APPENDIX E1

Dealing with Evasive Underground Nuclear Testing

Several schemes for clandestine testing have been proposed since 1959 when the concept of decoupling was first introduced (i.e., setting off nuclear explosions in large, deeply-buried underground chambers).

In the last 10 years, the deployment of sensitive seismographs and arrays—especially in a swath across China, Russia and other former (now independent) states of the Soviet Union, the Middle East, and other parts of Asia—provides an abundance of data that was not available previously to the United States. Access to seismic data at high frequencies and closer distances and for much smaller events makes the two most serious cheating scenarios—decoupled nuclear testing and mine masking—easier to detect and identify. This appendix describes the obstacles facing a country that wants to conduct a clandestine test in secret and yet not have one or several aspects detected by either the United States or other countries. It addresses whether decoupled testing can be done at yields of possible military significance by countries of special concern to the United States. Whereas detection capabilities have increased greatly during the last 10 years, relatively little is new in computer modeling of decoupling and in experiments with chemical explosives.

Decoupled Testing

If a nuclear explosion is detonated in a large cavity deep underground, the seismic waves generated can be reduced in amplitude. This method of evasion is called decoupling, and the amount of seismic-wave reduction is called the decoupling factor (DF). For a fully2 decoupled nuclear explosion, most of the explosive energy goes into increasing the gas pressure in the cavity by as much as 100 times atmospheric pressure.3 This is in contrast to a normal “well coupled” underground explosion where much of the energy goes into melting and deforming the surrounding rock and in generating larger seismic waves.

Although the concept of decoupling was proposed 50 years ago, data on decoupled nuclear explosions are very sparse and mostly 25 to 50 years old. This is surprising because decoupling as a cheating scenario has been mentioned repeatedly for 50 years as one of the main impediments to verifying a treaty that would ban underground testing. In the era of nuclear testing before September 1996, when the United States, Russia, the United Kingdom, France, and China signed the CTBT and began a moratorium, nuclear experiments by the United States to test this scenario were not considered important enough for them to be given priority and financial support except at very small yields.

A country wanting to conduct a clandestine explosion as large as 1 to 5 kilotons with a high probability that it would not be detected would have to meet all of the many criteria described below. The greater the amount of equipment deployed before, during, and after a

__________________

1 This appendix is drawn from the work of the Seismology Subcommittee (see Appendix A), which was directed by the main committee to investigate issues related to seismology and evasion.

2 The word fully simply means that rocks subjected to explosive energy in a decoupled test are not stressed beyond their elastic limit . Fully decoupled does not mean seismic waves are reduced to zero amplitude. The decoupling factor is not increased by yet greater enlargement of the cavity.

3 Computed for 1 kt explosion in a cavity in salt of radius of about 25 m at a depth of 1 km as scaled from data of the Sterling decoupled nuclear explosion (Denny and Goodman, 1990).



The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 161
APPENDIX E1 Dealing with Evasive Underground Nuclear Testing Several schemes for clandestine testing have been proposed since 1959 when the concept of decoupling was first introduced (i.e., setting off nuclear explosions in large, deeply- buried underground chambers). In the last 10 years, the deployment of sensitive seismographs and arrays—especially in a swath across China, Russia and other former (now independent) states of the Soviet Union, the Middle East, and other parts of Asia—provides an abundance of data that was not available previously to the United States. Access to seismic data at high frequencies and closer distances and for much smaller events makes the two most serious cheating scenarios—decoupled nuclear testing and mine masking—easier to detect and identify. This appendix describes the obstacles facing a country that wants to conduct a clandestine test in secret and yet not have one or several aspects detected by either the United States or other countries. It addresses whether decoupled testing can be done at yields of possible military significance by countries of special concern to the United States. Whereas detection capabilities have increased greatly during the last 10 years, relatively little is new in computer modeling of decoupling and in experiments with chemical explosives. Decoupled Testing If a nuclear explosion is detonated in a large cavity deep underground, the seismic waves generated can be reduced in amplitude. This method of evasion is called decoupling, and the amount of seismic-wave reduction is called the decoupling factor (DF). For a fully2 decoupled nuclear explosion, most of the explosive energy goes into increasing the gas pressure in the cavity by as much as 100 times atmospheric pressure.3 This is in contrast to a normal “well coupled” underground explosion where much of the energy goes into melting and deforming the surrounding rock and in generating larger seismic waves. Although the concept of decoupling was proposed 50 years ago, data on decoupled nuclear explosions are very sparse and mostly 25 to 50 years old. This is surprising because decoupling as a cheating scenario has been mentioned repeatedly for 50 years as one of the main impediments to verifying a treaty that would ban underground testing. In the era of nuclear testing before September 1996, when the United States, Russia, the United Kingdom, France, and China signed the CTBT and began a moratorium, nuclear experiments by the United States to test this scenario were not considered important enough for them to be given priority and financial support except at very small yields. A country wanting to conduct a clandestine explosion as large as 1 to 5 kilotons with a high probability that it would not be detected would have to meet all of the many criteria described below. The greater the amount of equipment deployed before, during, and after a 1 This appendix is drawn from the work of the Seismology Subcommittee (see Appendix A), which was directed by the main committee to investigate issues related to seismology and evasion. 2 The word fully simply means that rocks subjected to explosive energy in a decoupled test are not stressed beyond their elastic limit. Fully decoupled does not mean seismic waves are reduced to zero amplitude. The decoupling factor is not increased by yet greater enlargement of the cavity. 3 Computed for 1 kt explosion in a cavity in salt of radius of about 25 m at a depth of 1 km as scaled from data of the Sterling decoupled nuclear explosion (Denny and Goodman, 1990). 161

