but in exchange it largely controlled the agenda. The meeting was mainly a forum for presentation of work by one consulting group, Jaycor, which worked under a contract from the predecessor to DSWA, the Defense Nuclear Agency. Jaycor included a number of former employees of DNA.

  Speakers from Jaycor claimed that important nuclear testing could be carried out evasively by Iran, Libya, and North Korea. The meeting, however, did not provide an opportunity to evaluate fairly the prospects for detecting evasive testing. Most of the rest of us not connected with Jaycor were only able to ask questions or make short remarks. Jaycor proposed a site in Iran, which they called “El Cheato,” and claimed that Iran could use it for clandestine nuclear testing. My sense was that Jaycor’s knowledge of evasive testing was poor. For some reason they did not use AFTAC’s classified information for that site, even though they had access to it. Calling a site in Iran El Cheato might have been considered cute if many people in the audience had not taken them seriously.

  In 2001 I participated in a forum at the secret level on decoupled testing organized by the State Department’s Bureau of Verification, Livermore Lab, and the Department of Energy. Several people who had worked on decoupling for many years, such as Lew Glenn of the Livermore Lab, claimed that large decreases in seismic wave amplitudes could be produced by large decoupled tests. Glenn, unrestrained by the moderator, interrupted me and other speakers repeatedly. William Leith of the U.S. Geological Survey reiterated his previous statements about the possibility of using very large holes in the ground for decoupled nuclear testing. With only three slides allotted to me, I chose to speak about the Azgir partially decoupled explosion of 1976.

  The State Department itself had no expertise on decoupled testing—neither its physical basis nor geological and containment constraints—until it hired Robert Nelson a decade later. Little was accomplished at that meeting to narrow the issues involved with evasive testing.

  I contend that decoupling is no longer a problem today for nuclear explosions of military significance. Huge cavities would be needed to fully decouple nuclear explosions with yields of one to a few kilotons. I return later to claims about decoupled nuclear testing during the Senate hearings in 1999 on the Comprehensive Test Ban Treaty and the 2012 Report by the National Academies.

  5

  U.S. OVERESTIMATION OF SIZES OF SOVIET UNDERGROUND EXPLOSIONS: 1961–1974

  One of the key issues that greatly intensified the concerns of many of us in the scie
ntific community was how the United States determined the yield—the size or energy release—of Soviet underground nuclear explosions. For many decades, the U.S. government calculated Soviet yields based on seismic data and yields from explosions detonated in Nevada, a procedure that led to a major overestimation of the yields of Soviet explosions.

  A factor in the yield question in the late 1950s and 1960s concerned how many earthquakes occurred in the USSR whose seismic magnitude equaled that of a given yield (size) of an underground test. These numbers were important at the time of the negotiations for a full test ban treaty in 1963 because the seismic signals of many small earthquakes could not be distinguished from those of small underground nuclear explosions.

  INCORRECT DETERMINATION OF YIELDS OF SOVIET UNDERGROUND EXPLOSIONS

  Two factors affected the determination of the yields of Soviet underground nuclear explosions:

  1. Seismic magnitudes vary with the types of rocks in which explosions of a given yield are detonated.

  2. Seismic magnitudes also vary because seismic P waves travel easily or poorly beneath various testing areas.

  As a result of these two factors, several values of yield can be calculated for the magnitude of a single underground explosion.

  I make use here of the seismic magnitude mb (Table 3.2) to calibrate the yield or energy release of Soviet underground nuclear explosions. Later I will use the surface wave magnitude Ms as well.

  VARIATION OF YIELD WITH ROCK TYPE

  Prior to the resumption of nuclear testing by the USSR in October 1961, the United States proposed prohibiting underground tests of seismic magnitude mb of 4.75 and larger. This magnitude is determined from the first-arriving, short-period P waves from either an explosion or an earthquake. At the time, the United States took an mb of 4.75 to be equivalent to a yield of about 15 to 20 kilotons. It was widely thought that the seismic signals from smaller explosions in the Soviet Union could not be distinguished from those of small earthquakes.

