Seismic Verification of Nuclear Testing Treaties

Seismic Verification of Nuclear TestingTreaties

May 1988

NTIS order #PB88-214853

Recommended Citation:U.S. Congress, Office of Technology Assessment, eism”c Verification of NuclearTesting Treaties, OTA-ISC-361 (Washington, DC: U.S. Government Printing Office,May 1988).

Library of Congress Catalog Card Number 88-600523

For sale by the Superintendent of DocumentsU.S. Government Printing Office, Washington, DC 20402-9325

(order form can be found in the back of this report)

Foreword

Since the advent of the atomic bomb there has been interest from both anarms control and environmental perspective to restrict the testing of nuclear weap-ons. Although the debate over nuclear testing has many facets, verification is acentral issue to the consideration of any treaty. At the requests of the SenateSelect Committee on Intelligence, the House Committee on Foreign Affairs, andthe House Permanent Select Committee on Intelligence, OTA undertook an assess-ment of seismic capabilities to monitor underground nuclear explosions.

Like an earthquake, the force of an underground nuclear explosion createsseismic waves that travel through the Earth. A satisfactory seismic network tomonitor such tests must be able to both detect and identify seismic signals inthe presence of “noise,” for example, from natural earthquakes. In the case ofmonitoring a treaty that limits testing below a certain size explosion, the seismicnetwork must also be able to estimate the size with acceptable accuracy. All ofthis must be done with an assured capability to defeat adequately any credibleattempt to evade or spoof the monitoring network.

This report addresses the issues of detection, identification, yield estimation,and evasion to arrive at answers to the two critical questions:

● Down to what size explosion can underground testing be seismically moni-tored with high confidence?

● How accurately can the yields of underground explosions be measured?In doing so, we assessed the contribution that could be made if seismic stationswere located in the country whose tests are to be monitored, and other coopera-tive provisions that a treaty might include. A context chapter (chapter 2) has beenincluded to illustrate how the technical answers to these questions contribute tothe political debate over:

● Down to what yield can we verify Soviet compliance with a test ban treaty?● Is the 1976 Threshold Test Ban Treaty verifiable?● Has the Soviet Union complied with p-resent testing restrictions?In the course of this assessment, OTA drew on the experience of many organi-

zations and individuals. We appreciate the assistance of the project contractorswho prepared background analysis, the U.S. Government agencies and privatecompanies who contributed valuable information, the project advisory panel andworkshop participants who provided guidance and review, and the many addi-tional reviewers who helped ensure the accuracy and objectivity of this report.

u JOHN H. GIBBONSDirector

I l l

Seismic Verification of Nuclear Testing Treaties Advisory Panel

Howard Wesley Johnson, ChairHonorary Chairman of the CorporationMassachusetts Institute of Technology

Walter AlvarezProfessorDepartment of Geology and GeophysicsUniversity of California, BerkeleyThomas C. BacheManager, Geophysics DivisionEarth Sciences OperationScience Applications International Corp.

William E. Colbyformer Director of Central Intelligence

AgencyF. Anthony DahlenProfessorDepartment of Geological and Geophysical

SciencesPrinceton University

H.B. DurhamDistinguished Member of Technical StaffSandia National Laboratories

John R. FilsonChief, Office of Earthquakes, Volcanos,

and EngineeringU.S. Geological SurveyW. J. Hannon, Jr.Assistant Program LeaderAnalysis & AssessmentVerification ProgramLawrence Livermore National LaboratoryUniversity of CaliforniaRoland F. HerbstMember of President’s StaffRDA Associates

Eugene T. Herrin, Jr.ProfessorDepartment of Geological SciencesSouthern Methodist University

Franklyn K. Levinformer Senior Research Scientist, Exxon

Raymond J. McCroryformer Chief, Arms Control Intelligence

StaffCentral Intelligence AgencyKaren McNallyDirector, Charles F. Richter Seismological

LaboratoryUniversity of California, Santa Cruz

John R. MurphyProgram Manager and Vice PresidentS-CUBEDWolfgang K.H. PanofskyDirector EmeritusStanford Linear Accelerator CenterStanford UniversityPaul G. RichardsProfessorDepartment of Geological SciencesLamont-Doherty Geological ObservatoryColumbia University

Jack RuinaDirector, Defense & Arms Control Studies

ProgramCenter for International StudiesMassachusetts Institute of Technology

Lynn R. SykesHiggins Professor of Geological SciencesLamont-Doherty Geological ObservatoryColumbia University

Thomas A. WeaverDeputy Group Leader, Geophysical GroupLos Alamos National LaboratoryUniversity of California

NOTE: OTA appreciates and is grateful for the valuable assistance and thoughtful critiques provided by the advi-sory panel members. The panel does not, however, necessarily approve, disapprove, or endorse this report.OTA assumes full responsibility for the report and the accuracy of its contents.

iv

OTA Project Staff—Seismic Verification of Nuclear Test Ban Treaties

Lionel S. Johns, Assistant Director, OTAEnergy, Materials, and International Security Division

Peter Sharfman, International Security and Commerce Program Manager

Gregory E. van der Vink, Project Director

Administrative Staff

Jannie Home Marie C. Parker Jackie Robinson

Contractors

Steven M. Day Zoltan A. Der Frederick K. Lamb Robert P. Masse´

Acknowledgments

In addition to the advisory panel and contractors, OTA gratefully acknowledgesthe valuable contributions made by the following:

Ralph W. Alewine, III, Defense Advanced Research Projects AgencyShelton S. Alexander, Pennsylvania State University

Charles B. Archambeau, University of ColoradoRobert R. Blandford, Defense Advanced Research Projects Agency

John S. Derr, United States Geological SurveyDon Eilers, Los Alamos National Laboratory

Jack Evernden, United States Geological SurveyFrederick E. Followill, Lawrence Livermore National Laboratory

Robert A. Jeffries, Los Alamos National LaboratoryJames F. Lewkowicz, Air Force Geophysics LaboratoryJ. Bernard Minster, Scripps Institute of Oceanography

Otto W. Nuttli, St. Louis UniversityMary Peters, Naval Research LaboratoryRobert A. Phinney, Princeton University

Keith Priestley, University of NevadaAlan S. Ryan, Center for Seismic Studies

Robert H. Shumway, University of CaliforniaStewart Smith, Incorporated Research Institutions for Seismology

Sean Solomon, Massachusetts Institute of TechnologyJack Rachlin, United States Geological Survey

Irvin Williams, Department of EnergyRobert Zavadil, Air Force Technical Applications Center

Department of Energy, Office of Classification

Workshop I: Network Capabilities

Stewart Smith, ChairPresident, Incorporated Research Institute for Seismology

Charles B. ArchambeauProfessorCooperative Institute for Research in

Environmental ScienceUniversity of Colorado, Boulder

Thomas C. BacheManager, Geophysics DivisionEarth Sciences OperationScience Applications International Corp.

John S. DerrGeophysicistU.S. Geological Survey

Jack EverndenResearch GeophysicistU.S. Geological Survey

Willard J. HannonAssistant Program Leader for Analysis &

AssessmentVerification ProgramLawrence Livermore National LaboratoryUniversity of CaliforniaKeith NakanishiAssistant Program Leader for Seismic

ResearchLawrence Livermore National LaboratoryUniversity of CaliforniaPaul G. RichardsProfessorDepartment of Geological SciencesLamont-Doherty Geological ObservatoryColumbia University

Frederick E. Followill Robert J. ZavadilSeismologist Chief, Evaluation DivisionLawrence Livermore National Laboratory Directorate of GeophysicsUniversity of California Air Force Technical Applications Center

Workshop II: Identification

J. Bernard Minster, ChairVisiting Professor of Geophysics

Scripps Institute of OceanographyUniversity of California, San Diego


Environmental ScienceUniversity of Colorado, BoulderThomas C. BacheManager, Geophysics DivisionEarth Sciences OperationScience Applications International Corp.