OCR for page 161
162 The CTBT- Technical Issues for the U.S. clandestine test, the greater the chance that it would be detected. This is even more true for a series of tests of a new nuclear device of military significance.4 Strong views exist today as they have for 50 years—both pro and con—of the feasibility of conducting a secret decoupled explosion of significant yield. In general, experts agree that seismic signals from an underground nuclear explosion can be reduced by a large amount but that the technique is impractical for yields above 10 kilotons (Turnbull, 2002). In this appendix, the Seismology Subcommittee argues that decoupled testing with yields of 1 to 10 kilotons with decoupling factors of 50 to 100 is not credible for countries of concern to the United States and that such tests likely would be detected with present monitoring capabilities. The following sections discuss three decoupled nuclear explosions and what can be learned from them, cavities created by past nuclear explosions in salt that might be used for future clandestine testing, use of large cavities in thick salt deposits, and testing in mined cavities in hard rock. Salt is emphasized because cavities likely exist in that material from past nuclear explosions in the Former Soviet Union, and very large cavities at depth are easiest to construct in salt. Thick salt deposits at suitable depths for decoupled testing exist in some countries but not in others, such as North Korea. The feasibility of evasive testing is very much a function of the size or yield of an explosion a country wishes to test. Finally, this appendix lists the several significant obstacles a country would face in deciding to conduct a decoupled test and have a high likelihood that it would not be detected. Known decoupled nuclear explosions The database of decoupled nuclear explosions is very meager. It includes the only one that was nearly fully decoupled (Sterling), one partially decoupled (Azgir), and one small U.S. nuclear explosion that may have been decoupled significantly but by an unknown amount. They are the following: • Sterling, a 0.38 kiloton (380 ton) nuclear explosion, was detonated in 1966 in the cavity generated by the 5.3 kiloton fully-coupled Salmon nuclear explosion in a salt dome in Mississippi. A decoupling factor of about 72 + 8 was calculated. A factor of about 70 has occurred repeatedly since then in discussions about evasive testing. Assertions such as “This means a 70 kiloton test can be made to look like a 1-kiloton test, which the CTBT monitoring system will not be able to detect” is doubly false, in that a 70 kt explosion cannot be fully decoupled, and the IMS will confidently detect a signal produced by a 1 kt test. What Sterling showed, in fact, was that a 0.38-kiloton test could be decoupled by a factor of about 70. Detection and identification, which we address in this report, have improved immensely since 1966. Both decoupling and detection of larger decoupled explosions are discussed. • In 1976, the Soviet Union conducted a partially decoupled nuclear explosion of 8 to 10 kilotons in a huge cavity of mean diameter of 243 feet (74 m) in a salt dome at Azgir, which is now in the Republic of Kazakhstan (Sykes, 1996; Sultanov et al., 1999; Murphy, 2009). That cavity had been created at a depth of nearly 3,000 feet (1,000 meters) by a well-coupled nuclear explosion in 1971 with a yield of 64 kilotons (magnitude 6.06). Even in 1976—prior to the subsequent increase in deployed seismic instruments—that event was well recorded by many stations in Europe and Asia and as far away as Canada with 4 This appendix emphasizes decoupling factors larger than 3, since reduction in seismic amplitudes at about this level can be obtained by testing in regions where either seismic waves propagate less efficiently, explosions are detonated at greater depths than past nuclear weapons tests or in weak rock geologies.

OCR for page 161
163 Appendix E: Evasive Underground Nuclear Testing magnitude 4.06 (Sykes, 1996). It was decoupled by a factor (DF) of 12 to 15 times.5 According to news reports at the time, it was promptly identified as originating from a decoupled nuclear test. Because the yield of the 1976 explosion was more than 20 times that of Sterling, its data are crucial to arguments about the detection and identification of decoupled nuclear explosions larger than 1 kiloton. Using the above data for the 1971 and 1976 Azgir explosions, magnitudes of 2.4 and 3.4 are obtained for fully decoupled nuclear explosions (DF = 70) of 1 and 10 kilotons at the same depth in salt at Azgir. • Mill Yard was a U.S. nuclear explosion detonated in 1985 at the Nevada Test Site (NTS) in soft rock in a hemispherical cavity of radius 32 feet (11 m) (Garbin, 1986; Murphy, 2009). The amount it was decoupled is uncertain (Stevens et al., 1991, Sykes, 1996, Murphy, 2009). A cavity suitable for full decoupling of 1 kiloton in the same rock type would have to be much larger (about 40-m radius), a significant engineering achievement compared to the mining of the Mill Yard cavity (a factor of 96 in volume). Purging of the Mill Yard tunnel released 5.9 curies of radioactive material into the atmosphere (OTA, 1989 p.4). Two other explosions in cavities of the same size but of somewhat greater yield may have been decoupled but by small amounts (Stevens et al., 1991). Computer calculations of decoupling Computer codes have been used over the past 50 years in attempts to estimate decoupling factors (DF) for cavities of different size for explosions in salt and hard rock (e.g., Stevens et al., 1991; Glen and Goldstein, 1994; Murphy, 1996; 2009). The Seismology Subcommittee judged that some of those calculations for salt have overestimated DF for fully decoupled nuclear explosions and all of them for partially decoupled tests. Just prior to the Sterling test in 1966, a DF of about 125 was computed for it, whereas 72 was observed (Figure E-1). More recent code calculations that were constrained to fit the decoupling factor of 72 for Sterling (Glen and Goldstein, 1994; Murphy, 2009) give a DF of 50 to 65 for the 8-10 kiloton partially decoupled explosion at Azgir. However, actual observations of DF for it at large distances (Figure E-1) are significantly smaller–12 to 15. The estimates of DF made at close distances in Figure E-1 have a much larger uncertainty but still fall well below the two code calculations. 5 The details of the decoupling factor calculation are given in Sykes (1996). Note that the magnitude of these Azgir events are slightly smaller (by ~0.2 magnitude units) than predicted by the standard magnitude-yield formula given on page 144 due to the greater than normal depth of the events. However, this depth effect affects both the 1971 and 1976 event equally, so the magnitude-yield formula can still be used to determine the partial decoupling factor from the difference between measured magnitudes of the two events given their announced yields. See Sykes (1996) for further discussion.

OCR for page 161
164 The CTBT- Technical Issues for the U.S. FIGURE E-1: Decoupling factor (DF) for the Sterling and Azgir nuclear explosions in salt, and code calculations by Glen and Goldstein (1994) and Murphy (2009) as a function of the yield divided by the maximum yield allowed according to the Latter criterion for containment. That criterion states that the pressure on the cavity wall should be no larger than 0.5 ρgh, where ρ is the average rock density, g is gravitational acceleration, and h is cavity depth. Because most rocks are weak in tension and strong in compression, the criterion is such as to keep the rock surrounding a cavity in compression during and soon after an explosion. Full decoupling involves keeping the surrounding rock in the elastic domain so that no permanent deformation occurs. The uncertainties (one standard deviation) in the decoupling factors for the Azgir partially decoupled explosion obtained at stations at distances between 785 and 2,760 km (491 and 1,725 mi) (Sykes, 1996) are comparable to the size of the two red squares, whereas they are much larger for the closer measurements between 2 and 110 km (1.2 and 69 mi), as shown by the vertical black bars (Glen and Goldstein, 1994). SOURCE: Adapted from Glen and Goldstein, 1994; Murphy, 2009; and Sykes, 1996 Observed decoupling factors for the Cowboy chemical explosions in salt in 1959 also are smaller than those from code calculations (Murphy, 2009). These overestimates are likely attributable to the properties of salt in the region surrounding the cavity being different from those derived from measurements of physical properties either on a laboratory scale or on salt that has not been subjected to a previous nuclear shock or to solution mining. Different decoupling factors as a function of cavity size for several rheological models of the strength of the salt surrounding the Salmon and Azgir cavities have been calculated by Glen and Goldstein (1994) and Murphy (1996, 2009). The uncertainty is expressed by the sub-title of Goldstein and Glen’s (1993) “Simulation is easy. Prediction is difficult!”

