Research Required to Support Comprehensive Nuclear Test Ban Treaty Monitoring (1997)

After a half century in which nuclear weapons were developed, tested, and used, a Comprehensive Nuclear Test Ban Treaty (CTBT) banning all nuclear explosions has been negotiated and signed by 142 countries (as of February 18, 1997) including the United States. Although the U.S. Senate has yet to give advice and consent to U.S. ratification of the CTBT (two nations had ratified the CTBT as of February 18, 1997), it appears likely that nuclear explosion testing is over after a history of more than 2090 explosions. (The July 29, 1996, underground explosion in China may have been the last nuclear test.) Verification of compliance with the CTBT will be a major concern of many nations in both the short and the long-term, and it requires a vigorous research program to enhance capabilities to identify violations, minimize false alarms, and thus maintain confidence in compliance.

This chapter discusses the technical challenges of the CTBT in the context of previous nuclear arms control treaties. It then describes the importance of the Presidential safeguards for monitoring and the contributions of the International Monitoring System (IMS) and basic research toward achieving these goals. Finally, the chapter describes current programs for basic research in support of nuclear monitoring.

The Limited Test Ban Treaty (LTBT) of 1963 prohibited explosions in the oceans, atmosphere, and space by the signatories, bringing to an end the perils of radioactive fallout from testing. However, the LTBT did not ban underground nuclear explosions. Significant new nuclear weapons development and underground testing took place in the ensuing decades. To monitor this treaty, the United States used a combination of atmospheric infrasound, seismic, hydroacoustic, radionuclide, and satellite methods to ensure that explosions in banned environments could be detected. However, given that testing could take place underground, these monitoring efforts were limited in scope. At the same time, it was recognized that seismological monitoring of underground nuclear testing in other countries provided a means by which to monitor advances in weapons technologies reflected in their underground testing practices. Following the recommendations of the Berkner panel (Berkner et al., 1959), the United States deployed unclassified and classified seismological recording systems to enhance the national capability to detect, identify, and characterize underground nuclear explosions. Since 1959, the Department of Defense (DoD) has sustained a

seismological research program to support the analysis of these data, recognizing the central role that this discipline has played in monitoring the development of nuclear weapons, as well as in monitoring for treaty compliance.

The bilateral 1974 Threshold Test Ban Treaty (TTBT) placed an upper limit on the yield of U.S. and Soviet underground nuclear explosions equivalent to 150 kilotons (kt) of TNT. The monitoring challenge of this treaty was to estimate accurately the yield of the largest Soviet underground nuclear explosions. The seismic magnitude of these explosions was approximately 6.1. There was little difficulty in locating or identifying such large events because they produced detectable signals over the entire surface of the Earth. (The seismic wave amplitudes from a 150 kt event are about 50 times larger than those from a 1 kt explosion.) Because seismological monitoring was the primary means to verify the TTBT, there was a vigorous research program to address the question of yield estimation using seismic data. Progress in this area eventually enabled accurate yield estimates. It also documented significant systematic variability of seismic magnitudes at fixed yields arising from variability in wave transmission properties near the major U.S. and Soviet test sites.

The CTBT prohibits all nuclear explosions, effectively extending the LTBT to include underground tests. The 90-page text of the CTBT is about 50 times longer than the text of the LTBT, in large part because of extensive provisions (in the CTBT) for verification. The formal treatment of verification issues in the CTBT will continue to be developed and documented in extensive detail over the next few years. The treaty and its Protocol mention six different Operational Manuals, not yet written, that will spell out the technical and operational requirements:

1. Seismological Monitoring and the International Exchange of Seismological Data;

2. Radionuclide Monitoring and the International Exchange of Radionuclide Data;

3. Hydroacoustic Monitoring and the International Exchange of Hydroacoustic Data;

4. Infrasound Monitoring and the International Exchange of Infrasound Data;

5. the International Data Center; and

6. On-Site Inspections.

The CTBT and its Protocol specify an IMS consisting of 170 seismic stations, 80 stations monitoring relevant airborne radionuclides, 11 hydroacoustic and T-phase⁵ monitoring stations, and 60 infrasound stations, with associated global communications and integrated signal processing infrastructure; an International Data Center (IDC) to collect, archive, process, and distribute data and processing products; as well as procedures for On-Site Inspection (OSI). Importantly, however, the responsibility for determining treaty compliance rests with the States Parties, not with the CTBT organization. Thus, the United States and other nations can use IMS data, along with any additional sources of objective information available to them, to monitor the treaty. However, if a suspect event occurs, an OSI request must be based on information collected by the IMS, on any relevant information obtained by national technical means (NTM) in a manner consistent with generally recognized principles of international law, or on a combination of these. Monitoring the CTBT is anticipated to entail a long-term effort, in which it is desirable to keep costs as low as possible while achieving monitoring objectives. One of the primary purposes of this report is to assess what research activities will be required by the United States to ensure effective monitoring of the CTBT. A previous report of the National Research Council (NRC, 1995) addressed ways in which the CTBT seismic monitoring effort can contribute to independent areas of national concern, such as earthquake monitoring and basic research on the Earth system. IMS data from other monitoring technologies similarly have potential multiple use for research and Earth system monitoring, and ensuring that these data are generally available is important. An extensive history of CTBT negotiations is presented in Pounds (1994) and in several papers in Husebye and Dainty (1996).

The CTBT is a zero-yield treaty, meaning that no nuclear explosions are allowed, including any that might be deemed "Peaceful Nuclear Explosions" (PNE's) under the 1976 Peaceful Nuclear Explosions Treaty. When President Clinton announced that the U.S. negotiating policy for the CTBT would adopt the zero-yield decision, he recognized that this places great demands on any treaty monitoring system and stated, "I recognize that our present monitoring systems will not detect with high confidence very low-yield tests. Therefore, I am committed to pursuing a comprehensive research and development program to improve our treaty monitoring capabilities and operations."⁶ In fact, no viable system can monitor compliance with the CTBT with high confidence to the smallest possible yields, because explosions with a nuclear yield below a few pounds of TNT equivalent can be carried out, and it is technically possible to make such small explosions undetectable by the standard monitoring techniques. Given this reality, the issue of CTBT verification capability involves a political determination of the risk that is acceptable as a function of the yield level, the confidence level that is desired, the resources that are available; and other factors.