AssessmentVerification ProgramLawrence Livermore National LaboratoryUniversity of California

Robert R. Blandford John R. MurphyProgram Manager for the Nuclear Program Manager and Vice President

Monitoring Office S-CUBEDDefense Advanced Research Projects

Agency

vi

Keith PriestleyProfessorSeismological LaboratoryUniversity of Nevada

Jack RachlinGeologistU.S. Geological SurveyPaul G. RichardsProfessorDepartment of Geological SciencesLamont-Doherty Geological ObservatoryColumbia University

Robert H. ShumwayProfessorDivision of StatisticsUniversity of California, DavisLynn R. SykesHiggins Professor of Geological SciencesLamont-Doherty Geological ObservatoryColumbia UniversityRobert J. ZavadilChief, Evaluation DivisionDirectorate of GeophysicsAir Force Technical Applications Center

Workshop III: Evasion

Sean C. Solomon, ChairProfessor, Department of Earth, Atmospheric & Planetary Science

Massachusetts Institute of Technology


Environmental ScienceUniversity of Colorado, BoulderThomas C. BacheManager, Geophysics DivisionEarth Sciences OperationScience Applications International Corp.Robert R. BlandfordProgram Manager for the Nuclear

Monitoring OfficeDefense Advanced Research Projects

Agency


Eugene T. Herrin, Jr.ProfessorDepartment of Geological SciencesSouthern Methodist University

Keith PriestleyProfessorSeismological LaboratoryUniversity of Nevada

Jack RachlinGeologistU.S. Geological SurveyLynn R. SykesHiggins Professor of Geological SciencesLamont-Doherty Geological ObservatoryColumbia University

Willard J. HarmonAssistant Program Leader for Analysis &

AssesmsentVerification ProgramLawrence Livermore National LaboratoryUniversity of California

vii

Workshop IV: Yield Determinations

Robert A. Phinney, ChairChairman, Department of Geological & Geophysical Sciences

Princeton University

Ralph W. Alewine, IIIDirector of Nuclear Monitoring OfficeDefense Advanced Research Projects

Agency


Environmental ScienceUniversity of Colorado, BoulderThomas C. BacheManager, Geophysics DivisionEarth Sciences OperationScience Applications International Corp.Donald EilersAssociate Group LeaderLos Alamos National LaboratoryUniversity of California


AssessmentVerification ProgramLawrence Livermore National LaboratoryUniversity of California

Eugene T. Herrin, Jr.ProfessorDepartment of Geological SciencesSouthern Methodist UniversityRobert A. JeffriesProgram Director for Verification &

SafeguardsLos Alamos National Laboratory

Frederick K. LambProfessorDepartment of PhysicsUniversity of Illinois, UrbanaKeith PriestleyProfessorSeismological LaboratoryUniversity of NevadaPaul G. RichardsProfessorDepartment of Geological SciencesLamont-Doherty Geological ObservatoryColumbia UniversityAlan S. RyallManager of ResearchCenter for Seismic Studies

Robert H. ShumwayProfessorDivision of StatisticsUniversity of California, DavisLynn R. SykesHiggins Professor of Geological SciencesLamont-Doherty Geological ObservatoryColumbia University

Robert J. ZavadilChief, Evaluation DivisionDirectorate of GeophysicsAir Force Technical Applications Center

Vlll

Contents

PageI. Executive Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32. Seismic Verification in the Context of National Security . . . . . . . . . . . . . . . . . 233. The Role of Seismology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 414. Detecting Seismic Events . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 555. Identifying Seismic Events . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 776. Methods of Evading a Monitoring Network . . . . . . . . . . . . . . . . . . . . . . . . . . . . 957. Estimating the Yields of Nuclear Explosions . . . . . . . . . . . . . . . . . . . . . ......113Appendix. Hydrodynamic Methods of Yield Estimation . ...................129

Chapter 1

Executive Summary

CONTENTS

PageIntroduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3The Test Ban Debate.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3The Meaning of Verification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5Aspects of Monitoring Underground Nuclear Explosions . . . . . . . . . . . . . . . 6

Detecting Seismic Events . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6Identifying Seismic Events as Nuclear Explosions.. . . . . . . . . . . . . . . . . . 8Evading a Seismic Monitoring Network. ....... . . . . . . . . . . . . . . . . . . . 11

How Low Can We Go?. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13Leve 1–Above 10 kt . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14Level 2–Below10 kt but Above 1-2 kt . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14Level 3–Below 1-2 kit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14Level 4–Comprehensive Test Ban . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14

Estimating the Yield of Nuclear Explosions. . . . . . . . . . . . . . . . . . . . . . . . . . 15Soviet Compliance. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17Verification of the TTBT and the PNET. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19A Phased Approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19

BoxesBox Pagel-A. NRDC/Soviet Academy of Sciences . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101-B. CORRTEX. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18

FiguresFigure No. Pagel-1. Signal From Semipalatinsk . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7l-2. Explosion and Earthquake . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12

Chapter 1

Executive Summary

Technologies that were originally developed to study earthquakes may nowenable the United States

limitto verify a treaty with the Soviet Union to furtherthe testing of nuclear weapons.

INTRODUCTION

Seismology now provides a means to moni-tor underground nuclear explosions down tolow yields, even when strenuous attempts aremade to evade the monitoring system. By do-ing so, seismology plays a central role in verify-ing arms control agreements that limit the test-ing of nuclear weapons. Seismology, however,is like any other technology: it has both strengthsand limitations. If the capabilities of seismicmonitoring are to be fully realized, it is neces-sary to understand both how the strengths canbe used and how the limitations can be avoided.

To a great extent, the capabilities of anygiven seismic monitoring network are deter-mined by how the monitoring task is approachedand what supplementary provisions are nego-tiated within the treaty. If agreements can benegotiated to reduce uncertainty, then seismol-ogy can be very effective and extremely lowyields could be monitored with high confidence.

This report addresses two key questions:

1. down to what size explosion can under-ground testing be seismically monitoredwith high confidence, and

2. how accurately can the yields of under-ground explosions be measured seismically?

The answers to these questions provide thetechnical information that lies at the heart ofthe political debate over:

1. how low a threshold test ban treaty withthe Soviet Union we could verify,

2. whether the 1976 Threshold Test BanTreaty is verifiable, and

3. whether the Soviet Union has compliedwith present testing restrictions.

Seismic monitoring as discussed in this studyis evaluated without specific references to theparticular treaty regime to which it is to ap-ply. There will always be some limit to the ca-pability of any given monitoring network, andhence there will always be a threshold belowwhich a seismic network could not monitorwith high confidence. Consequently, should atotal test ban be enacted there will be a verylow threshold below which seismic methodscannot provide high confidence monitoring.Such a treaty could still be considered to bein the national interest if, taking both seismicand nonseismic verification methods into ac-count, the significance of undetected violations(if they were to occur) would be outweighedby the benefits of such a treaty.

THE TEST BAN DEBATE

Test ban treaties are a seemingly simple ap- advantages in one area must be weighed againstpreach to nuclear arms control, yet their im- advantages in another. Consequently, all aspectspact is complex and multi-faceted. The deci- of a new treaty must be considered togethersion as to whether a given test ban treaty is and the cumulative impact evaluated in termsin our overall national interest is dependent of a balance with the Soviet Union. Finally,on many questions concerning its effects. Dis- the total net assessment of the effects of a

3

—

4

Photo credit: Department of Energy

Signal cables and test device being lowered down test hole.

treaty on our national security must be weighed of a nuclear deterrent and the extent to whichagainst the alternative: no treaty. arms control can contribute to national secu-

rity. It is perhaps because test ban treaties goOne’s opinion about the effects of a test ban, to the very heart of nuclear weapons policy that

and thus its desirability, is largely dependent the debate over them remains unresolved. Threeon one’s philosophical position about the role decades of negotiation between the United

5

States and the U.S.S.R. have produced onlythree limitation treaties, two of which remainunratified:

1.

2.

3.

1963 Limited Nuclear Test Ban Treaty(LTBT). This treaty bans nuclear explo-sions in the atmosphere, in outer space,and under water. It was signed by theUnited States and the U.S.S.R. on August5, 1963 and has been in effect since Oc-tober 10, 1963. Over 100 other countrieshave also signed this treaty.1974 Threshold Test Ban Treaty (TTBT).This treaty restricts the testing of under-ground nuclear weapons to yields no greaterthan 150 kilotons (kt). It was signed bythe United States and the U.S.S.R. onJuly 3,1974. Although the TTBT has yetto receive the consent of the U.S. Senate,both nations consider themselves obligatedto adhere to it.1976 Peaceful Nuclear Explosions Treaty(PNET). This treaty is a complement tothe TTBT and restricts individual peace-

ful nuclear explosions (PNEs) to yields nogreater than 150 kt, and aggregate yieldsin a salvo to no greater than 1500 kt. Itwas signed by the United States and theU.S.S.R. on May 28,1976. Although PNEThas yet to receive the consent of the U.S.Senate, both nations consider themselvesobligated to adhere to it.