OCR for page 161
165 Appendix E: Evasive Underground Nuclear Testing In their thorough re-examination of the data from Salmon and Sterling, Denny and Goodman (1990) find that the Sterling explosion was nearly fully decoupled, even though its yield exceeded the Latter criterion by 1.8 times (0.9 times the overburden pressure). They conclude that decoupling factors larger than about 70 could not be obtained by further increase in cavity size. It is clear from the data in Figure E-1 that the decoupling factor for explosions in salt drops precipitously between 1 and 2 times the overburden pressure (2 to 4 times the Latter criterion). Denny and Goodman also find that nonlinear effects of the Sterling explosion extended outward to about 80 m, much larger than its cavity radius of 16.7 m, but that the seismic corner frequency was closely related to cavity size. Hence, more attention needs to be paid to the actual data on cavity size and decoupling factors for the Sterling and Azgir explosions (Figure E-1) and to the analyses of Denny and Goodman than to the more vague concept of “an elastic radius.” Claims such as that by Stevens et al. (1991) that very large decoupling factors can be obtained for explosions that are overdriven6 by large amounts in salt compared with those for Sterling, are not supported by the data of the 1976 Azgir nuclear explosion and the 1961 Cowboy chemical explosions. Murphy (2009) acknowledges this misfit for overdriven explosions whose yield is too large for the size of its cavity (i.e., the pressure on the cavity wall exceeds the Latter criterion). A prospective cheater relying on the single data point for the small Sterling explosion and code calculations for salt either would be forced to make very conservative assumptions about decoupling or forego that mode of clandestine testing. Decoupling factor at high frequency Decoupling leads to a seismic signal that looks like that of a smaller explosion. So, for example, a 10 kt explosion looks basically like a fully coupled explosion of 150 tons using a decoupling factor of 70, if such large cavities could be achieved at depth. Figure E-2 compares the computed amplitudes (on the vertical axis) as a function of frequency (on the horizontal axis) of coupled and fully decoupled nuclear explosions, each with yields of 10 kilotons. The two spectra are nearly flat (horizontal) at low frequencies and then drop off (decay) at frequencies higher than what is called the “corner frequency.” The corner frequencies in this case are 2 Hz for the coupled explosion and 12.5 Hz for the decoupled event; lower-yield explosions will have higher corner frequencies, as do earthquakes. 6 An overdriven explosion is one for which the pressure on the cavity wall places the surrounding rock, in this case salt, in tension. It results in partial decoupling.

OCR for page 161
166 The CTBT- Technical Issues for the U.S. FIGURE E-2: Comparison of computed seismic source functions for various frequencies (f) in Hz (cycles per second). W = yield and H = depth. SOURCE: Murphy and Barker, 1995 FIGURE E-3: Decoupling factor (DF) as a function of frequency for the fully decoupled explosion in Figure E-2 compared with that of the well-coupled explosion. W = yield and H = depth. SOURCE: Murphy and Barker, 1995 Figure E-3 compares the amplitude ratio (i.e., the decoupling factor, DF) for the decoupled and coupled calculations shown in Figure E-2. DF is about 70 at low frequency and drops to about 15 at a frequency of about 10 Hz (cycles per second). For explosions of 1 kt, the drop to a DF of 30 would occur at about 15 Hz and to a DF of 10 at about 25 Hz. Scaling the corner frequency of Sterling to 1 kiloton gives 25 Hz (Denny and Goodman, 1990). Modern seismic instruments, especially those at regional distances from seismic sources, record high-frequency signals. Frequencies up to 20 Hz are often recorded for regions characterized by efficient propagation of seismic waves—such as most of Russia, the areas near the Chinese and Indian test sites, and all of North Korea. Hence, for areas characterized by efficient propagation of seismic waves of high frequencies, the detection of decoupled tests of 1 kiloton and larger can

OCR for page 161
167 Appendix E: Evasive Underground Nuclear Testing be accomplished because a well coupled test whose yield is 1 kiloton divided by 70 (that is, about 15 tons) can be detected in such regions. In fact, it may not be necessary to go to frequencies above the corner frequency where the DF is less in order to see decoupled explosions if the equivalent fully-coupled explosion is detectable at that yield. For example, as discussed in Appendix D, a 15-ton shot (1kt decoupled by 70) is detectable at Novaya Zemlya using the IMS. Kim and Richards (2007) describe detections, across a broad band of frequencies, of chemical explosions whose yield was just one or two tons, at distances of about 300 km (188 mi) in northeast China just to the north of North Korea. Decoupled testing in existing cavities created by past explosions in salt Both the decoupled 1966 Sterling and the partially decoupled Azgir nuclear explosions were detonated in cavities in salt domes created by well-coupled nuclear explosions. Salt is one of the few Earth materials in which a cavity produced by a nuclear explosion is not likely to collapse on a time scale of months to years. Nevertheless, it takes a fully coupled explosion about 14 to 20 times larger to create a cavity in salt suitable for conducting a subsequent, fully decoupled explosion in it. Hence, a past explosion in salt of 1 kiloton is suitable for conducting only a fully decoupled test of 0.07 kilotons (70 tons) or smaller. While the U.S.S.R. conducted a number of fully coupled nuclear explosions in thick salt deposits (OTA, 1989; Sultanov et al., 1999), all of the possible sites of cavity-producing explosions in salt of 2 kilotons or larger are known and can be monitored readily. All are located in regions of very low earthquake activity. Thus, a seismic event at or near one of those sites would be suspicious immediately and receive intense scrutiny. All 8 cavities created by past Soviet explosions in salt that are suitable for fully-decoupled explosions of one kiloton (and up to a maximum of 4.2 kilotons) are located either at Azgir in Kazakhstan or Bukhara in Uzbekistan (Sykes, 1996). All explosion-produced cavities in salt in the Russian Republic are suitable only for fully decoupled tests of 0.5 to 0.9 kilotons or smaller. Hence, the breakup of the Soviet Union greatly limited possible opportunities for conducting future decoupled nuclear explosions by Russia in explosion-produced cavities because of the exclusion of regions in Central Asia. Testing in cavities in salt constructed by either solution or conventional mining Turnbull (1995, 2002), Leith (2001), and others propose that large cavities could be used for decoupled testing with yields much larger than 1 kiloton. It is generally agreed that to fully decouple a 5-kiloton explosion in salt at a depth of about 3,000 feet (900 meters) a spherical cavity with a diameter of about 240 feet (73 meters) is required. Obviously, this would entail a substantial construction effort. Very large cavities have been created in salt domes by conventional and solution mining (Berest and Minh, 1981; Sykes, 1996; Leith, 2001), mainly for storage of gas, oil and toxic waste. Salt domes, sometimes called salt diapirs, are large bulbous geological structures consisting mainly of the mineral halite (sodium chloride) but often with up to 5 to 10 percent of other minerals (Leith, 2001). Salt domes are known traps for petroleum. The economic value of salt deposits has led to their being mapped extensively and either described in the open literature (e.g., Zharkov, 1984) or held as proprietary data by petroleum companies. The Pre-Caspian depression, which contains the world’s largest known concentration of salt domes, is located mainly in Kazakhstan but extends into adjacent parts of Russia. Another huge area of salt domes is found along the Gulf coast of the United States. In terms of countries of proliferation concern with respect to evasive testing, Leith (2001) reports other widespread salt deposits in China and Iran, limited deposits in Pakistan (Davis and Sykes, 1999), very