It has long been recognized that verification capability for a CTBT is set by overall political agendas, which differ widely between countries. In that setting, this report addresses the ability of technical systems to monitor at various levels and the research that is required to achieve and enhance these abilities. In general, the technical systems put in place to monitor the CTBT will be under pressure to detect, locate, and identify small "events" underground, underwater, and in the atmosphere with high confidence and accuracy. This pressure translates into requirements for research to constantly improve the capabilities and results from technical monitoring systems in a cost-effective way.

The United States has specified precise monitoring levels for international CTBT compliance. The monitoring levels are not geographically uniform, and the specifics are classified. For the purpose of this report, an appropriate distillation of the President's requirements for CTBT monitoring is the phrase "a few kilotons evasively tested" in selected areas of the world (see, for example, the U.S. working paper for the Conference on Disarmament of May 1994 [United States, 1994]). Evasive testing involves any of a number of scenarios for masking or muting the signals from a nuclear explosion. For underground testing, detonation in a large cavity can reduce the magnitude of seismic signals significantly. This is of concern in limited geographic areas of the world, and there are limited opportunities for evasive testing in the oceanic and atmospheric environments. The Presidential Safeguard calling for sustained research and development efforts in support of CTBT monitoring recognizes that evasion scenarios provide strong motivation to continue research in monitoring technologies.

Although any nuclear explosion that is detected will be of interest to the United States and will be a violation of the CTBT if a signatory nation is involved, specification of a target threshold for monitoring guides the assessment of research priorities. In this report the panel focuses on U.S. national needs because international requirements are not defined (in general, based on negotiation history, the United States appears to have more stringent CTBT monitoring requirements than most other countries, although all nations that sign the treaty intend for there to be a total ban on testing). Although some estimates of the potential monitoring capability of the IMS suggest a high-confidence threshold of around 1 kt, nonevasively tested, on a global basis, this is not a formal design criterion. The network design was constrained strongly by considerations of cost and uniformity of coverage. Because the system is not yet fully deployed, validation of the projected monitoring capability will take several years, and even if a fully coupled 1 kt level is achieved, it will not satisfy the U.S. monitoring objective for areas in which evasion is considered viable. The U.S. NTM will augment IMS capabilities, even with the enhancement of the latter as research progresses.

Throughout the past four decades the United States has monitored foreign nuclear testing in

order to assess the associated programs of nuclear weapons development, as well as monitor compliance with previous test ban treaties. The present situation, with a signed CTBT, is different in that the primary driver is monitoring compliance with an arms control treaty that bans all nuclear explosions, and the goals of detecting, identifying, and characterizing frequent (large) nuclear explosion signals are absent. Even more so than in the past, solid technical grounds are imperative for CTBT monitoring, because the lower signal-to-noise ratios inherent in low-yield monitoring and the plethora of nonnuclear events with signals similar to nuclear explosions make it far more difficult to monitor this treaty than previous test bans.

The IMS technologies will provide data for identifying a probable nuclear explosion if the signals can be distinguished from natural phenomena or mining explosions. Additional technologies such as satellite radiation sensors will play a key role in monitoring the atmosphere and space environments. In addition, satellite imaging capabilities can be used to detect testing operations such as drilling or device emplacement, along with sensing environmental changes associated with a nuclear test (ground disruption, crater formation, test analysis facilities).

Any CTBT monitoring system will have practical limits in terms of the capabilities of the system to detect, locate, and identify events. These limits are imposed both by cost considerations that constrain data acquisition and processing and by intrinsic constraints of monitoring technologies. A complete interpretation of monitoring limits must allow for the possibility of various evasion approaches, such as muffling a nuclear explosion signal by detonation of the device in a preexisting cavity (decoupling) or obscuring the explosion signal by simultaneous detonation with an earthquake, quarry blast, or mine collapse. More than 50 years of research underlies the present ability to use the various wavetypes in diverse environments for monitoring applications. The significant progress that has been achieved has provided the technical basis for moving forward with CTBT negotiations. However, national objectives for ensuring international compliance with a total ban on nuclear explosions place extreme demands on all of the monitoring technologies and operational systems, and there is a need for continuing research to enhance the entire U.S. CTBT monitoring system. Furthermore, none of the technologies mentioned above can monitor hydronuclear tests (experiments for which the fissile component of the device is modified to reduce the nuclear yield to levels on the order of a few hundredths of a pound of TNT).

The physical processes associated with nuclear explosions produce distinctive sources of acoustic waves (in the atmosphere or ocean), elastic waves (underground), possible releases of radioactive materials (underground events can result in some immediate release of radioactive gases and particulates or delayed seepage of radioactive gases), or characteristic radiation (in space). These signals and products then propagate through or are advected by the Earth system with various transmission effects and eventually may be detected by different types of sensors placed around the planet's surface or on satellites. The background noise, comprised of signals from nonnuclear events including weather, and/or the physical limitations of the sensors constrain the signals that can be detected and the attributes of the signals that are recorded (e.g., frequency bandwidth, rate of time sampling, background levels of radioactive materials). Signals recorded at different sensors must be retrieved from the field and associated with a common source using general knowledge of how such signals propagate, and the time of origin and location of the source must be estimated. Attributes of the recorded signals, corrected for propagation and instrumentation effects, are then used to identify the type of source, ideally distinguishing nuclear explosion signals from earthquakes or other nonnuclear phenomena. All monitoring technologies share these fundamental elements: source excitation, signal propagation or advection, recording instrumentation, event association, event location, and event identification. They also share the technological challenges of data retrieval and automation of data analysis.