Nuclear explosions compliant with thesetreaties can only be conducted by the Unitedand the U.S.S.R. underground, at specific testsites (unless a PNE), and with yields no greaterthan 150 kt. Although they have had impor-tant positive environmental and arms controlimpacts, these treaties have not prevented thedevelopment of new types of warheads andbombs. For this reason, public interest in acomplete test ban or a much lower thresholdremains strong, and each year a number ofproposals continue to be brought before theCongress to limit further the testing of nuclearweapons.

THE MEANING OF VERIFICATIONFor the United States, the main national

security benefits derived from test limitationtreaties are a result of the Soviet Union beingsimilarly restricted. In considering agreementsthat bear on such vital matters as nuclearweapons development, each country usuallyassumes as a cautious working hypothesis thatthe other parties would cheat if they were suffi-ciently confident that they would not be caught.Verification—the process of confirming com-pliance and detecting violations if they occur—is therefore central to the value of any suchtreaty.

“To verify” means to establish truth or ac-curacy. Yet in the arena of arms control, theprocess of verification is political as well astechnical. It is political because the degree ofverification needed is based upon one’s percep-

tion of the benefits of a treaty compared withone’s perception of its disadvantages and thelikelihood of violations. No treaty can be con-sidered to be either verifiable or unverifiablewithout such a value judgment. Moreover, thisjudgment is complex because it requires notonly an understanding of the capabilities ofthe monitoring systems, but also an assess-ment as to what is an acceptable level of risk,and a decision as to what should constitute sig-nificant noncompliance. Consequently, peoplewith differing perspectives on the role of nu-clear weapons in national security and on themotivations of Soviet leadership will differ onthe level of verification required.l

‘This issue is discussed further in chapter 2, Seism”c Verifi-cation in the Context of National Security.

6

ASPECTS OF MONITORING UNDERGROUNDNUCLEAR EXPLOSIONS

Like earthquakes, the force of an undergroundnuclear explosion creates seismic waves thattravel through the Earth. A seismic monitor-ing network must be able both to detect andto identify seismic signals. Detection consistsof recognizing that a seismic event has oc-curred and locating the source of the seismicsignals. Identification involves determiningwhether the source was a nuclear explosion.In the case of a threshold treaty, the monitor-ing network must also be able to estimate theyield of the explosion from the seismic signalto determine if it is within the limit of thetreaty. All of this must be done with a capa-bility that can be demonstrated to adequatelydefeat any credible attempt to evade the mon-itoring network.

If the seismic signals from explosions are notdeliberately obscured or reduced by special ef-forts, seismic networks would be capable of de-tecting and identify with confidence nuclearexplosions with yields below 1 kt. What stopsthis capability from being directly translatedinto a monitoring threshold is the requirementthat the monitoring network be able to accom-plish detection and identification with highconfidence in spite of any credible evasion sce-nario for concealing or reducing the seismic sig-nal from a test explosion.

Demonstrating that the monitoring capabil-ity meets this requirement becomes complexat lower yields. As the size of the explosionbecomes smaller:

●

●

●

there are more opportunities for evadingthe monitoring network,there are more earthquakes and industrialexplosions from which such small clandes-tine explosions need to be distinguished,andthere are more factors that can stronglyinfluence the seismic signal.

The cumulative effect is that the lower theyield, the more difficult the task of monitor-ing against possible evasion scenarios.

The threat of evasion can be greatly reducedby negotiating within a treaty various coop-erative monitoring arrangements and testingrestrictions. However, there will eventually bea yield below which the uncertainty of anymonitoring regime will increase significantly.The point at which the uncertainties of the mon-itoring system no longer permit adequate veri-fication is a political judgment of the point atwhich the risks of the treaty outweigh the benefits.

Determining the credibility of various eva-sion methods requires subjective judgmentsabout levels of motivation and risk as well asmore objective technical assessments of thecapability of the monitoring system. To sepa-rate the technical capabilities from the subjec-tive judgments, we will first describe our ca-pability to detect and identify seismic eventsand then will show how this capability is lim-ited by various possible evasion methods. Allconsiderations are then combined to addressthe summary question: How low can we go?

Detecting Seismic Events

The first requirement of a seismic networkis that it be capable of detecting a seismicevent. If the Earth were perfectly quiet, thiswould be easy. Modem seismometers are high-ly sophisticated and can detect remarkablysmall motions. Limitations are due not to theinherent sensitivity of the instrument but ratherto phenomena such as wind, ocean waves, andeven rush hour traffic. All of these processescause ground vibrations that are sensed byseismometers and recorded as small-scale back-ground motion, collectively referred to as“noise.” Seismic networks, consisting of groupsof instruments, are designed to distinguishevents like earthquakes and explosions fromthis ever-present background noise. The extentto which a seismic network is capable of de-tecting such events is dependent on many fac-tors. Of particular importance are the typesof seismic stations used, the number and dis-tribution of the stations, the amount of back-

7

ground noise at the site locations, and the effi-ciency with which seismic waves travel fromthe source to the receiving station through thesurrounding geologic area.

Detecting seismic events can be accomplishedwith high certainty down to extremely smallyields. A hypothetical seismic network withstations only outside the Soviet Union wouldbe capable of detecting well-coupled2 explo-sions with yields below 1 kt anywhere withinthe Soviet Union and would be able to detect

2A “well-coupled” explosion is one where the energy is welltransmitted from the explosion to the surrounding rock.

even smaller events in selected regions. Theexisting seismic array in Norway, for exam-ple, has easily detected Soviet explosions withyields of a fraction of a kt conducted 3,800kilometers away at the Soviet test site in East-ern Kazakhstan (figure l-l).

Seismic stations within the Soviet Unionwould further improve the detection capabil-ity of a network. In principle, almost anydesired signal detection level could be achievedwithin the Soviet Union if a sufficient numberof internal stations were deployed. In thissense, the detection of seismic events does notprovide a barrier for monitoring even the lowestthreshold treaties. From a monitoring stand-

Figure 1-1 .-Signal From Semipalatinsk

Noressarraybeam

I I I I I I I I I0 10 20 30 40 50

Time (seconds)

Seismic signal from a 0.25 kt explosion at the Soviet test site near Semipalatinsk. Recorded 3,800 km away at the NORESSseismic array in Norway on July 11, 1985. The signal to noise ratio is about 30 indicating that much smaller explosions couldbe detected even at this great distance.

SOURCE: R.W. Alewire Ill, “Seismic Sensing of Soviet Tests,” Defense 85, December 1985, pp. 11-21.

.

8

point, stations within the Soviet Union aremore important for improving identificationcapabilities than for further reduction of thealready low detection threshold.

Identifying Seismic Events asNuclear Explosions

Once a seismic signal has been detected, thenext task is to determine whether it was cre-ated by a nuclear explosion. Seismic signalsare generated not only by nuclear explosions,but also by natural earthquakes, rockburstsin mines, and chemical explosions conductedfor mining, quarry blasting, and construction.

Every day there are many earthquakesaround the globe whose seismic signals are the

same size as those of potential undergroundnuclear explosions. Several methods can be ap-plied to differentiate earthquakes from under-ground nuclear explosions. Note, however, thatno one method is completely reliable. It is theset of different identification methods takenas a whole and applied in a systematic fashionthat is assessed when summaries on capabil-ity are given. In this sense, identification isa “winnowing” process.

The most basic method of identification isto use the location and the depth of the event.Over 90 percent of all seismic events in theU.S.S.R. can be classified as earthquakes sim-ply because they are either too deep or not ina plausible location for a nuclear explosion. Forseismic events that are in a location and at a

Photo credit: Department of Energy

Craters formed by cavity collapse in Yucca Flat, Nevada Test Site.

9

depth that could bean explosion, other meth-ods of discrimination based on physical differ-ences between earthquakes and explosions areused.

When a nuclear device explodes underground,it applies uniform pressure to the walls of thecavity it creates. As a result, explosions areseen seismically as highly concentrated sourcesof compressional waves, sent out with approx-imately the same strength in all directions fromthe point of the detonated device. An earth-quake, on the other hand, occurs when twoblocks of the Earth’s crust slip past each otheralong a fault. An earthquake generates shearwaves from all parts of the fault that rupture.

These fundamental differences between earth-quakes and explosions are often exhibited intheir seismic signals. As a result, seismologistshave been able to develop a series of methodsto differentiate the two sources based on thedifferent types of seismic waves they create.The combination of all methods, when appliedin a comprehensive approach, can differenti-ate with high confidence between explosions,down to low yields, and earthquakes.