OCR for page 161
168 The CTBT- Technical Issues for the U.S. limited quantities in India and Israel, and no known salt deposits in North Korea, a country of very old crustal rocks. Significant thicknesses of salt for decoupled testing are not present at the following test sites: Novaya Zemlya, Russia; Lop Nor, China; Pokharan, India; Pakistan (two sites); and France’s now closed test site in the Pacific. Thick beds of salt that have not been deformed into salt domes are known in other parts of Russia, especially to the north of Lake Baikal, and in some of the now independent countries of the former Soviet Union (Zharkov, 1984; OTA, 1989). Bedded salt, however, is generally not as favorable for the construction of very large underground cavities because salt is typically inter- bedded with other rocks such as dolomite, anhydrite, gypsum, limestone and sandstone and sometimes with weak layers of potash. Salt domes typically contain fewer of these sedimentary rocks, making them more suitable for construction of very large cavities. Solution mining is the least expensive method for either mining salt itself or forming large cavities for other uses. In solution mining, fresh water is pumped in a pipe that extends from the surface into a salt formation at depth. A mixture of salt and water (brine) is pumped back out in another pipe. This avoids constructing tunnels and shafts as in conventional mining. The U.S. Strategic Petroleum Reserve is stored in large cavities created by solution mining at Bryan Mound, Louisiana. Large pumps, tanks and other equipment are quite visible at that site. Brine in very large cavities created by solution mining has rarely been pumped out and replaced by air. An explosion in a brine-filled cavity would be well coupled, not decoupled. Emptying a large commercial cavity of this type without replacing its content (e.g., oil or pressurized gas) by brine or seawater is strictly discouraged or forbidden because these expensive and fragile structures may collapse. Relatively little is known about the strength of salt near the wall of a cavity formed by solution mining. Hence, extrapolating the properties of salt in the region surrounding a cavity formed by a nuclear explosion such as Salmon to a solution-mined cavity is uncertain. Disposal of brine is a major problem because a rule of thumb in the industry is that the solution mining of 1 cubic foot of salt requires about 7 cubic feet of injected fresh water (Leith, 2001). In some locations, the fresh water could be pumped from an aquifer, and the resulting brine pumped into another aquifer perhaps at greater depth, thus avoiding large surface flows of water and brine. Having enough fresh water to construct a very large cavity by solution mining would be a major problem for salt deposits in arid parts of Iran, the Middle East, northwestern China, Pakistan and India.

OCR for page 161
169 Appendix E: Evasive Underground Nuclear Testing FIGURE E-4: Stability range for an air-filled cavity in salt. Depth range is bounded at the shallow end by need for containment of a decoupled nuclear explosion and at the deep end by the need to keep the cavity open long enough to conduct explosion. Diagonal lines indicate minimum cavity radius (in meters) required for full decoupling. A given size of cavity is suitable only for smaller yields as depth decreases. SOURCE: Modified from Davis and Sykes, 1999 Unlike most rocks, salt deforms at relatively shallow depths in the crust of the earth, which limits the depth at which cavities can be constructed for clandestine testing (Figure E-4). Air filled cavities in salt typically are not stable at depths greater than about 2,900 to 4,200 feet (880 to 1,280 meters) (Berest and Minh, 1981). Those depths are a function of the temperature gradient in the Earth and the presence of other minerals including small amounts of water. Cavities in salt cannot be shallower than that needed to ensure containment of decoupled explosions. The 1966 Salmon and the 1976 Azgir explosions were detonated at nearly optimum depth, about 3,000 feet (about 900 meters), so as to insure both containment and cavity stability. Larger cavities would have been needed to ensure containment if those explosions had been conducted at shallower depths. Leith (2001) references large cavities created by solution mining in the former Soviet Union and Germany at depths to 200 feet (60 meters). Though large, their very shallow depth would not ensure containment for a decoupled explosion larger than a very small fraction of a kiloton. When the United States was testing in Nevada, containment policy since 1970 was to detonate explosions of any yield at depths of at least 600 feet (183 meters) (OTA, 1989). A number of very large cavities in salt at depths greater than about 3,000 feet (900 meters) have either totally or partly collapsed (Sykes, 1996; Leith, 2001). Fifteen nuclear explosions with yields of about 3 to 15 kilotons were conducted during the 1980s at depths of 3,000 to 3,600 feet (900 to 1,100 meters) in bedded salt to the north of Astrakhan, Russia, near the mouth of the Volga River. All were intended to create cavities for the storage of gas condensates from nearby gas fields. Most, and perhaps all, of the cavities had completely closed within several years of the explosions that created them.7 7 Written communication to L. Sykes from W. Leith. September 7, 2001.