Basic research contributes to monitoring the CTBT by enhancing the performance of monitoring elements, and nuclear test monitoring research programs in the past four decades have carried out research on all of the essential elements mentioned above. The primary technical challenge associated with the CTBT is related to the fact that even very small tests are banned. Signals from small events are more difficult to detect, and the number of earthquakes, chemical explosions, and natural or man-made radioactive sources whose signals have

characteristics similar to those of nuclear explosions increases as the size of the events of concern decreases.⁷ For buried, well-coupled underground explosions, a 1 kt explosion produces seismic magnitudes of about 3.8–4.5 (depending on the source environment). Thus, CTBT monitoring for an explosion yield threshold of 1 kt for underground explosions that are detonated without special concealment efforts would require monitoring capabilities that provide high-confidence detection and identification of events with seismic magnitudes of 3.8–4.5 or larger on a global basis. The projected capabilities of the IMS are for global seismic magnitude monitoring levels of about 4.0 (for high-confidence event detection and location; reliable identification probably will have a higher threshold).⁸ It is possible to reduce the seismic magnitude for a given yield by decoupling, but such evasion efforts also must consider overhead imaging and radionuclide monitoring capabilities.

Decoupling and other evasion scenarios appear to be limited geographically conditions, but in areas of concern (typically involving states with advanced technological capabilities and prior nuclear testing experience), high-confidence detection and identification of events down to seismic magnitudes of about 2.5 is required to monitor fully decoupled 1 kt explosions. Typically, the experience from teleseismic monitoring ⁹ is that reliable identification of events requires a detection threshold approximately 0.5 magnitude units below the target threshold. This increases the number of events that must be examined and, if possible, identified. Regional seismic signals (involving waves that travel in the crustal waveguide) might exhibit a smaller difference in identification and location thresholds.

U.S. monitoring goals apparently will not be met by the stations of the IMS alone but may be met by the use of additional stations, possibly including mobile monitoring capabilities. To put these numbers in perspective, current global earthquake bulletins produced by the earthquake monitoring community are complete for magnitudes of about 4.5 and larger, and complete catalogs are obtained for magnitudes as small as 2.0 in localized areas with regional earthquake monitoring systems. In this setting, assessments as to whether a given monitoring system is adequate for treaty verification will be driven, in part, by perceptions of the plausibility of treaty evasion.

As the magnitude threshold decreases, the number of detected earthquakes and chemical explosions increases. At the same time, the number of stations in a fixed network that will detect a given event and the distance range at which detections are made decrease. (Wave amplitudes tend to decrease with distance as the energy spreads outward.) Given these factors, the detection, location, and identification of small events by combined IMS and NTM assets involve the analysis of signals recorded at regional distances where wave propagation is often complicated and regionally varying. For example, "regional" seismic waves within 1000 km of the source typically reverberate in the crustal waveguide, potentially obscuring the source type and complicating event location. Even when only a few stations provide data, it is necessary to have a high-confidence location with an area of uncertainty no greater than 1000 km². This requirement reflects operational requirements and is mandated by the On-Site Inspections provisions of the treaty (Protocol to the CTBT, Part IIA). The challenge of precisely locating and confidently identifying all small events at some low-magnitude threshold given sparse monitoring networks is formidable. Meeting it requires a sustained basic research program in support of CTBT monitoring.

Current U.S. treaty monitoring capabilities are founded on knowledge and research discoveries that emerged from university research programs over the past five decades. For example, in the realm of seismic monitoring, almost all of the key concepts and methods originated at universities.

These include seminal research on event detection, location estimation, discrimination, yield estimation, seismic magnitude scales, global velocity models, seismic wave attenuation, seismic coupling, seismic regionalization, seismic wave excitation, and broadband seismic recording systems. Much of this university research was funded by DoD research programs. Given the historical evolution of current monitoring capabilities, it is reasonable to assert that the long-term development of new or enhanced monitoring capabilities is largely dependent upon sustaining a fundamental research program in relevant areas. This is also critical for training the next generation of scientifically capable personnel to support long-term CTBT monitoring operations.

Scientific support of CTBT monitoring is not restricted to fundamental research programs conducted by universities. Other types of scientific research are essential for national monitoring capabilities. These include applied or exploratory developmental research, which also involves the development and testing of new scientific understanding and technologies for the functions of treaty monitoring. For example, a proposed seismic magnitude scale for regional distance observations may require extensive testing and validation prior to acceptance (or rejection) by the monitoring organization. This type of work is usually conducted by private contractors, who pursue sophisticated developmental work on specified problems, often using data (which may be classified) from actual monitoring operations. This activity serves a key role in the transition of scientific advances from universities to government agencies. The technical expertise of these private companies is provided mainly by university graduates who are then trained in test ban monitoring by their work with private contractors.

Even more directly linked to operational needs is the implementation of scientific systems such as communication, computer, and software platforms. A major element of this effort for CTBT monitoring involves automated processing by computer software. This advanced developmental work usually involves state-of-the-art technology, and again private contractors play a major role. Some agencies sustain internal scientific capabilities (e.g., the Department of Energy's National Laboratories) for pursuing a range of fundamental, exploratory developmental, and advanced developmental research, but by their very nature these tend to be mission-oriented programs. Innovative basic research is pursued more easily in a university setting.

Ideally, the national treaty monitoring capability built on this scientific infrastructure should be established prior to a political agreement; however, the political forces driving treaty negotiations are generally not coupled with treaty monitoring capabilities. Although the lag of a monitoring capability may not preclude signing a treaty, it can be a reason for delay in ratification, as in the case of the 1974 TTBT, which was not ratified by the United States until 1990, in part because confidence-building measures that would have calibrated monitoring systems were not set in place until the late 1980s.