As the size of the seismic events gets smaller,nuclear explosions must be distinguished notonly from earthquakes, but also from otherkinds of explosions. Industrial chemical explo-sions (e.g. in a quarry operation) pose a par-ticularly difficult problem because their seismicsignals have physical characteristics similarto those of nuclear explosions. Consequently,the seismic methods that are routinely usedto differentiate earthquakes from explosionscannot distinguish between some legitimatechemical explosions for mining purposes anda clandestine nuclear test explosion. Fortu-nately, industrial explosions in the range of1 to 10 kt are rare (less than one a year). Largeexplosions are usually ripple fired so as to min-imize ground vibration and fracture rock moreefficiently. Ripple firing is often accomplishedwith bursts spaced about 0.2 seconds apartover a duration of about a second. Recent worksuggests that this ripple-firing has an identifi-able signature apparent in the observed seis-mic signals, and therefore can be used to iden-

tify such chemical explosions. However, theabsence of evidence for ripple firing cannot betaken as evidence that the event is not a chem-ical explosion. Because of the size considera-tion, industrial explosions are not an identifi-cation problem for normal nuclear explosionsabove 1 kt. The difficulty, as we will see later,comes in distinguishing between a small decou-pled nuclear test and a large salvo-fired chem-ical explosion. This difficulty can be limitedthrough such treaty provisions as options forinspections and constraints on chemical ex-plosions. 3

Because a seismic signal must be clearly de-tected before it can be identified, the thresh-old for identification will always be greaterthan the threshold for detection. As describedin the previous section, however, the detectionthreshold is quite low. Correspondingly, evena hypothetical network consisting of stationsonly outside the Soviet Union would be capa-ble of identifying seismic events with magni-tudes corresponding to about 1 kt if no at-tempts were made to evade the monitoringsystem.4 Seismic stations within the SovietUnion would further improve the identifica-tion capability of a network.

It has been argued that the use of high fre-quency seismic data will greatly improve ourcapability to detect and identify low-yield nu-clear explosions.56 Recent experiments con-ducted by the Natural Resources DefenseCouncil together with the Soviet Academy ofSciences are beginning to provide high fre-quency seismic data from within the SovietUnion that shows clear recordings of small ex-plosions (see box l-A). There remains, however,a lack of consensus on the extent to which theuse of higher frequency data will actually im-

3See chapter 6, Eva&-ng a Monitoring Network, for a discus-sion of treaty constraints.

‘See chapter 5, Identifying Seism”c Events.‘For example, much lower identification thresholds have been

defended by J.F. Evernden, C.B. Archarnbeau, and E. Cran-swick, “An Evaluation of Seismic Decoupling and UndergroundNuclear Test Monitoring Using High-Frequency Seismic Data,”Reviews of Geophysics, vol. 24, May 1986, pp. 143-215.

‘See chapter 4, De&ctingSeismz”cE vents, and chapter 5 Iden-tifying Seism”c Events, for discussions of high frequency mon-itoring.

10

Box 1-A.—NRDC/Soviet Academy of SciencesNew seismic data from the Soviet Union is becoming available through an agreement between

the Natural Resources Defense Council and the Soviet Academy of Sciences. The agreement pro-vides for the establishment of a few high-quality seismic stations within the Soviet Union aroundthe area of the Soviet test site in Kazakhstan. The agreement also included experiments in whichthe Soviet Union detonated chemical explosions of known yield near the test site so that the testsite could be calibrated.

The seismograph below is from a 0.01 kt chemical explosion detonated near the Soviet test siteand recorded 255 km away. The signal of the explosion can be clearly seen along with the coinciden-tal arrival of seismic waves caused by a large earthquake that occurred south of New Zealand. Thethree components of ground motion are east-west (E), north-south (N), and vertical (Z).

Start of explosionrecording

Start of earthquakerecord 1

SOURCE: Natural Resources Defense Council.

prove monitoring capabilities. The lack of con- til more extensive data can be collected at re-sensus is due to differences in opinion as to gional distances from areas throughout thehow well U.S. experience and the limited ex- Soviet Union. Nevertheless, there is generalperience near the Soviet test site can be ex- agreement among seismologists that goodtrapolated to an actual comprehensive moni- data is obtainable at higher frequencies thantoring system throughout the Soviet Union. those used routinely today, and that this dataConsequently, the debate over improved ca- offers advantages for nuclear monitoring thatpability will probably remain unresolved un- should continue to be explored.

11

Photo credit: Sand/a National Laboratories

An example of what an internal seismic station might look like.

Evading a Seismic Monitoring Network

To monitor a treaty, nuclear tests must bedetected and identified with high confidenceeven if attempts are made to evade the moni-toring system. As mentioned earlier, it is thefeasibility of various evasion scenarios thatsets the lower limit on the monitoring capa-bility. The major evasion concerns are:

●

●

●

that the signal of an explosion could behidden in the signal of a naturally occur-ring earthquake,that an explosion could be muffled by det-onating it in a large underground cavity,orthat a nuclear test could be disguised asor masked by a large legitimate industrialexplosion.

The hide-in-earthquake scenario assumesthat a small nuclear test could be conducted

by detonating the explosion during or soon af-ter an earthquake. If the earthquake were suffi-ciently large and the small explosion properlytimed, the seismic signal of the explosion wouldbe hidden in the seismic signal of the earth-quake. However, it is not practical for anevader to wait for an earthquake that is in theimmediate vicinity of the test site. Therefore,the masking earthquake would have to be largeand at some distance. The smaller nuclear ex-plosion will produce higher frequency signalsthan the earthquake, and filtering the signalswill reveal the signals from the explosion (fig-ure 1-2). For this and other reasons (chapter6), the hide-in-earthquake scenario need nolonger be viewed as a credible evasion threatfor explosions above 5-10 kt. To counter thisthreat for explosions below 5-10 kt may requireaccess to data from seismic stations within theSoviet Union, because the higher frequencyseismic signals from explosions below 5-10 kt

12

Figure l-2.—Explosion and Earthquake

Conventional Recording Using Low Frequencies

50 100 150Large ½ kiloton

earthquake explosion Time (seconds)

Same Recording But With High Frequency Passband

o

Both the upper andis the conventionalends, the arrival of

50 1(Large

earthquake

I I150 200

½ kilotonexplosion

Time (seconds)

lower seismograms were recorded in Norway and cover the same period of time. The upper seismogramrecording of low-frequency seismic waves. Both it and the lower recording show, at the time of 30 sec-waves from a large earthquake that occurred in the eastern part of the Soviet Union. About one minute

after the earthquake, at a time of 100 seconds, the Soviet Union conducted a very small (about ½ kt) underground nuclearexplosion at their Kazakhstan test site. With the standard filter (upper seismogram), the signal of the explosion appears hid-den by the earthquake. Using a passband filter for higher frequency seismic waves (lower seismogram), the explosion is revealed.

SOURCE: Semiannual Technical Summary for Norwegian Seismic Array for 1984, Royal Norwegian Council for Scientific and Industrial Research, Scientific ReportNo. 1-84/85.

13

may not always be picked up by stations out-side the Soviet Union.

Decoupling appears to be potentially themost effective of all evasion methods. It in-volves detonating a nuclear device in a largeunderground cavity so as to muffle the seis-mic signal. If the explosion occurs in a largecavity, the explosive stresses are reduced be-cause they are spread over the large area ofthe cavity wall. At a certain cavity size, thestresses will not exceed the elastic limit of therock. At such a point, the explosion is said tobe “fully decoupled” and the seismic signal becomes greatly reduced. At low seismic frequen-cies, a fully decoupled explosion may have asignal 70 times smaller than that of a fully cou-pled explosion. At high frequencies the decou-pling factor is probably reduced to somewherebetween 10 and 7.

Above 10 kt, decoupling is not consideredto be a credible evasion scenario because: 1)the clandestine construction of a cavity largeand strong enough to decouple a 10 kt explo-sion is not feasible; and 2) even if such an ex-plosion were somehow fully decoupled, the seis-mic signal would stand a good chance of beingdetected and possibly identified. Therefore,decoupling could be most effective for smallexplosions up to a few kt, particularly whendone in conjunction with a legitimate indus-trial explosion. For example, a potential eva-sion method would be to secretly decouple asmall nuclear explosion in a large undergroundcavity and mask or attribute the muffled sig-nals to a large chemical explosion that is simul-

taneously detonated under the guise of legiti-mate industrial activity.