OCR for page 161
170 The CTBT- Technical Issues for the U.S. Salt domes are commonly capped by sedimentary rocks up to a few hundred meters thick. Most cap rocks are highly fractured (Leith, 2001) from the deformation that results from the formation of salt domes. Ground water often circulates in cap rocks at shallow depths. Rock salt (halite) is often considered to be a viscous material that never deforms by brittle faulting. Fractures and faults, however, are observed very locally in some salt domes (Davison, 2009). A potential evader would need to select a salt dome whose top and deformed cap rocks are well below the surface. Earthquakes are very rare in most areas of thick salt deposits, including those in Russia, the Ukraine, the Pre-Caspian depression of Kazakhstan and the Gulf coast of the U.S. Large chemical explosions are not used in salt mining. Hence, the near absence of those seismic events makes monitoring of those areas for clandestine testing relatively easy. Earthquakes do occur in the vicinity of a number of salt deposits of Iran. Much is known about the geology of Iran, including the locations of salt deposits, from petroleum exploration and other geologic mapping over the last century. Iran, a moderate-size country, can be monitored using seismic stations in several surrounding countries. One decoupling option would be to use an abandoned salt mine for decoupled testing. Such mines, however, typically have exploratory drill holes and other openings, all of which must be known and sealed before a nuclear-explosion test so that they are not conduits for radioactive materials to the surface. An early plan for permanent disposal of radioactive waste from commercial nuclear reactors in an abandoned salt mine near Lyons, Kansas, was cancelled in part because all previous boreholes had not been cataloged and containment could not be assured. No decoupled nuclear explosions are known to have been conducted in cavities created in salt by either solution or conventional mining. Hence, a country wanting to conduct a clandestine test in such a manner would have to be wary about containment of bomb-produced radioactive isotopes, cavity stability, and detection by the much-increased capabilities of seismology and by the various national technical means available to the United States. The very small levels of earthquake activity in most areas of thick salt deposits make them relatively easy to monitor. Only the United States and Russia are known to have conducted decoupled tests. The capabilities of other countries to conduct decoupled tests in total secrecy must be questioned given their lesser testing experience (if any), their limited testing in a variety of rock types and less experience with containment. Testing in cavities in hard rock Hard rock is much more abundant globally than are thick salt deposits. Lacking salt, it is the medium in which construction of large cavities would have to be undertaken for clandestine testing by North Korea and at the Russian, Chinese, Pakistani, and Indian test sites. Mining of large cavities in hard rock, however, is much more difficult and expensive than is cavity construction in salt. Secret disposal of the material excavated is likely to be more difficult compared with the disposal of salt brine. The volumes and diameters of the largest existing cavities in hard rock are much smaller than those formed by solution mining of salt. Cavities of the size used for the tiny Mill Yard explosion are far smaller than those needed for full decoupling of explosions of even a fraction of a kiloton in hard and soft rock. Many engineering reports on the construction of large underground openings emphasize that hard rock masses are seldom monolithic but are penetrated by numerous joints, faults and other discontinuities on many length scales. Traditional continuum codes (computer simulations) are not sufficient for simulating dynamic block motion for underground nuclear explosions in such media (Heuzé et al., 1991). The failure of codes to fit both data points for the two decoupled explosions in salt—Sterling in 1966 and Azgir in 1976 (see Figure E-1)—indicates

OCR for page 161
171 Appendix E: Evasive Underground Nuclear Testing that similar calculations for hard rock are likely to be more unreliable. Faults and joints on the scale of 10 to 1,000 meters (30 to 3,000 feet) also present a major problem for containment of radioactive products produced by a nuclear explosion, especially for the containment of noble gases like xenon. Faults and cracks in hard rock (in contract to salt) do not heal after a nuclear explosion. Containment methods perfected for fully-coupled explosions are largely irrelevant for decoupled explosions. The rock is not liquefied by the shockwave, and leaks are far more likely. Construction of spherical cavities in hard rock that are large enough for full decoupling of explosions over 1 kt is expensive and requires technological sophistication not widely available (Leith, 2001). Leith’s estimate, however, assumes a 20 meter (66 foot) radius for full decoupling of 1 kt in hard rock at a depth of about 3,000 feet. Murphy (2009) now concludes that the required radius is nearly identical for both salt and hard rock. Using 25 rather than 20 meters, which we identified earlier for salt, results in a volume increase of a factor of two and hence in the above construction results in difficulty starting at 0.5 rather than 1 kt. Large non-spherical cavities in hard rock of the same volume are easier to construct and have been proposed for clandestine testing (Stevens et al., 1991; Leith, 2001; Murphy, 2009). The surface area of non-spherical cavities, however, is greater than that of a sphere of the same volume. Hence, the chance that more faults, cracks and joints would be encountered at the surface of a non-spherical cavity increases the chance that radionuclides could escape and be detected. The shortest dimension of non-spherical openings in hard rock will experience the largest non-elastic pressure pulse compared with that experienced on the wall of a fully decoupled nuclear explosion of the same size in a spherical cavity. A number of the large cavities in hard rock described by Leith (2001) are too shallow for containment to be assured except for an explosion with a yield of a small fraction of a kiloton. One example is the underground skating arena in hard rock created in Norway for the 1994 Winter Olympics. Whereas that arena has an unsupported span of 200 feet (61 meters), the depth below the surface of its top is only 82 to 165 feet (25 to 50 m). Those depths are too shallow for containment of radioactive products if a similar cavity were to be used for decoupled nuclear testing. Room-and-pillar mines in hard rock and salt also are poor choices for decoupled testing because damaged or destroyed pillars may well result in collapse of underground openings. Turnbull (1995) claims that nuclear explosions were 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. But they have, in fact, been detected. In his review of Soviet peaceful nuclear explosions (PNEs), Nordyke (1975) describes a proposed ore-breaking project using a 1.8- kiloton PNE. A Soviet list of PNEs (Sultanov et al., 1999) contains a 2.1 kt explosion on September 4, 1972, on the Kola Peninsula in a well-known mining area. It was recorded by 47 open stations with a magnitude of 4.6. Hence, it was well coupled, not decoupled. In 1992, The New York Times reported a nuclear explosion of 1/3 kiloton at noon on September 16, 1979, in a mine at Yunokommunarsk, Ukraine. Sultanov et al. (1999) list it as occurring in sandstone within a coal mine with a yield of 0.3 kt. From seismic arrivals at the NORSAR seismic array, Ringdal and Richards (1993) computed an event time at noon Moscow time and a magnitude of 3.3. They state that it would have been much better recorded if it had occurred in 1993, when an advanced regional seismic array was in operation in northern Europe. It would be even better recorded and located today. The yield computed for magnitude 3.3 was about 3 times smaller than if it had occurred in hard rock. Its smaller magnitude is reasonably attributed to its occurrence in soft rock, not to decoupling. Explosions in coal and similar soft rocks do not produce as large seismic waves as those in hard rock. Decoupling factors of about 20 to 40 have been observed for chemical explosions in hard rock that range in size from a few pounds to about 10 tons. There remains uncertainty in how to map the different energy density of chemical explosions into a nuclear decoupling factor. This uncertainty complicates hard rock decoupling scenarios for a potential evader. Salt is the only