When scientific capabilities and understanding evolve in parallel with the political process, there is often confusion or turmoil with respect to the role and adequacy of the scientific contributions. Such was the case for the TTBT, when ongoing research was coming to an understanding of the variations of seismic magnitudes associated with 150 kt explosions in different regions of the world at about the same time (and in some cases even before) that political assertions of treaty violation were being made based on earlier (flawed) scientific understanding. The ultimate validation of the seismological research results brought about by the bilaterally (U.S.-U.S.S.R.) monitored explosions of the Joint Verification Experiment (JVE) in 1988 reinforced the value of seismological monitoring of treaties. This experience emphasized the need for the operational regime to be fully informed and responsive to research advances and to implement them rapidly in operations; for policy makers to be aware of the technical limitations of the monitoring systems; and for the technical community to be aware of the policy applications of its results and the need to be clear about their strengths and limitations.

The fundamental technical bases for most nuclear test monitoring technologies have not changed substantially over the past few decades. As discussed in Chapter 2, a solid foundation of physical principles underlies all of the monitoring technologies anticipated to play key roles for the CTBT. However, every technology has intrinsic capabilities and limitations imposed by the monitoring system, the physical nature of the technology, the number and nature of nonnuclear

events with signal characteristics similar to those of nuclear explosions, and the heterogeneity of the Earth. Enhancing the operational capabilities of all disciplines can be achieved by a combination of a full spectrum of coordinated research and development conducted by universities, private companies, and government laboratories and systematic implementation of the methodologies with continuing empirical calibration of each monitoring element. Development of new, unexpected monitoring capabilities also requires a basic research effort. The very nature of the CTBT monitoring environment is that monitoring capabilities steadily will evolve and improve, but they will never achieve 100 per cent confidence in being able to identify and attribute small or evasively conducted nuclear test explosions. It is desirable to establish rapidly the monitoring capabilities of the system when all IMS and NTM assets are in place and empirically calibrated, (a process that will take several years) and then to monitor improvements in these capabilities brought about by continuing research efforts and discoveries. Throughout the monitoring process there will be problem events that defy identification by standard means, and it is important to develop a system that can exploit all available data to resolve the cases that are of concern. Clear communication among the research, operational, and policy arenas about the state of the monitoring capabilities and their continuing evolution must be promoted by the monitoring infrastructure in order to avoid misunderstandings at any level. As the President has stated, the U.S. decision to support the CTBT necessitates a strong commitment to a research program to enhance monitoring capabilities.

The level of effort in basic research that must be sustained involves a policy decision that should be informed by appraisals of the present capabilities of the system, assessments of the potential for improvement, and the benefits of enhanced levels of treaty compliance. For global monitoring systems, this value-added assessment should take into account dual-use benefits between CTBT monitoring and other national interests such as earthquake and volcano hazards, meteorite monitoring, and nuclear reactor emission monitoring. All of these activities will benefit greatly from multiuse of the data collected by the IMS and NTM for CTBT monitoring purposes, and they will build confidence in compliance with the CTBT.

It is important to recognize that the role of science is multifaceted, particularly for monitoring technologies associated with complex Earth systems. For example, the effort to understand seismic or acoustic signals from nuclear explosions should not be conducted in isolation from understanding such signals resulting from earthquakes, quarry blasts, landslides, or other sources of wave motions, some of which have properties similar to those of nuclear explosions. Similarly, one cannot simply study sources in one part of the world and generalize the results to the entire globe. Geological processes have produced great heterogeneity in the planet's interior that influences the propagation of seismic and acoustic energy on each path from source to receiver. Similarly, wind patterns, ocean currents, and ocean floor topography vary from place to place, and human activities are region dependent. There is an intrinsic need to attain an understanding of the source and propagation effects for all monitoring technologies for all significant source types and specific regions of the world.

Awareness of the complex range of scientific issues associated with nuclear test monitoring has led to the establishment of broadly defined research efforts in the past, addressing issues from fundamental source and propagation theory to the calibration of specific paths. In addition to stimulating basic scientific investigations, treaty monitoring efforts motivate many technical research endeavors, such as those leading to automation of signal processing, enhanced telecommunications, and sensor development. It is essential to maintain some separation between immediate operational needs and the basic research effort because focus on the former may not allow one to develop the new perspective that completely alters the context and the panoply of available resources. Availability of new data also results in unexpected research problems that can be solved only by a broadly based program.

Although the balance of fundamental research investigations and applied developmental efforts may be expected to shift somewhat with time as any field matures, the CTBT is a context in which continued basic research will play a key role even as current U.S. monitoring objectives are achieved over the next decade. This importance follows from the very nature of CTBT monitoring, which involves ''problem events" that have characteristics

similar to those of nuclear explosions. These could be violations of the treaty or "false alarms," politically and economically costly mis-identifications of natural events as presumed nuclear tests (with purported high confidence). There are a vast number of detected events that routine capabilities will dispose of readily, but some problem events will always remain (which has been well established by past monitoring efforts). This is particularly true for the small events important to CTBT monitoring if efforts at evasion are deemed to be of serious concern. By definition, conventional processing fails to give high-confidence identification for problem events, and innovative approaches will be needed to reduce the yield equivalent at which problems arise. The new approaches may come from unexpected directions that must be sustained by basic research endeavors. The panel returns to this issue in Chapter 4.