As discussed in the section on identification,differentiating a small nuclear explosion froma legitimate industrial explosion associatedwith mining and quarry blasting is difficultbecause both produce similar seismic signals.This is not a problem for identification undernormal circumstances because industrial ex-plosions in the 1-10 kt range occur less thanonce a year. However, industrial explosionsmay create a problem when considering thedecoupling evasion scenario. That is, some low-yield decoupled explosions might produce seis-mic signals comparable to those observed fromlarge chemical explosions and no routine ca-pability has yet been developed to differentiate,with high confidence, between such signals.

None of the evasion scenarios poses any seri-ous problem for monitoring explosions above10 kt. However, to provide adequate monitor-ing capability below 10 kt, efforts must bemade to limit decoupling opportunities. Thiswould include an internal seismic network andprovisions within the treaty (such as pre-notifi-cation with the option for on-site inspection)to handle the large numbers of chemical ex-plosions. At a few kt, decoupling becomes pos-sible when considered as part of an evasion scenario. Arranging such a test, however, wouldbe both difficult and expensive; and it is notclear that the evader could have confidencethat such a test (even if it were successful)would go undiscovered. However, the possi-bility remains that a few tests of small magni-tude could be conducted and remain unidenti-fied as nuclear explosions.

HOW LOW CAN WE GO?

Given all the strengths and limitations ofseismic methods in detecting and identifyingSoviet nuclear explosions, combined with thecredibility of the various evasion scenarios, theultimate question of interest for monitoringany low-yield threshold test ban treaty is es-sentially: How low can we go? The answer to

that question depends largely on what is ne-gotiated in the treaty. As we have seen, thechallenge for a monitoring network is to dem-onstrate a capability to distinguish credibleevasion attempts from the background of fre-quent earthquakes and legitimate industrialexplosions that occur at low yields.

14

The monitoring burden placed on the seis-mic network by various evasion scenarios canbe greatly lessened if seismology gets somehelp. The sources of help are varied and numer-ous: they include not only seismic monitoringbut also other methods such as satellite sur-veillance, radioactive air sampling, communi-cation intercepts, reports from intelligenceagents, information leaks, interviews withdefectors and emigres, on-site inspections, etc.The structure of any treaty or agreement shouldbe approached through a combination of seis-mic methods, treaty constraints, and inspectionsthat will reduce the uncertainties and difficul-ties of applying seismic monitoring methods toevery conceivable test situation. Yet with theseconsiderations in mind, some generalizationscan still be made about monitoring at variouslevels.

Level l–Above 10 kt

Nuclear tests with explosive yields above 10kt can be readily monitored with high confi-dence.7 This can be done with external seismicnetworks and other national technical means.The seismic signals produced by explosions ofthis size are discernible and no method of evad-ing a seismic monitoring network is credible.However, for accurate monitoring of a 10 ktthreshold treaty it would be desirable to havestations within the Soviet Union for improvedyield estimation, plus treaty restrictions forhandling the identification of large chemicalexplosions in areas where decoupling couldtake place.

Level 2—Below 10 kt butAbove 1-2 kt

Below 10 kt and above 1-2 kt, the monitor-ing network must demonstrate a capability todefeat evasion scenarios. Constructing an un-derground cavity of sufficient size to fullydecouple an explosion in this range is believedto be feasible in salt, with dedicated effort and

7The United States and the Soviet Union presently restricttheir testing to explosions with yields no greater than 150 kt.The bomb dropped on Hiroshima was estimated to have a yieldof 13 kt.

resources. Consequently, the signals from ex-plosions below 10 kt could perhaps be muffled.The seismic signals from these small muffledexplosions would need not only to be detected,but also distinguished from legitimate chemi-cal industrial explosions and small earthquakes.Demonstrating a capability to defeat credibleevasion attempts would require seismic sta-tions throughout the Soviet Union (especiallyin areas of salt deposits), negotiated provisionswithin the treaty to handle chemical explo-sions, and stringent testing restrictions to limitdecoupling opportunities. If such restrictionscould be negotiated, most experts believe thata high-quality, well run network of internal sta-tions could monitor a threshold of around 5kt. Expert opinion about the lowest yields thatcould reliably be monitored ranges from 1 ktto 10 kt; these differences of opinion stem fromdiffering judgments about what technical pro-visions can be negotiated into the treaty, howmuch the use of high frequencies will improveour capability, and what levels of monitoringcapability are necessary to give us confidencethat the Soviet Union would not risk testingabove the threshold.

Level 3–Below 1-2 kt

For treaty thresholds below 1 or 2 kt, theburden on the monitoring country would bemuch greater. It would become possible todecouple illegal explosions not only in saltdomes but also in media such as granite, allu-vium, and layered salt deposits. Although itmay prove possible to detect such explosionswith an extensive internal network, there isno convincing evidence that such events couldbe confidently identified with current technol-ogy. That is, additional work in identificationcapability will be required before it can be de-termined whether such small decoupled explo-sions could be reliably differentiated from thebackground of many small earthquakes androutine chemical explosions of comparablemagnitude.

Level 4—Comprehensive Test Ban

There will always be some threshold belowwhich seismic monitoring cannot be accom-

15

plished with high certainty. A comprehensive weigh the significance of any undetected clan-test ban treaty could, however, still be consid- destine testing (should it occur) below the mon-ered adequately verifiable if it were determined itoring threshold.that the advantages of such a treaty would out-

ESTIMATING THE YIELD OF NUCLEAR EXPLOSIONS

For treaties that limit the testing of nuclearweapons below a specific threshold, the moni-toring network not only must detect and iden-tify a seismic event such as a nuclear explosionbut also must measure the yield to determinewhether it is below the threshold permitted bythe treaty. This is presently of great interestwith regard to our ability to verify compliancewith the 150-kt limit of the 1974 ThresholdTest Ban Treaty.

The yield of a nuclear explosion maybe esti-mated from the seismic signal it produces.Yield estimation is accomplished by measur-ing from the seismogram the size of an identi-fied seismic wave. When corrected for distanceand local effects at the recording station, thismeasurement is referred to as the seismic mag-nitude. The relationship between seismic mag-nitude and explosive yield has been determinedusing explosions of known yields. This rela-tionship is applied to estimate the size of un-known explosions. The problem is that the rela-tionship was originally determined from U.S.and French testing and calibrated for the Ne-vada test site. As a result, Soviet tests aremeasured as if they had been conducted at theNevada test site unless a correction is made.No correction would be needed if the U.S. andSoviet test sites were geologically identical,but they are not.

The Nevada test site, in the western UnitedStates, is in a geologically young and activearea that is being deformed by the motion be-tween the North American and Pacific tectonicplates. This recent geologic activity has cre-ated an area of anomalously hot and possiblyeven partially molten rock beneath the Nevadatest site. As a result, when an explosion oc-curs at the Nevada test site, the rock deep be-neath Nevada absorbs a large proportion ofthe seismic energy. The Soviet test site, on the

other hand, is more similar to the geologyfound in the eastern United States. It is a geo-logically old and stable area, away from anyrecent plate tectonic activity. When an explo-sion occurs at the Soviet test site, the cold,solid rock transmits the seismic energy strongly.As a consequence, waves traveling from themain Soviet test site in Eastern Kazakhstanappear much larger than waves traveling fromthe Nevada test site. Unless that differenceis taken into account, the size of Soviet explo-sions will be greatly overestimated.

The geological difference between the testsites can result in systematic error, or “bias,in the way that measurements of seismic wavesare converted to yield estimates. Random er-ror is also introduced into the estimates by themeasurement process. Thus, as with any meas-urement, there is an overall uncertainty asso-ciated with determining the size of an under-ground nuclear explosion. This is true whetherthe measurement is being made using seismol-ogy, hydrodynamic methods, or radiochemicalmethods. It is a characteristic of the measure-ment. To represent the uncertainty, measure-ments are presented by giving the most likelynumber (the mean value of all measurements)and a range that represents both the randomscatter of the measurement and an estimateof the systematic uncertainty in the interpre-tation of the measurements. It is most likelythat the actual value is near the central num-ber and it is increasingly unlikely that the ac-tual number would be found towards either endof the scatter range. When appropriate, the un-certainty range can be expressed by using whatis called a “factor of uncertainty. ” For exam-ple, a factor of 2 uncertainty means that thebest estimate of the yield (the “measured cen-tral value”) when multiplied or divided by 2,defines a range within which the true yield willfall in 95 out of 100 cases. This is the high de-

76

gree of certainty conventionally used in seis-mology and may or may not be appropriatein a verification context.8

The uncertainty associated with estimatesof the systematic error can be greatly reducedby negotiating provisions that restrict testingto specific test sites and by calibrating the testsites. If such calibration were to be an integralpart of any future treaty, the concern over sys-tematic errors of this kind would become min-imal. The majority of errors that would remainwould be random. As discussed in chapter 2,a country considering cheating could not takeadvantage of the random error, because itwould not be possible to predict how the ran-dom error would act on any given evasion at-tempt. In other words, if a country attemptedto test above the threshold, it would have torealize that with every test the chances of ap-pearing in compliance would decrease and atthe same time the chance that at least one ofthe tests would appear in unambiguous viola-tion would increase. For this reason, the rangeof uncertainty should not be considered as arange within which cheating could occur.