OCR for page 161
172 The CTBT- Technical Issues for the U.S. medium in which a decoupling factor as large as 70 has been obtained for chemical explosions in underground cavities. Thus, a maximum decoupling factor for hard rock of 40, not 70, seems more appropriate to assume for monitoring at 90 percent confidence. A country not wanting to be caught cheating likely would need to use a smaller factor, say one not larger than 10 to 20, and not to attempt decoupled testing larger than 500 tons (half a kiloton) in hard rock. Evaluation of the cavity-decoupling scenario as the basis for a militarily significant nu- clear test program therefore raises a number of different technical issues for a country considering an evasive test: 1. Is there access to a region with appropriate geology for cavity construction? • Is that geological medium nearly homogeneous on a scale of hundreds of meters? • Can cavities of suitable size, shape, depth and strength be constructed clandestinely in the chosen region? 2. For a cavity in salt formed by solution mining: • Is enough water available? • Can it be pumped out and the brine disposed clandestinely--eight times the cavity volume, plus the final brine fill? • How should the very limited experience with conducting decoupled nuclear explosions in salt be taken into account? • Can decoupling factors as high as 70 be attained for yields much larger than sub-kiloton (i.e., larger than the 1966 Sterling test)? • Can the layered properties of rock sequences for bedded salt be dealt with? 3. For a cavity in hard rock: • Can mined rock be disposed of clandestinely? • Can a country afford the price of mining a large cavity in hard rock? • Can uncertainties in rock properties and in orientations and magnitudes of principal stresses be dealt with? • Can presence of joints and faults in hard rock be detected and dealt with? • Can flow of water into cavity—in either hard rock or salt—be dealt with? • Can cavities that depart significantly from a spherical shape8 be used? • Should a decoupling factor no larger than 10 to 20 be assumed? 4. Can collapse of cavity during construction and decoupled test be avoided? • Can surface deformation potentially detectable by interferometric synthetic aperture radar (InSAR) both during and after construction and following the test be minimized? 5. Can radionuclides be fully contained from a decoupled explosion? • Take into account that noble gases can be detected today at much smaller concentrations than a decade ago. • Take into account that radionuclides have leaked from many previous nuclear explosions in hard rock at Novaya Zemlya and eastern Kazakhstan and the few in granite at the Nevada Test Site. 6. Can the site be chosen to avoid seismic detection and identification, given the detection thresholds of modern monitoring networks and their capability to record high frequency regional signals? • Can the limited practical experience with nuclear tests in salt, and very low-yield chemical explosions in hard rock, be extrapolated to predict the signals associated with nuclear testing in cavities in hard rock? • Can the size of a test be made small enough to deal with future advances in detection and identification capabilities? 8 Dimensions that differ by more than 1:4 (Murphy, 2009).

OCR for page 161
173 Appendix E: Evasive Underground Nuclear Testing 7. Is there such a region that is suitably remote and controllable, and that can handle the logistics of secret nuclear weapons testing? • Can secrecy be successfully imposed on all of the people involved in the cross-cutting technologies of a clandestine test program, and on all who need to know of its technical results? • Can the tester avoid compromising security by conducting a nuclear test in a region containing a hostile ethnic group or a civil war? Can the test be conducted outside one’s own territory? 8. Can nuclear explosions of large enough yield be carried out secretly, and repeated as nec- essary, to support the development of a deployable weapon? • Can those carrying out the decoupled test be sure that the yield will not be larger than planned, and thus only partially decoupled? • Can a minimum of drill holes, cables, and specialized equipment be used and yet obtain necessary information about the characteristics of nuclear device(s). • Can the site be cleaned up before an on-site inspection team arrives? 9. Can a clandestine test in a mining area be hidden in one of a series of ongoing large chemical explosions? • Can suitable rock for a decoupled test be found below coal, other minerals and sedimentary rock in which large chemical explosions are used in mining? Mine Masking The 2002 Report briefly described the possibility of evasive nuclear-explosion testing in an active mining region. Many types of mining operations routinely use chemical explosives, sometimes in impressively large amounts (exceeding ten kilotons of chemical explosive for some shots, and annual totals amounting to megatons of explosive per year for the largest industrial countries). An issue of concern in the early 1990s was whether large mine blasts might generate seismic signals in such numbers that efforts at CTBT monitoring for nuclear explosions could be overwhelmed, but it later became understood and accepted that the commercial purpose of mine blasting entails practices that greatly reduce seismic amplitudes and only a small fraction of mine blasts would even be detected. The issue with mine-blast signals then became whether detectable blasting activity could be used to mask or disguise the signals from an underground nuclear explosion. This section provides further details, additional references to papers and a website that describe relevant aspects of the seismic signals from chemical explosions, and some specific mining regions where blasting activity is detected at monitoring networks, and summarizes assessments of the size of the largest underground nuclear explosion whose seismic signals might be successfully masked by mine-blasting. To estimate the overall scale of mine blasting Richards et al. (1992) surveyed blasting operations in the mining industry (emphasizing the United States, with operations in Russia and Europe being comparable) and concluded: “The main point…from the perspective of those concerned with nuclear explosion monitoring and the question of discriminating between chemical and nuclear explosions, is that a large industrialized country can be expected to carry out large numbers of chemical explosions.” The industry would call a shot larger than 50 tons “large,” and on the order of 30 such shots occur each day in the United States, including one at 200 tons or even larger. Several shots at the kiloton level occur each month, and some amount to more than ten kilotons. A key point is that almost all industrial shots larger than 1 ton are “ripple fired” with the total charge broken up into much smaller units, typically less than 100 kilograms (0.0001 kilotons) of chemical explosive, that are fired in sequence to achieve the commercial purpose of breaking or moving large amounts of rock. For the largest chemical explosions, the sequence of separate blasts typically takes tens of seconds to execute. A net