Given the U.S. national objective of monitoring the CTBT to a level of a few kilotons evasively tested in key areas of the world, it is possible to estimate the operational requirements of the monitoring network based on historical experience with nuclear explosions of known yields. For each monitoring technology a theoretical monitoring threshold can be established, which is defined in terms of a high confidence level (often defined as 90 per cent confidence of recognizing a violation if one is attempted) in detection and identification of all events (as nuclear or non-nuclear) with a certain monitoring characteristic. The relevant characteristics vary among technologies, depending on the particular measurement procedures. For example, in seismic monitoring, many measures involve seismic magnitudes (logarithmic scales based on the amplitude of ground shaking produced by different events). For hydro-acoustic signals, the relevant measurements are typically described in terms of logarithms of pressure units such as micropascals (µPa). These measures, corrected for predictable propagation effects such as the decrease of acoustic and seismic wave amplitudes with increasing distance from the source caused by expansion of the wavefront, are indirect measures of the source properties. For example, the precise fraction of nuclear explosion energy or earthquake energy released into the elastic wavefield is not know (it is on the order of at most a few percent) and must be established empirically. Thus, to translate the monitoring capability of seismology into a corresponding bound on nuclear explosion size requires an empirical calibration of magnitude in terms of explosion yield.

Given the past 50 years of nuclear testing there is an empirical basis for relating current-day thresholds for different monitoring systems to equivalent nuclear explosion yields, albeit with some uncertainty. For example, a 1-kilogram explosion (chemical or nuclear) detonated in the SOFAR channel (SOund Fixing and Ranging; see Appendix E) of most regions of the world's deep oceans will be detected by several hydroacoustic stations. A 1 kt explosion in the SOFAR channel in most oceanic regions would drive most of the hydroacoustic instruments around the world off-scale (there are, however, oceanic areas that may be blocked). Thus, the favorable sound transmission properties of the ocean define a low effective detection level, at which the challenge will be in distinguishing conventional explosions (e.g., for seismic exploration) from other events.

The solid Earth is far less efficient in transmitting seismic waves, but there is an extensive basis for relating seismic magnitude to explosion yield for certain test sites. For example, based on announced yields for various test sites around the world, one particular seismic magnitude, m_b(ISC) (the 1-second period P-wave magnitude determined by the International Seismological Centre [ISC]; see Appendix D for a discussion of various magnitudes) has been found in one study to have the following mean values for a 150 kt explosion at each site (Adushkin, 1996):

• 5.74 (Nevada Test Site, tuff);

• 6.12 (Semipalatinsk, East Kazakhstan);

• 5.97 (Novaya Zemlya, Russia);

• 5.93 (Lop Nor, China); and

• 6.04 (Mururoa, French Polynesia).

Such numbers form the basis for seismic monitoring of compliance with the TTBT.

Linear relationships between seismic magnitude and the logarithm of the explosion yield make it possible to determine "yield-scaling" curves that

allow prediction of the yield, and its uncertainty, for any given magnitude. For a 1 kt explosion, such relations for m_b(ISC) predict corresponding magnitudes of

• 3.87 (Nevada Test Site, tuff);

• 4.45 (Semipalatinsk, East Kazakhstan);

• 4.32 (Novaya Zemlya, Russia);

• 4.3 (Lop Nor, China); and

• 4.5 (Mururoa, French Polynesia).

These magnitudes have higher uncertainty than the 150 kt values above, because 1 kt is near or below the level of the smallest events that were available for calibrating the yield-scaling curves. Also, the effort over the past two decades was focused on the 150 kt level associated with TTBT compliance, and the uncertainty increases as the yields vary from this value.

Such data provide a basis for translating a monitoring system capability defined in terms of seismic magnitude threshold into a corresponding yield threshold. The magnitudes given above typically correspond to the central point of the distribution for a given yield, and somewhat lower values (by about 0.2 magnitude units) are needed to provide a 90 per cent confidence level at 1 kt.¹⁰ Thus, the above values suggest that teleseismically detecting and identifying 90 per cent of the 1 kt nuclear explosions near the Chinese test site would require a seismic monitoring system with capabilities to detect and identify all events above m_b(ISC) of 4.3-0.2 = 4.1. Typically, however, teleseismic identification requires magnitudes about 0.5 units above the detection and location threshold,¹¹ thereby lowering the detection and location threshold for the system to 3.6. Evasive decoupling would lower the seismic detection level of the monitoring system by another 1.5–1.85 magnitude units (factors of 30 to 70 reductions in amplitudes for a given yield.¹²) Thus, the experience from teleseismic monitoring suggests that a detection and location capability near magnitude 1.8–2.1 would be required to confidently identify an evasively tested 1 kt explosion near the Chinese test site.¹³

At these low magnitudes, however, monitoring will be done at regional distances. Importantly, the experience with regional signals suggests that the difference between the thresholds for detection or location and identification will be less than 0.5 (as in the teleseismic case). In the best case, this offset could be zero, and the detection requirement would be increased to approximately 2.3–2.6.

In this way, monitoring goals expressed in terms of yield levels and assumptions about decoupling capability are translated into monitoring capabilities that a seismometer network and an associated data center must be designed to handle. An obvious complication is that there are large (0.7 magnitude units, corresponding to factor of 5 seismic ground motion amplitude variations) systematic differences in magnitudes expected for a 1 kt explosion, and this variation is found even among the only five calibrated sites mentioned above. The reason is that variations in coupling of energy from the explosion into the seismic signals (caused largely by rock strength and porosity at each site) and variations in seismic wave attenuation properties (associated with crustal and upper-mantle tectonic history and thermal structure) affect the observed seismic motions. Thus, a globally uniform seismic magnitude monitoring capability implies a nonuniform explosion yield monitoring capability. Fortunately, an understanding of the reasons for differences in magnitude levels between calibrated test sites gives us some predictive capabilities. For example, the highly attenuating structure of the tectonically active western United

States is mainly responsible for the relatively low seismic magnitudes for Nevada Test Site (NTS) explosions, and other events in the same region experience similar systematic reductions in seismic wave amplitudes for a given source energy level. One can predict, with reasonable confidence, that other tectonically active areas around the world will have similar magnitude reductions relative to geologically stable areas.