Most of the systematic error associated withestimating the yields of Soviet nuclear explo-sions is due to geological differences betweenthe U.S. and Soviet test sites and in the coup-ling of the explosion to the Earth. Therefore,the single most important thing that can bedone to reduce the uncertainty in yield esti-mation is to calibrate the test sites. Calibra-tion could be accomplished through an ex-change of devices of known yield, or throughindependent measures of the explosive yieldsuch as can be provided by radiochemical orhydrodynamic methods (See box l-B).

Our present capability to estimate seismi-cally the yields of Soviet explosions is oftencited as a factor of two.9 While this may re-flect present operational methods, it is not anaccurate representation of our capability. Our

‘See chapter 2, Seismic Verification in the Context of NationalSecurity.

‘U. S. Department of State, “Verifying Nuclear Testing Limi-tations: Possible U.S.-Soviet Cooperation, Special Report No.152, ” Aug. 14, 1986.

capability could be greatly improved by incor-porating new methods of yield estimation.Most seismologists feel that if new methodswere applied, the resulting uncertainty formeasuring explosive yields in the range of 150kt at the Soviet test site would be closer toa factor of 1.5 than a factor of two.10 Presentmethods are stated to be accurate only to afactor of two in part because they have not yetincorporated the newer methods of yield esti-mation that use surface waves and Lg waves. 11

The uncertainty of this comprehensive ap-proach could be further reduced if calibrationshots were performed and testing were re-stricted to areas of known geologic composi-tion. It is estimated that through such measures,the uncertainty in seismically measuring Soviettests could be reduced to a level comparable tothe uncertainty in seismically measuring U.S.tests. An uncertainty factor of 1.3 is the currentcapability that seismic methods are able toachieve for estimating yields at the Nevada testsite.

As with detection and identification, yieldestimation becomes more difficult at low yields.Below about 50 kt, high-quality Lg-wave sig-nals can only be reliably picked up by stationswithin the Soviet Union less than 2,000 kmfrom the test site. For explosions below 10 kt,the uncertainty increases because small explo-sions do not always transmit their signals effi-ciently to the surrounding rock. For small ex-plosions, the uncertainty could be reduced byrestricting such tests to depths below the water

iOThis reduction of uncertainly derives from using More thanone statistically independent method. Consider as an examplethe situation where there are three independent methods of cal-culating the yield of an explosion, all of which (for the sake ofthis example) have a factor of two uncertainty in a log normaldistribution:

# of Methods Resulting Uncertainty1 2.02 1.63 1.5

By combining methods, the uncertainty can be reduced be-low the uncertainty of the individual methods. This methodol-ogy, however, can only reduce random error, Systematic error,such as differences between the test site, will remain and limitthe extent to which the uncertainty can be reduced unless cali-bration is performed.

‘‘See chapter 7, Estimating the Yields of Nuclear Explosions.

17

table, so that their signal will be transmitted the capability of seismic methods could be in-effectively. proved to a point comparable to the accuracy

Our capability to estimate explosive yieldswould depend to a large degree on the approachthat was taken and what was negotiated in fu-ture treaties. If new methods of yield determi-nation were incorporated into the measure-ments and calibration shots were performed,

of other methods, such as CORRTEX, that re-quire a foreign presence and equipment at thetest site. In any case, if the objective of reduc-ing the uncertainty is to reduce the opportu-nities for cheating, small differences in randomuncertainty do not matter.

SOVIET COMPLIANCEThe decision as to what constitutes adequate

verification should represent a fair assessmentof the perceived dangers of non-compliance.This necessarily involves a weighing of the ad-vantages of the treaty against the feasibility,likelihood, and significance of noncompliance.Such decisions are subjective and in the pasthave been influenced by the desirability of thetreaty and the political attractiveness of par-ticular monitoring systems.12 Specific concernover compliance with test ban treaties has beenheightened by findings by the Reagan Admin-istration that:

“Soviet nuclear testing activities for a num-ber of tests constitute a likely violation of le-gal obligations under the Threshold Test BanTreaty.”13

Although the 1974 Threshold Test BanTreaty and the 1976 Peaceful Nuclear Explo-sions Treaty have remained unratified for over10 years, both nations have expressed their in-tent to abide by the yield limit. Because nei-ther the United States nor the Soviet Unionhas indicated an intention not to ratify the trea-ties, both parties are obligated under interna-tional law (Article 18, the 1969 Vienna Con-vention on the Law of Treaties) to refrain fromacts that would defeat their objective andpurpose.

“%e chapter 2, Seisnn”c Verification in the Context of Na-tional Security.

“’’The President’s Unclassified Report on Soviet Noncom-pliance with Arms Control Agreements, ” transmitted to theCongress, Mar. 10, 1987.

In examining compliance with the 150-ktthreshold, seismic evidence is currently con-sidered the most reliable basis for estimatingthe yields of Soviet underground nuclear ex-plosions.14

The distribution of Soviet tests in-dicates that about 10 (out of over 200) Sovietexplosions since the signing of the ThresholdTest Ban Treaty in 1974 could have estimatedyields with central values above the 150 ktthreshold limit, depending on how the estimateis made.15 These 10 tests could actually be ator below the 150 kt limit, but have higher yieldestimates due to random fluctuations in theseismic signals. In fact, when the same meth-ods of yield estimation are applied to U.S. tests,approximately the same number of U.S. testsalso appear to be above the 150 kt thresholdlimit. These apparent violations, however, donot mean that one, or the other, or both coun-tries have cheated; nor does it mean per se thatseismology is an inadequate method of yieldestimation. It is inherent in any method ofmeasurement that if several tests are per-formed at the limit, some of these tests willhave estimated central values above the yieldlimit. Because of the nature of measurements(using any method), it is expected that abouthalf the Soviet tests at 150 kt would be meas-ured as slightly above 150 kt and the other halfwould be measured as slightly below 150 kt.

“Conclusion of the Defense Intelligence Agency Review Panelas stated in a letter from Roger E. Batzel, Director of LawrenceLivermore National Laboratory to Senator Claiborne Pen onFeb. 23, 1987.

*’See chapter 7, Estimating the Yields of Nuclear Explosions.

18

IBox 1-B.–CORRTEX

CORRTEX (Continuous Reflectometry for Radius versus Time Experiments) is a technique thatwas developed in the mid-1970s to improve yield estimation using non-seismic (hydrodynamic) meth-ods. In the CORRTEX technique, a satellite hole is drilled parallel to the emplacement hole of thenuclear device and an electrical sensing cable is lowered down the hole (see figure). When the explo-sion occurs, a shock wave moves outward crushing and electrically shorting the cable.

By measuring with electronic equipment at the surface the rate at which the cable is shortedout, the rate of expansion of the shock wave can be calculated. From the rate of expansion of theshock wave and the properties of the surrounding medium, the yield of the nuclear device can beestimated. A full assessment of this method is presented in the Appendix, Hydrodynamic Methodsof Yield Estimation.

CORRTEX CORRTEXrecorder recorderr 1 t

Surface 1 t Surface

-

S e n s i n g

c a b l e

o s t i c

point dTypical cable emplacement

in satellite hole

‘T-Sensingcable

//I

\Y

point

\

\

\

/\\

Shock front /progression

Mov ing shock wave f r o mnuclear detonation

crushes and shortens cable

19

All of the estimates of Soviet and U.S. testsare within the 90 percent confidence level thatone would expect if the yields were 150 kt orless. Extensive statistical studies have exam-ined the distribution of estimated yields of ex-plosions at Soviet test sites. These studies haveconcluded that the Soviets are observing a yield

limit consistent with compliance with the 150kt limit of the Threshold Test Ban Treaty.16 17

‘81bid.ITone of the fir9t ~oupS to carry out such statistical studies

was Lawrence Livermore National Laboratory. Their conclu-sion was reported in open testimony before the Senate ArmedServices Committee on Feb. 26, 1987 by Dr. Milo Nordyke,Leader of the Treaty Verification Program.