OCR for page 161
174 The CTBT- Technical Issues for the U.S. effect is great reduction in amplitude of seismic signal compared with the strength of nuclear explosion signals, where all the energy is released almost instantaneously. Khalturin et al. (1998) surveyed chemical explosion activity on the territory of the former Soviet Union, finding that this reduction in signal strength (compared to a single-fired shot, or an underground nuclear explosion) was by factors typically in the range of 30 to 100. They wrote that “The reason for the inefficiency of generating seismic signal is presumably because the usual commercial purpose of chemical explosions entails the need to fracture rock into small pieces, which necessitates firing practices (such as ripple firing) in which much of the explosive energy goes into rock fragmentation. A smaller fraction is then radiated seismically than would be the case for a well-coupled single fired shot” (p. 13). They reported a small number of locations where the reduction is less, around a factor of 3 for quite large chemical yields (several hundred tons), and noted that “Such explosions, which appear to be uncommon and declining as blasting practices are modernized, may require special attention in the context of verification of the Comprehensive Test Ban Treaty” (p. 1). Concerning the numbers of potentially problematic mine blasts, Khalturin et al. (1998) also wrote that: “… there are a limited number of regions in which mine blasting is seismically detectable over large distances. The Kuzbass mining region of western Siberia, Russia, and the region near Abakan farther to the east, appears to be associated with explosions with magnitude greater than 3.5 that are likely to be detected a few times each month at considerable distances” (p. 13). Recognizing the potential for ambiguity in interpreting seismic signals both from large mine blasts and mine ground failure (such as cavity collapse), a working group looked at this problem in the late 1990s. The working group created a draft report in 1997 that was reviewed by the National Research Council in 1998 (NRC, 1998). The working group used the NRC review to produce a final document (Chiapetta et al., 1999). As noted in that report the CTBT encourages the voluntary reporting of any industrial explosion greater than 300 tons and has a provision for consultation and clarification designed to address questionable events in an unobtrusive way. The report also notes that there are discrimination techniques to identify large mine collapses and ripple fired mine blasts, though these are not foolproof. Finally the report notes that as the mining industry makes greater use of precision timing systems in conducting ripple firing, which allows better control over the rock fragmentation, the overall size of many of these mine blasts may be lowered to less than magnitude 3.5. The number of mines where there are routine blast events greater than magnitude 3.5 is limited. We examined the CTBTO’s Reviewed Event Bulletin (REB) for 2007-2008 to look for regions of routine blasting that have seismic signals greater than 3.5. Figure E-5 shows six regions that were identified because the seismic events all occur during daylight hours, whereas earthquakes are more random and occur equally during the night and day. These regions each contain multiple large open pit mining operations easily visible in Google Earth. Table E-1 gives the characteristics of these regions. The coal mining region in Wyoming with large mine blasts is very well known (e.g., Chiapetta et al., 1999; Arrowsmith et al., 2008; Zhou et al., 2006). The mine events often have very different characteristics from earthquakes making them easy to identify. For example, the largest event, with IDC maximum likelihood of mb = 3.9, has a very low local magnitude (REB ML = 3.5, USGS ML = 3.2), very different from earthquakes. The repeating nature of mining events also allows the use of waveform correlation methods to obtain very precise relative time, location, and size differences between events. The United States has set a good example by providing publically available information about seismic events associated with mining. Since the late 1990s, the U.S. Geological Survey (USGS) has documented U.S. mine blasts detected seismically, including those that have been

OCR for page 161
175 Appendix E: Evasive Underground Nuclear Testing reported by the REB of the CTBTO. The USGS website9 describes this activity, gives links to archives of information on detected mining blasts, and notes that the work is done in the context of confidence building measures associated with the CTBT. Authorities in both Russia and China have provided information on specific very large blasts on their territories. FIGURE E-5: Map of 41,728 Reviewed Event Bulletin (REB) events from 2007-2008 with depth 0 colored by time of day they occurred. Approximate daytime events are shown in red, and approximate nighttime events are shown in green. Six prominent mining regions are marked. SOURCE: William Walter, Seismology Subcommittee member The REB mb values in the last row of Table E-1are typically derived from a single IMS station and hence may be unrepresentative of the magnitudes that would be reported by a network of detecting stations. Station magnitudes typically exhibit a scatter in values about the mean value, so that use of only one station that rises above noise levels usually results in a magnitude that is biased toward high values. With respect to mine masking, mines typically adopt locally-appropriate practices of blasting so that the infrastructure of the mine and its environs are not stressed too much by local vibrations. Thus part of the work of monitoring is to compare new signals against an archive of previous signals from the same region. But if an underground nuclear test in a mining area were carried out at nearly the same time and place as a mine blast typical of the region, then what magnitude level of signals might result? And what are the possibilities for concealment, via this approach, of a treaty violation? Answers can come from taking examples of signals from large mine blasts, and signals from small underground nuclear explosions, then adding them together before subjecting them to the methods used to discriminate between various types of seismic events. What is typically found is that the maximum size of the identifiable waves (for example, 9 See: http://earthquake.usgs.gov/earthquakes/eqarchives/mineblast/.

OCR for page 161
176 The CTBT- Technical Issues for the U.S. the P-waves) from the mine blast is about that expected from individual sub-blasts (commonly called “delays”), and these amplitudes are spread out over a longer time in seismograms. For this reason, the use of mine-blasts for masking nuclear explosion signals, though they might afford some possibilities, are not very effective for concealing large releases of energy. The seismic signal from an underground nuclear event is an expression of the instantaneous release of nuclear energy, and unless the yield is very small, it stands out in comparison with the size of the energy released in a sub-blast “delay.” TABLE E-1: Example Mining Events in the 2007-2008 REB Catalog. Mining 1 2 3 4 5 6 region in Figure E-5 Wyoming, Zouirat, SW Northern Siberia, Eastern Region, USA Mauritania Russia Kazakhstan Russia Australia Country Approximate 43oN 23oN 52oN 51oN 53-55oN 22oS o o o 74oE 85-92oE 148oE 105 W 12 W 35 E location Coal Iron Coal Various Various Coal Mineral Resources 250 22 34 12 157 33 Number of events in REB 2007- 2008 Largest REB 3.9 3.7 3.6 3.5 3.9 4.0 mb (maximum liklihood) SOURCE: Committee A study of mine masking possibilities by Smith (1993) used several different examples of mine-blast seismograms together with single-fired explosion records and found a number of features that could be used to identify a simultaneous shot within a ripple-fired blast. He concluded that to conceal a single-fired deep detonation (depth is required for containment of radionuclides), the single explosive shot should not exceed 10 percent of the total explosive. This assumes an underground nuclear explosion that is well coupled. If the latter explosion were in a large cavity (which potentially might be possible to the extent that a large mining operation could include necessary equipment for creation of an underground cavity), then all the complication of executing a cavity-decoupled shot would be added to the procedures for carrying out the masking shot (that would itself—to the extent it were detected—attract the attention of monitoring agencies). The discussion of mine-blasting and associated charge sizes is very different for deep mining and shallow mining. The latter is often called surface mining. The largest total charge sizes of chemical explosions in mining (on the order of kilotons) are associated with shallow mining, to remove surface layers in a procedure called cast-blasting (in which a long strip of sediments is thrown sideways) or to break up the uncovered target (usually coal, or iron ore). Surface mining operations differ from deep mining in that underground facilities and the dangers associated with them are absent, so that local ground vibrations can be larger. (Blasting in a deep mine is unusual if it involves more than 10 tons of explosive, because of the damage that large vibrations could do to local underground infrastructures.) Mine masking of an underground nuclear explosion would entail both shallow mining with its large charge sizes and deep mining to get a nuclear device down to levels at which its explosive energy would not open up paths for