To generalize the above discussion, a rational assessment of monitoring capability for the CTBT will require three steps. The first involves an assessment of the intrinsic monitoring capability in terms of signals acquired for each of the monitoring technologies. Second, there is a need to assess the opportunities to merge the data streams from different technologies to enhance detection and identification capabilities. The third step involves an empirical translation of all monitoring measurements into a corresponding explosion yield monitoring threshold, false-alarm rate, and uncertainty. With this background, a policy decision can be made about the performance of the monitoring system and the level of compliance assurance. If it is not sufficient, additional assets must be deployed or research carried out to bring the level down sufficiently.

The CTBT includes specific plans for an extensive IMS that will collect, process, and distribute seismic, hydroacoustic, infrasonic, and radionuclide data collected from global arrays of sensors. The seismic, hydroacoustic, and infrasonic data will be transmitted continuously (or archived at the site and available by dial-up request) to an International Data Center (IDC). The mission of the IDC will be "to support the verification responsibilities of States Parties to the CTBT by providing objective, scientifically-demanding and technically-demanding products and services necessary for effective global monitoring." This will involve the automated and interactive combination of data ("fusion") from different monitoring technologies and the production of a list of event locations (without specific identification), compiled in the IDC Reviewed Event Bulletin (REB). This bulletin, along with raw and processed data, will be provided by the IDC to National Data Centers (NDCs) of the States Parties, which can in turn incorporate them into their national treaty monitoring activities. In addition, the IDC may screen events and carry out screening operations upon request of and for a State Party.

The U.S. NDC will be operated by the Air Force Technical Applications Center (AFTAC), which has historically had the lead U.S. role in monitoring foreign nuclear weapons testing. During the Group of Scientific Experts Technical Test 3 (GSETT-3,¹⁴) the U.S. NDC considered three mechanisms to provide the scientific research community with open access to all IMS seismic data:

1. an interactive World Wide Web data request procedure;

2. continuous telemetry of windowed seismogram segments to the U.S. Geological Survey (USGS) National Earthquake Information Service (NEIS) for earthquake monitoring applications; and

3. continuous telemetry of all IMS seismic signals to the Data Management Center (DMC) of the Incorporated Research Institutions for Seismology (IRIS) for incorporation into the IRIS data archives, which are the primary data source for U.S. seismological researchers.

In practice, only limited amounts of IMS data (primarily from U.S. stations) have been transmitted by the above mechanisms during GSETT-3. In part these data transfers were limited by logistical problems of establishing the communication infrastructure and protocols from the U.S. NDC. Open distribution of the IMS data by these mechanisms has also been limited, however, by the lack of clear policy on access to international IMS data.¹⁵ As this report was being written, the panel was aware that such policies are still undefined and that they may only be resolved through international discussions at the Preparatory Commission. Recognizing that this is an issue of

continuing discussion and negotiation in the U.S. Government and at the Preparatory Commission, the panel notes that the treaty contains a strong statement supporting open access to IMS data for scientific research and that there is no language in the treaty suggesting data restrictions. Specifically,

A previous NRC report (NRC, 1995) detailed the benefits to CTBT monitoring from multiuse of seismic data streams. The report concluded that open access to IMS seismic data streams would encourage the dual use of such data for earthquake and CTBT research and that it would lead to mutual and unpredictable benefits to both fields. The current panel endorses the data access recommendations from NRC (1995), emphasizing that they should be applied to all IMS technologies. In the panel's view, the benefits of basic research to CTBT monitoring will be severely limited if IMS data is only provided to a limited group of U.S. researchers supported by DoD and DOE CTBT programs. In the panel's view, the most effective strategies for improving U.S. monitoring capabilities will facilitate research contributions from the broadest segments of the scientific community by allowing open access and multiuse of IMS data streams. To facilitate this research, the panel strongly recommends that the US Government should formulate a policy supporting open distribution of IMS data for scientific research.

For the purposes of this report, the panel discusses and recommends research in seismology, hydroacoustics, infrasound, and radionuclide monitoring assuming that there will open access to all of the IMS data for scientific research. Recognizing that this issue has not been resolved, the panel justifies its approach in two ways. First, by discussing the benefits to monitoring from multiuse of IMS data, the panel intends to contribute to policy debate on this issue. Second, the panel's charge from NTPO clearly anticipates the importance of the IMS data as a foundation for a strong research program. Specifically,

The monitoring stations of the IMS are specified in Annex 1 to the Protocol to the CTBT. The seismic monitoring network will involve 50 stations comprising a Primary Network, all providing continuous seismic recordings to the IDC via satellite and telephone communications systems. About 30 of the Primary stations will involve small-aperture, vertical-component high-frequency seismic arrays, and all stations will have a three-component broadband seismic recording system. The IMS seismic system will also include 120 Auxiliary Network stations, all with three-component sensors. Only two of these will include arrays. Data from the Auxiliary Network will be available for dial-up or Internet access from the IDC, as well as being maintained in archives at the responsible NDCs for special data requests. Telemetry of selected Auxiliary stations (usually those closest to an initial event location) will be requested by the IDC according to criteria established during the automated event location and processing steps. This achieves significant economy relative to continuous telemetry of all of the data but limits the role of Auxiliary Network data in the initial definition and association of events. The on-demand status of Auxiliary Network data implies that its most important use will be identification, with lesser applications for location and detection. Because of its comparatively lower use, Auxiliary Network data will not be calibrated to the same degree as Primary Network data, vis-ô-vis detection and identification, except for stations processed for national purposes. The Auxiliary Network does provide a backup to the Primary Network should data flow be interrupted.