VERIFICATION OF THE TTBT AND THE PNETAs noted above, the 1974 Threshold Test

Ban Treaty and the 1976 Peaceful Nuclear Ex-plosions Treaty have not been ratified. Mostrecently, the Senate failed to consent to ratifi-cation at least in part because of concerns thatthe size of Soviet explosions cannot be meas-ured with adequate confidence.18 As a result,for over 10 years the United States has con-tinued to abide by the treaties, yet refused toratify them, ostensibly because they cannotbe adequately verified. Note that if the trea-ties were ratified, the protocols would comeinto force calling for an exchange of data which,if validated, would then improve the verifica-tion.19 20 However, the treaty contains no pro-visions for an independent verification of thedata, most importantly, the yields of the cali-bration shots. Therefore, many experts ques-tion the value of such data, unless the data canbe validated.

As a solution to the problem of uncertaintyin yield estimation, the administration has re-cently proposed the use of an on-site measure-ment system called CORRTEX for all testsabove 50 kt. The CORRTEX system is stated

‘“Threshold Test Ban Treaty and Peaceful NucJear ExplosionsTreaty, Hearings before the Senate Committee on Foreign Re-lations, Jan. 13 and 15, 1987, S. Hrg. 100-115.

‘The protocol of the TTBT calls for an exchange of geologi-cal data plus two calibration shots from each geophysically dis-tinct testing area.

‘“Monitoring would also be improved by the data that wouldbe obtained from all PNE shots that have occurred in the So-viet Union since 1976.

to have a factor of 1.3 uncertainty for measur-ing yields greater than 50 kt at the Soviet testsite.21 22 The drawbacks are that it requires pre-notification and cooperation of the host coun-try to the extent that foreign personnel andtheir equipment must be allowed at the testsite for each test. Also, CORRTEX has limitedapplication for monitoring a low-yield treatyand none for detecting clandestine testing, andso it would not improve our ability to monitorlow-yield testing thresholds.

Alternatively, advanced seismic methodscould be used. The advantage of seismic meth-ods is that a continued presence of foreign per-sonnel at the test site would not be necessary.Additionally, our ability to monitor all Soviettesting (not just testing above 50 kt) wouldbe improved. If the Soviet test site wascalibrated and advanced seismic methods wereutilized, the uncertainty in seismic yield estima-tion could be reduced to a level comparable toCORRTEX. In fact, CORRTEX could be usedto confirm independently the yields of the cal-ibration shots. The Soviet Union has alreadyagreed to the use of CORRTEX for one or twosuch explosions at the Soviet test site.23

*’U. S. Department of State, Bureau of Public Affairs, “U.S.Policy Regarding Limitations on Nuclear Testing, Special Re-port No. 150, ” August 1986.

“see appendix, Hydrodynm”c Methods of Yield Estimation.‘sStatement of the Secretary of State of the United States

and Minister of Foreign Affairs of the Soviet Union, Dec. 9, 1987.

A PHASED APPROACH

If the policy decision were made that trea-ties further restricting or eliminating the test-ing of nuclear weapons are in our national in-

terest, then from a verification viewpoint thereis much to be said for a phased approach tothis goal. Conceptually, it would begin with

20

a limit that can be monitored with high confi-dence using current methods, but would estab-lish the verification network for the desiredlowest level. The threshold would then be lo-wered as information, experience, and confi-dence increase.

For example, the United States and the So-viet Union could begin with a treaty that pro-hibited testing above 10 kt. This is a level thatcan currently be well monitored seismically .24The verification network negotiated within thetreaty, however, would be designed with thegoal of monitoring down to the 1-2 kt level.This would include a network of advanced seis-mic stations throughout the Soviet Union todetect off-site testing plus negotiated provi-sions to reduce evasion possibilities.25 Principalamong these would be to restrict testing to be-low the water table and to a specified test sitewhere decoupling opportunities would be lim-ited, to require some special handling (such aspre-notification and on-site inspection) for thedetonation of large chemical explosions, andto institute measures to confirm a prohibitionon decoupling. To reduce the systematic un-certainty in yield estimation, a series of cali-bration shots would need to be conducted ateach test site using either an independentmethod such as CORRTEX or the exchangeof devices of known yield.

24some exWrts be~eve that decoupling is not feasible above5 kt and consequently, that a 5 kt threshold could be well moni-tored with existing methods and facilities; while others wouldplace the threshold somewhere between 5 and 10. However, vir-tually all experts agree that tests above 10 kt can be well moni-tored, even assuming the monitored country is intent oncheating.26CJ& c h a p t e r G , Eva&”ng a Mom”toring Network.

After such a network had been in operationfor some time, many of the disagreements con-cerning hypothetical networks would be re-solved. It would be known how well seismicwaves travel through the geology of the So-viet Union and what noise levels exist in vari-ous areas. Through such a process, experiencewith large-scale monitoring would be gainedand there would be more accurate knowledgeof what level of monitoring effort is needed.After this information and experience had beenobtained, provisions within the treaty couldcall for the further reduction of the threshold.At that point, it would be better known downto what level monitoring could be accom-plished. If it were determined that a lowerthreshold could be verified, the testing thresh-old could be set to the new verifiable limit. Bythe continuation of periodic reviews, the treatywould always be able to use developments inseismology to maximize the restriction of nu-clear testing.

This procedure, however, does not take intoaccount any considerations other than seismicverification. It simply presents the maximumrestrictions that could be accomplished froma seismic verification standpoint. Considera-tions other than seismic verification may re-sult indifferent thresholds being more desira-ble. Lower thresholds, or even a complete banon testing, may be chosen if the political ad-vantages are seen to outweigh the risk, andif the significance of minor undetected cheat-ing is seen to be small when all monitoringmethods are considered. Higher thresholdsmay be chosen to permit certain types of test-ing or to avoid placing a threshold at a bound-ary where particularly significant tests couldoccur at yields only slightly above thethreshold.

Chapter 2

Seismic Verification in the Contextof National Security

CONTENTS

TheTheAre

PageRole of Verification. . . . . . . . . . . . . . . . . . . . . ... , . . . . . . . . . . . . . . . . . . 23Definition and Value Judgment of “Verification” . . . . . . . . . . . . . . . . . 24Test Ban Treaties in Our National Interest?. . . . . . . . . . . . . . . . . . . . . . 25

Evaluating the Capability of a Verification Network. . . . . . . . . . . . . . . . . . . 29Uncertainty and Confidence Levels, , . . . . . . . . . . . . . . . . . . . . . . . . . . . , , . 29The Relationship Between Uncertainty and Cheating Opportunities . . . . 33What Constitutes Adequate Verification? . . . . . . . . . . . . . . . . . . . . . . . . . . 35

The Question of Determining Compliance . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36

BoxesBox Page2-A. First Interest in Test Ban Treaties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 252-B. Recent Interest in Nuclear Test Limitations. . . . . . . . . . . . . . ., . . . . . . 28

FiguresFigure No. Page2-1. Nuclear Testing, July 16, 1945-December 31, 1987 . . . . . . . . . . . . . . . . . 272-2. Measurements of 150 kt With Factor-of-2 Uncertainty . . . . . . . . . . . . . 322-3. Fifty-Percent Confidence Level for Measurements of 150 kt With

Factor-of-2 Uncertainty . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32

TableTable No. Page2-1. Confidence Intervals for an Explosion Measured at 150 kt . . . . . . . . . 32

Chapter 2

Seismic Verification in the Context ofNational Security

Seismic monitoring is central to considerations of verification, test bantreaties, and national security.

THE ROLE OFFor an arms control agreement to be success-

ful, each participating country must feel thatthe provisions of the treaty will enhance itsown national security. This requires an evalu-ation by each country of the costs and bene-fits to its national security of the treaty’srestrictions. In the case of nuclear test bantreaties, the cost is accepting restrictions onthe ability to test nuclear weapons. In returnfor paying this price, each country gains thedirect benefit of similarly restricting the othercountry. In addition to the direct benefits ofthe agreement, participating countries can alsogain the political and non-proliferation bene-fits of working for arms control, as well as theenvironmental benefits of reducing the hazardsof radioactive contamination of the environ-ment from testing.