  Originally, however, U.S. estimations of Soviet yields from magnitudes were based solely on underground tests conducted in 1957 and 1958 in Nevada in tuff, a soft rock formed by consolidation of volcanic ash, and on data from a test site characterized by poor propagation of P-waves to large distances. Very small magnitudes were computed for those early U.S. underground nuclear explosions for their known yields, which led to biased and incorrect determinations of Soviet yields. It is understandable that the Soviet Union, with hard rocks at its test sites, was not interested in a ban on underground tests based on a magnitude threshold.

  Donald Springer of Livermore and colleagues indicated in 2002 that the United States tested nuclear explosions underground in a great variety of rock types at the Nevada Test Site (NTS) but overwhelmingly in two soft rocks, tuff and alluvium. The United States conducted very few tests at NTS in hard rocks like granite and only a few in rocks of intermediate strength like dolomite and rhyolite.

  Alluvium consists of unconsolidated sand and gravel. Unlike hard rocks, it often can be dug with a shovel and has very low seismic wave speeds. It is not a suitable material in which to construct a large cavity for decoupled (muffled) testing because a cavity in it would collapse quickly. U.S. nuclear explosions in alluvium were set off in two environments—dry and water saturated. Tests in dry alluvium generated the smallest seismic magnitudes of any rock type for a given yield. This is understandable because much energy is expended in the closure of air spaces between sand and gravel.

  Very few places in the world—the Nevada Test Site (NTS) and Namibia, but none in Russia—have thicknesses of dry alluvium suitable for conducting tests larger than one or two kilotons. Figure 5.1 shows many collapsed craters in alluvium in Yucca Valley, the site of many U.S. tests within NTS. Those craters were (and still are) easily visible on satellite imagery.

  FIGURE 5.1

  Aerial view of craters produced by underground nuclear explosions in the Yucca Flat portion of the Nevada Test Site. View is looking from south to north.

  Source: Springer and colleagues, 2002.

  The United States conducted two tests of about one kiloton, called Unde and Ess, in alluvium in 1951 and 1955. Their depth of burial, 16 and 62 feet (5 and 19 m), was so shallow, however, that they produced subsidence craters at the surface. The much-discussed Rainer test in 1957 was detonated at a greater depth of 900 feet (274 m) in a tunnel in tuff. Blanca in 1958, with a yield of 22 kilotons, was detonated at 990 feet (301 m) in very soft tuff that had a low seismic wave speed and low strength. Springer and others indicated that the cavity it produced collapsed twelve seconds later. Tuff clearly is also not appropriate for the construction of a large cavity at depth in the Earth.

  When the United States resumed testing in late 1961, it detonated the nuclear explosion Fisher of 13.4 kilotons in dry alluvium at the Yucca testing area at a depth of 1194 feet (364 m). It too produced a collapsed cavity at the surface. In 2009 Carl Romney of the Department of Defense reported that Fisher, when adjusted for differences in yield, was forty times smaller in seismic amplitude than the underground explosion Gnome in salt in New Mexico. Clearly, for the same yield, large differences in the seismic magnitude mb were associated with explosions conducted in different rock types.

  In 1974 seismologist Romney of the Defense Department urged the United States government to base the maximum size of underground tests under the Threshold Test Ban Treaty (TTBT) on seismic magnitude. Yield, not magnitude, however, was accepted as the threshold in 1974 by the two parties to the treaty—the United States and the Soviet Union. Soviet negotiators had argued that yield, or energy release, was a physical quantity, whereas magnitude was not. That was correct. If magnitude had been the threshold, it would have allowed the United States to conduct explosions of much larger yield at the Nevada Test Site in tuff or alluvium than the Soviet Union could do at their two test sites in old, much harder rock.