The planned locations of the International Seismic Monitoring Stations are shown in Figure 1.1A. The network is far from being fully deployed (Figure 1.1B), but it is more complete and has a longer operational history than any other IMS technology. The prototype IDC, which has

operated since January 1, 1995, was receiving data from 35 (14 arrays, 21 three-component) Primary and 51 (15 of which are used infrequently because of poor cost-benefit and/or performance) Auxiliary stations as of January 1997. The primary function of this seismic system is to monitor underground explosions, but as discussed later, ground motion recordings may also provide information about explosions in the water and atmosphere. Figure 1.1A makes it clear that the political or technical process of CTBT negotiations brought about a relatively uniform global distribution of seismic stations, but even then, there are large regions of the continents and oceans that have few seismic stations. The complementary monitoring capabilities provided by different sensors and NTM must be assessed in these areas, and if adequate monitoring thresholds to meet U.S. CTBT monitoring goals for specific areas are not achieved, additional NTM assets may be deployed. This is true for all of the monitoring systems.

A new 60-station infrasound network is also being deployed for the IMS, involving sensitive atmospheric pressure gauges that can detect acoustic waves in the frequency band excited by atmospheric nuclear explosions. This system primarily will monitor atmospheric events but may also have a role in monitoring underground and underwater environments. The distribution of IMS infrasound stations is shown in Figure 1.1A. In a large number of cases the stations will be near seismic or hydroacoustic stations, which will allow comparison of signals recorded by the different sensors. There has been no extensive infrasound network in operation since the early 1960s, and even given the historical record from infrasound stations that recorded many atmospheric explosions in the 1950s, it is likely that further operational experience will be needed to understand all sources of background atmospheric sounds that must be discriminated from the signals from small explosions. Data from this network will provide valuable information about phenomena such as volcanic eruptions, meteors, and microbaroms (sounds associated with ocean waves). The prototype IDC was receiving data from four infrasound stations as of January 1997, with many stations yet to be deployed.

The oceanic environment will be monitored in part by the IMS hydroacoustic network of 11 systems: 6 single hydrophone sensors (underwater pressure gauges that detect pressure waves in the ocean) and 5 island seismic stations for recording seismic T-phases. The sparseness of the hydroacoustic network (shown in Figure 1.1A) stems from several factors:

1. the political resolution of several States Parties not to include sophisticated hydroacoustic monitoring systems in the IMS,

2. the high cost of hydrophone installations,

3. the efficient sound transmission properties of the oceans (which enable low monitoring thresholds at least in the deep oceans), and

4. the concentration of land-based sensors in the northern hemisphere.

The IMS hydrophone systems will lack the capabilities of hydrophone arrays, which are able to track underwater moving objects using array processing methods. This component of the IMS data collection effort is not exploiting state-of-the-art capabilities as a result of policy decisions and the interplay between seismic and hydroacoustic monitoring capabilities. Additional U.S. hydroacoustic systems (arrays) may provide extended capabilities for U.S. monitoring purposes, although the long-term status of those systems is highly uncertain. The availability of global, unclassified hydroacoustic data offers great research potential for studying underwater phenomena, such as volcanic eruptions, seaslides and turbidity currents, and submarine earthquakes, so there are again many dual-use applications of the data from this system. The prototype IDC was receiving data from four hydroacoustic stations and one T-phase seismic station as of January 1997.

The seismic, infrasonic, and hydroacoustic systems have similar requirements for data transmission and processing, and standardized formats, communications protocols, and parallel processing algorithms have been established for them by the IDC. The details of processing differ because of the distinctive noise conditions; for example, because of the low signal-to-noise ratios characteristic of infrasound signals, detection of infrasonic signals requires correlation detectors rather than ratios of short-term signal power to long-term signal power as used for seismic waves. The basic product of the analysis of each type of signal is an event bulletin, which will include events uniquely detected by one class of signals as well

as events detected by a combination of wavetypes from different technologies.

There will be 80 radionuclide stations in the IMS, all of which involve surface stations that sample large volumes of air. All of these systems will collect particulates, and 40 stations will include noble gas detectors. High-resolution gamma-ray spectra will be obtained daily from the particulate and gas samples, and these spectra will be transmitted to the IDC. The CTBT protocol identifies 16 radionuclide laboratories that will maintain specific analytic capabilities for confirming the presence of nuclear debris on the original samples when the gamma-ray spectra indicate radionuclide anomalies. The prototype IDC is presently using a four-level reporting system for the radionuclide methodology: Level 1—natural radionuclides within normal station observations, Level 2—natural radionuclides outside of normal station levels, Level 3—fission or activation products within normal station observations, Level 4—fission products outside normal station observations. A Level 4 report will be more comprehensive than the other reports, and all anomalous fission or activation products will be indicated in the Fission Product Event Bulletin. This report will trigger the use of atmospheric transport models to backtrack the radionuclides' source location. As of January 1997, 21 radionuclide stations were providing data to the prototype IDC.

Some of the stations currently providing data to the prototype IDC are temporary and will be replaced when final IMS stations are installed, and some will require upgrade of current capabilities before meeting IMS standards. However, the current stations are providing a realistic test of data flow for the projected IMS. When the IMS is fully deployed, it will produce a data stream of about 10 gigabytes per day flowing into the IDC. The complete data set acquired by the IDC will be transmitted to the U.S. NDC for the purpose of CTBT monitoring. There will be similar analysis for the detection of events conducted by both the IDC and the U.S. NDC; however, NDC operations will differ in two keys ways. First, the U.S. NDC will incorporate additional data streams from national seismic, hydroacoustic, infrasonic, radionuclide, and satellite sensors, as well as other NTM. Second, it will attempt to identify each event at or above specified monitoring thresholds to ensure that no nuclear tests have taken place. A list of suspect events, along with a bulletin of all events that have been analyzed, will be communicated to the U.S. government. The latter list and at least some of the additional NTM used by the U.S. NDC will not be available routinely to the research community, although under certain conditions, such as the case of problem events, classified research activities may be conducted by the research community using the restricted data. If the IMS data were available to the research community, the future CTBT research program would involve extensive research on many of the actual monitoring signals. This would greatly enhance the impact of unclassified research on the monitoring regime and enable applications of treaty monitoring data to other areas of national concern (such as hazard monitoring).