OCR for page 161
177 Appendix E: Evasive Underground Nuclear Testing radionuclide releases. Thousands of people could be involved in such combined operations, and concealment of activities in a region that intrinsically draws attention because of its large seismic signals (from routine operations) would per se be a challenge. If a large surface mining operation were to conduct a significant underground shot, the seismic signal would have different characteristics from routine surface shots, and such differences would be amenable to analysis through comparison of signals using correlation and other techniques. Thus for an evader, one of the main perceived advantages to mine masking— the fact that seismic signals from the region are common—is also a major disadvantage. This is because the routine signals allow very good calibration of the mine, hence permitting a detailed investigation into the nature of and differences between events. In addition, surface shots generate infrasound signals in ways that well-contained, deeply buried underground explosions do not. Infrasound signals in combination with seismic signals offer the potential to estimate the fraction of the shot that was underground. Taking the events from Table E-1 as representative of the largest routine mining events, the REB mb = 3.9 is about the magnitude of a single contained fully-coupled nuclear explosion of around 200 tons in a stable region to perhaps around 600 tons in tectonic regions (using the formulas of Murphy, 1996: mb = 4.45 + 1.0 log (Y) for stable and mb = 4.05 + 1.0 log (Y) for tectonically active). Using the Smith (1993) criterion of 10 percent of total yield leads to estimates of around 20 to 60 tons for a masked shot in a large mine blast. Given these small yields, theoretical evasion scenarios often invoke the potential of decoupling the masked test. In this case, the mine is simply a cover for decoupled shots. The issues surrounding decoupled explosions are presented earlier in this appendix. In addition to the seismic signals from blasting, mines can be seismic sources associated with a variety of other phenomena known collectively as “ground failures.” Caused principally by the introduction of cavities and thus stress-free surfaces at depth, where stresses previously had been large due to the overburden, these include coal bumps, mine collapses including pillar collapses, and rockbursts, whose seismic signals can be stronger than those from mine blasting (Chiapetta et al., 1999). Because neither the occurrence of these phenomena nor the size of their signals can be accurately predicted in time, they would not appear to afford practical opportunities for hiding the seismic signals from an underground nuclear explosion. Seismic signals from mine collapses have been intensively studied in the last 15 years because of specific examples of such events in 1995, for which the mix of seismic body waves and seismic surface waves initially seemed explosion-like (i.e., weak surface waves compared with the strength of body waves). Several authors have studied rock bursts and mine collapses with the goal of finding a method of discriminating their signals from those of underground nuclear explosions, and a reliable method has emerged that is based upon the distinguishing characteristic that in underground explosions the rocks in the vicinity of the shot point are pushed outward from the source, whereas in a rock burst or a mine collapse the rocks in the vicinity of the source move inward toward the source (see, for example, Bowers and Walter, 2002; Dreger et al., 2008). The polarity of seismic signals from a rock burst or mine collapse, whether for body waves or surface waves, is thus the opposite of that from an underground explosion. Summary Mining is a global multi-billion dollar industry that regularly uses large amounts of chemical explosives. The main economic purpose of industrial blasting is to fracture rock and expose resources. Therefore most large blasts are “ripple-fired,” spread out in time and space. This ripple firing reduces the seismic amplitudes and imparts characteristics to the signals that

OCR for page 161
178 The CTBT- Technical Issues for the U.S. usually allows them to be distinguished from single-fired contained explosions. There are a limited number of mines in the world that routinely generate seismic events greater than REB mb = 3.5, with the largest signals over the past several years being about 4.0. These large open pit mines take many years to develop and are visible by commercial satellite. Such mines and their blasting signals will be the subject of extra attention under the CTBT, noting that the Treaty provides means for voluntary reporting of shots greater than 300 tons (0.3 kilotons). Since the 1990s, the United States has voluntarily provided a separate list of mining events (times, locations, and magnitudes) through a website of the U.S. Geological Survey. The U.S. should encourage other countries with national seismic networks to do the same. Work done in the 1990s indicates that a masked underground contained explosion would need to be smaller than 10 percent the size of the masking large mine blast. For the largest mine blasts, this leads to estimates for fully coupled shots of a few hundred tons of nuclear yield or less (decoupled explosions are covered in another section). For an evader, mining operations provide cover for extensive excavations needed to contain a nuclear test and reasons for seismic signals. However, mine operations that routinely produce large seismic events (magnitude > 3.5) can be among the best calibrated areas on Earth due to their past record of numerous events. The record of many routine signals allows techniques such as waveform correlation and joint seismic and infrasound analysis to provide ways to flag and identify unusual signals. Mines that routinely produce large seismic signals offer opportunities to better calibrate the seismic network for improved detection, location and identification. Mine collapses that generate seismic signals greater than magnitude 3.5 are infrequent and are not easily controllable for masking purposes. Their unusual seismic signals make them a potential source of false alarms, but new algorithms have been developed that can distinguish such events from both explosion and earthquakes. Unsuccessful Proposals for Clandestine Underground Nuclear-Explosion Testing Several other scenarios have been proposed for clandestine underground testing, but none is considered likely to be successful: • Hiding signals of an explosion in that of an earthquake—Modern instruments detect seismic waves from earthquakes and nuclear explosions over a very broad range of frequencies and at many different distances. Signals from small nuclear explosions can be separated from those of large earthquakes by simple filtering in the frequency domain, using arrays to separate signals arriving from different azimuths and wave speeds, and looking at data from regional stations closest to the presumed explosion. Because earthquakes cannot be predicted, a wait of years typically would be involved after emplacing a nuclear device and then determining in a very short amount of time that a large nearby earthquake had occurred. • Set off series of explosions so seismic waves look like those from an earthquake—This idea was proposed about 40 years ago when identification was made using seismic waves with only two periods, about 1 and 20 seconds—the Ms/mb technique. This evasion proposal does not work when digital data are used, as they are today, with a broad range of periods. • Absorb energy by placing carbon in cavity containing an explosion—very small U.S. tests of this concept many decades ago were not successful. • Reduce size of seismic waves by testing in rubble zone of a previous nuclear explosion—Sites of past nuclear explosions that generated rubblized zones of any significant size are known and can be monitored. Rubblized zones are likely to be conduits for noble gases and other bomb-produced radionuclides.

OCR for page 161
179 Appendix E: Evasive Underground Nuclear Testing Seismic waves from nuclear explosions in the Degelen Mountain subsite of the Semipalatinsk Test Site, Kazakhstan, reportedly were about ten times smaller than expected for an explosion conducted close to the underground location of an earlier nuclear test (results reported by Sokolova, 2008). This modest “decoupling” factor, which is comparable to testing, say, in dry alluvium and thus not large enough to represent a significant problem, should be explored by simulation and non-nuclear experiment.

OCR for page 161