  VARIATION OF SEISMIC MAGNITUDE WITH PROPAGATION OF SEISMIC P WAVES IN THE UPPER MANTLE OF THE EARTH

  From 1958 through 1988, the U.S. government did not take into account a second important factor that affects the measured seismic magnitude, mb, of underground explosions of the same yield at the Nevada Test Site and the two main Soviet test sites. Seismic waves were not absorbed as much beneath those Soviet test sites at depths of 30 to 125 miles (50 to 200 km) as they were beneath the Nevada Test Site. The amount of absorption is related to temperatures at those depths. NTS experienced much younger volcanism and its associated heating than the geologically older rocks beneath the two main Soviet test sites. Soviet scientists had presented data on chemical explosions of several kilotons that indicated larger magnitudes than those the United States had measured for explosions of the same size in Nevada. Still the U.S. government ignored those results.

  For more than a decade, Romney and Eugene Herrin of Southern Methodist University, both seismologists, advocated the use of an incorrect method to calculate yields of Soviet underground explosions. Based on known yields and seismic magnitudes, mb, of explosions at NTS, the magnitudes of Soviet explosions were used to calculate unknown Soviet yields. U.S. estimates of Soviet yields were too large not only because of differences in rock type but also because the propagation of P waves beneath Soviet test sites was very efficient. These differences are illustrated in figure 5.2. P waves crossing parts of the upper mantle of the Earth, called the asthenosphere, where temperatures are high, become reduced in amplitude.

  FIGURE 5.2

  Illustration of differences in sizes (amplitudes) of seismic P waves for underground nuclear explosions of the same yield at Soviet and U.S. test sites. Seismic P waves from explosions at the Soviet Union’s two main test sites are much larger than P waves from an explosion of the same yield in Nevada.

  Source: Office of Technology Assessment, 1988.

  During the Reagan administration, accusations by the U.S. government that the Soviet Union was cheating on the Threshold Test Ban Treaty were based on these incorrect calibrations
and indirectly on the belief that the Russians could not be trusted. The U.S. yield estimation procedures were classified “secret” for decades. Hence, those lacking the appropriate clearances could not judge those yield calculations, making it difficult to prove that Romney’s and Herrin’s methods were erroneous. British and many U.S. scientists, including me, who worked on test ban verification, did not agree with the U.S. procedures for determining Soviet yields. Incorrect calculations also led to claims in the United States that the yields of deployed Soviet weapons were larger than they actually were by a factor of about three times.

  A variety of evidence indicated that the Earth’s uppermost mantle beneath Nevada differed from regions of older geology such as those found beneath the two main Soviet test sites and beneath the central and eastern United States. I described earlier this effect for the Gnome and Salmon explosions in New Mexico and Mississippi. In 1967 Jack Evernden, then at the Air Force Technical Application Center (AFTAC), used many observations of seismic waves to show that P-wave propagation and P-wave speeds were very different beneath Nevada and the central and eastern parts of the United States. Evernden’s work and the observations from Gnome and Salmon should have been a warning not to use data from Nevada to calibrate yields of explosions at Soviet test sites.

  In his 2009 book, Romney accepted major differences between the propagation of seismic waves in the western United States and regions of older crust and uppermost mantle. Nevertheless, I know he was an ardent foe of this view for many decades, especially at many classified meetings. The control of classified seismological data by a few persons, such as Romney and Herrin, prevented a resolution of this issue between 1974 and 1988. Because I was very involved in those deliberations, at both the classified and open levels, I knew their views well throughout that long debate.

  Romney also stated that, unknown to seismologists outside AFTAC, the Soviets fired their first underground nuclear explosion on October 11, 1961. He went on to say that six open seismic stations reported signals to the U.S. Coast and Geodetic Survey (USCGS) for their unclassified estimate of location and depth for the 1961 event. Romney says the USCGS location was near the Eastern Kazakhstan test site (which was correct) but their depth estimate of 19 miles (31 km) was not. USCGS, a government agency, was not permitted to state whether the event was an explosion or an earthquake, which it did not.