Monitoring all of Earth's environments for nuclear tests clearly requires a diversified effort that draws on scientific expertise in many fields. The U.S. capabilities for monitoring the CTBT are founded on a substantial infrastructure for data collection, analysis, and reporting that has been assembled over nearly 50 years of nuclear test monitoring. This includes extensive experience with seismic, hydroacoustic, infrasound, radionuclide (ground and airborne sensors), and satellite monitoring. The U.S. operational regime is supported by extensive basic and applied research activities that incorporates expertise from many agencies and the academic and private sectors. The monitoring efforts are further supported by the nuclear testing experience of the United States, as well as by a great variety of intelligence assets that serve to define monitoring capabilities in different regions of the world.

Although the full scope of U.S. national efforts is not described here, a few key roles are identified. Although nuclear test treaty monitoring is intrinsically an arms control issue, DoD and DOE have had primary responsibility for the U.S. effort for decades. Currently, the Air Force Technical Application Center has the primary operational task of monitoring nuclear testing treaties, and the U.S. NDC for CTBT monitoring is being established at AFTAC. AFTAC is implementing a computer analysis capability to access IDC and NTM data

and carry out the detection, location, and event identification procedures that will lie at the heart of U.S. CTBT monitoring.

For more than 35 years, AFTAC operations have been strongly linked to DoD research programs, with basic research (the so-called 6.1 program), exploratory development research (6.2 program), and advanced development research (6.3 program) components. In recent years the 6.1 program has been directed by the Air Force Office of Scientific Research (AFOSR), the 6.2 program by the Air Force Phillips Laboratory (AFPL), and the 6.3 program by AFTAC. An innovative technology and advanced developmental program (6.2/6.3) has been directed by the Advanced Research Projects Agency (ARPA). These programs and their interrelations, are discussed at some length in the NRC (1995) report. Although some internal DoD research has been conducted by AFTAC, AFPL, and ARPA personnel, the primary research effort has involved peer-reviewed, externally funded university and private contractor researchers, with budgets of about $7.6 million per year provided by AFOSR, ARPA, and AFTAC in the past two years. In addition, ARPA has supported the development of a prototype-IDC, involving extensive hardware and software development, with a budget level of about $15 million year for the past two years. The latter effort involves advanced developmental research, which is separate from the fundamental research of the peer-reviewed program.

The Department of Energy has had primary responsibility for U.S. nuclear weapons development and testing programs and has also sustained long-term research programs that support nuclear test treaty monitoring. The latter include large internal research programs on seismological and hydroacoustic monitoring, satellite systems development, On-Site Inspection methodologies, and modeling of nuclear explosion signals in all media. In FY 1995, DOE was assigned an expanded responsibility for research and development for monitoring and/or verifying compliance with the CTBT, which encompassed all anticipated monitoring technologies and systems. This expanded DOE program is strongly linked to the AFTAC operational effort. For the past two years it has included a substantial ($3.665 million [FY 1995] and $4.395 million [FY 1996]) external funding effort supporting university and private contractor research activities. It appears that this external program will be greatly reduced beginning in FY 1997, leaving the DoD program as the main support base for university and industry fundamental research in support of the CTBT.

Many other government organizations contribute directly to U.S. nuclear test monitoring efforts. These include (1) the National Oceanic and Atmospheric Administration (NOAA), which has responsibilities for weather forecasting and atmospheric modeling, (2) the USGS, which receives seismic data and bulletin information from thousands of stations around the world and plays the lead role in documenting the seismicity of the United States; (3) the Office of Naval Research (ONR), which supports research in long-range acoustic propagation in the oceans; and (4) the National Science Foundation (NSF), which supports basic research in many relevant areas.¹⁶ These other federal agencies support basic research in areas relevant to many of the CTBT monitoring technologies and can be viewed as indirect support of the U.S. monitoring capabilities. Earthquake monitoring conducted by the USGS-NEIS, the ISC, and many regional networks around the world provides independent determinations of event bulletins derived from much larger numbers of stations than the IMS. (2600 seismic stations currently report to the USGS.) These are a significant source of information about earthquake activity at large and small magnitudes that can support U.S. CTBT monitoring activities. As detailed in NRC (1995), this is an important example of the potential contributions to CTBT monitoring from research outside of DoD and DOE programs.

In 1996, the DoD program was substantially restructured in response to changing priorities. Although AFTAC continues to be tasked with

serving as the CTBT NDC, the AFOSR, AFPL, and ARPA research budgets were consolidated into a single DoD funding line, organized under the new Nuclear Treaty Program Office (NTPO), overseen by the Assistant to the Secretary of Defense for Nuclear, Chemical, and Biological Programs. The Department of Defense Appropriations Act of 1997 recommended a budget of $29.1 million for the associated Air Force arms control funding element, with $8.8 million for a peer-reviewed external CTBT monitoring research program ($7.1 million specifically for peer-reviewed basic research in the field of explosion seismology and $1.7 million for research in complementary disciplines such as hydroacoustics, infrasound, and radionuclide analyses). The other $20.3 million is for sustained development of the IMS IDC and involves operational systems development. The NTPO has tasked the Defense Special Weapons Agency (DSWA; formerly the Defense Nuclear Agency) to oversee the external research program, and an initial Program Research Development Announcement (PRDA) was issued by DSWA in November 1996. Initial contract awards are anticipated by mid-1997.

Consolidation of the DoD CTBT research and development program constitutes a major restructuring of the research community's support and has prompted widespread concern about future support for basic research in the field of seismology. There have been significant turmoil in the seismological research program over the past 15 years and philosophical disagreements over the balance and nature of the research program that will best service the CTBT monitoring effort. By eliminating past bureaucratic structures, DoD has an opportunity and a responsibility to set in place an effective CTBT research program for the future. Attendant issues are addressed in Chapter 4 of this report.