In considering agreements that bear on suchvital matters as a restriction on nuclear weap-on’s development, each country may assumeas a cautious working hypothesis that the par-ticipants would cheat if they were sufficientlyconfident that they would not be caught. Ver-ification is a process that is undertaken to con-firm treaty compliance and, therefore, the abil-ity of the United States to monitor Sovietactivity is central to the value of any suchtreaty.

Verification is most often viewed as a proc-ess that improves confidence in a treaty. Theconverse, however, is also true. Establishinga treaty generally improves our ability to mon-itor Soviet activity. In this way, monitoring

VERIFICATION

and treaties have a mutually beneficial rela-tionship. The United States monitors Sovietweapons developments whether or not a treatyexists, because the information is importantfor our national security. Treaties make mon-itoring easier and more accurate, because theyinclude provisions explicitly intended to aidverification. Additionally, treaties create pathsof communication that can be used to resolve,clarify, or correct ambiguous situations. In thissense, treaties have national security valuethat extends beyond their direct purpose.

Verification of treaties is a complex process.The question of whether a treaty is “verifiable”cannot be answered in any absolute or techni-cal sense. It can only be answered relativelyby referring to values that are influenced bya wide range of political and philosophical view-points. Consequently, verification involves notonly technical considerations, but also judge-ments as to how these technical considerationstranslate into the policy world.

In the past several years, Congress has beenasked to consider proposals for treaties thatprohibit testing above various thresholds.Each proposal has sparked controversy withinboth the technical and policy communities. Forany given treaty, some within both communi-ties will claim it is “verifiable,” whereas otherswill assert that the Soviet Union would be ableto cheat and hence that stricter verificationprovisions are needed to ensure our nationalsecurity. This chapter is intended to providea framework for understanding how to weighthe risks and benefits of such treaties.

23

76-584 0 - 88 : QL 3 - 2

24

THE DEFINITION AND VALUE JUDGMENT OF“VERIFICATION”

In the case of test ban treaties, measures aretaken to ensure that the advantages of thetreaty cannot be undermined by the othercountry testing clandestinely. These measures,assessments of the measures’ capabilities,evaluations of the risks and benefits, and thepolitical climate within which all of these judg-ments are made make up the process referredto as verification.

“To verify” means to establish truth or ac-curacy. Realistically, in the arena of arms con-trol, verification can never be perfect or abso-lute: it necessarily involves uncertainty andthis is often described in probabilistic terms.Because the process of verification involves de-termining acceptable levels of uncertainty, itis political as well as scientific. The degree ofverification needed is based on one’s percep-tion of the benefits of the treaty compared withone’s perception of the disadvantages and thelikelihood of violations. Consequently, the levelof verification required will always be differentfor people with different perspectives.

In U.S.-Soviet agreements, the concern aboutverification is exacerbated by societal asym-metries whereby monitoring compliance is usu-ally achieved more easily in the United Statesthan in the Soviet Union. These asymmetriesmay cause the United States to insist on stric-ter verification procedures than the Sovietswould judge are needed. This difference makesnegotiations difficult, and can create the im-pression that the United States is obstruct-ing negotiations.

A country considering cheating would haveto evaluate the risks and costs of being caughtagainst the benefits of succeeding. A countryconcerned about preventing cheating has toguess the other country’s values for makingthis decision and then evaluate them againsttheir own estimations of the advantages of the

treaty compared with the risk of violation. Ifthe countries lack insight into each other’svalue systems and decision processes, this un-certainty will result in the perception that ahigh degree of verification is needed. As a re-sult, the degree of verification needed to satisfythe concerned country may be higher thanwhat is really needed to discourage cheating.

To illustrate this argument, it is useful toconsider the analogy of a treaty restrictingeach party to one side of a river. If the riverfreezes over, one or both countries may con-sider crossing to the other side. If the wateris deep and there is nothing worth having onthe other side, then the ice does not have tobe very thin to discourage a party from cross-ing. If, on the other hand, the water is shallowand there is something of great value to be ob-tained from the other side, than the ice mustbe very thin to discourage a party from cross-ing. The thinness of the ice combined with thedepth of the water is the degree of deterrentavailable to dissuade a party from trying tocross the ice. How thin it has to be to actuallydeter depends on each party’s perception ofthe risk and the reward. In arms control, cross-ing the ice represents cheating on a treaty. Thelevel of verification capability needed to detercrossing (the thinness of the ice) depends oneach side’s perception of the risks and rewardsof cheating. The attraction of cheating (get-ting to the other side) would be the belief thatit could result in some sort of advantage thatwould lead to a significant improvement to thecountry’s national security. The consequenceof being caught (falling through the ice) woulddepend on the depth of the water. This wouldinvolve international humiliation, the possibleabrogation of the treaty resulting in the lossof whatever advantages the treaty had pro-vided, and the potential loss of all other presentand future agreements.

25

ARE TEST BAN TREATIES IN OUR NATIONAL INTEREST?

Test Ban treaties are a seemingly simple ap-proach to arms control, yet their impact is com-plex and multi-faceted. Determining the advan-tages of such a treaty depends upon weighingsuch questions as:

●

●

●

●

●

●

●

●

●

●

Is testing necessary to develop futureweapon systems? Do we want both theUnited States and the Soviet Union to de-velop new weapon systems?Is testing necessary to ensure a high de-gree of reliability of the nuclear stockpile?Do we want the nuclear arsenals of boththe United States and Soviet Union to behighly reliable?Is continued testing necessary to main-tain high levels of technical expertise inthe weapons laboratories? Do we want tocontinue high levels of expertise in boththe United States and Soviet weapons lab-oratories, and if so, for what purposes?Is testing necessary to ensure the safetyof nuclear devices?Could more conservative design practicesreduce the need for nuclear testing?Would the effects of a test ban impact theUnited States and the Soviet Union dif-ferently?Would a decrease in confidence in nuclearweapons’ performance increase or decreasethe likelihood of nuclear war?Would a test ban treaty discourage nu-clear proliferation? Could it be extendedto cover other nations?Would the effects of a treaty be stabiliz-ing or destabilizing?Overall, do the advantages outweigh thedisadvantages?

Due to the immense uncertainties associatedwith nuclear conflict, there are few definitiveanswers to these questions. One’s opinionabout the answers is largely dependent on one’sphilosophical position about the role of a nu-clear deterrent, and the extent to which armscontrol can contribute to national security.None of these questions, moreover, can be con-sidered in isolation. Disadvantages in one areamust be weighed against advantages in another.

Consequently, all aspects of a new treaty mustbe considered together and their cumulativeimpact evaluated in terms of a balance withthe Soviet Union. Such a net assessment is dif-ficult because even greater uncertainty is in-troduced when we try to guess how a given

Box 2-A.—First Interest in Test Ban TreatiesInterest in restricting the testing of nuclear

weapons began with an incident that occurredover 30 years ago. On February 26, 1954, anexperimental thermonuclear device, namedBravo, was exploded on the Bikini Atoll inthe Pacific Ocean. The explosion was theUnited States’ 46th nuclear explosion. Itproduced a yield equivalent to 15 million tonsof TNT, which was over twice what was ex-pected. The radioactive fallout covered anarea larger than anticipated and accidentlycontaminated an unfortunate Japanese fish-ing boat named Lucky Dragon. When theboat docked at Yaizu Harbor in Japan, twenty-three of the crew had radiation sickness re-sulting from fallout. The captain of the ves-sel, Aikichi Kuboyana, died of leukemia inSeptember 1954. In another such accident,radioactive rain caused by a Soviet hydrogenbomb test fell on Japan. These incidents fo-cused worldwide attention on the increasedlevel of nuclear testing and the dangers ofradioactive fallout. Soon after, the first pro-posal for a test ban was put forth.* The 1954proposal presented by India’s Prime Minis-ter Jawaharlal Nehru was described as:

. . . some sort of what maybe called “stand-still agreement” in respect, at least, of theseactual explosions, even if the arrangementsabout the discontinuance of production andstockpiling must await more substantialagreements among those principally con-cerned.Since that time over 1,600 nuclear explo-

sions have occurred and at least four morecountries (United Kingdom, France, People’sRepublic of China, and India) have success-fully tested nuclear devices.*See Bmce A. Bolt, Nuclear Explosions and Earthquakes, W.H.

Freeman and Company, 1976.

26

treaty would affect the Soviet Union. Finally,the total net assessment of the effects of atreaty on our national security must be weighedagainst the alternative: no treaty.

The first formal round of negotiations on acomprehensive test ban treaty began on Oc-tober 31,1958 when the United States, the So-viet Union, and the United Kingdom opened,in Geneva, the Conference on the Discontinu-ance of Nuclear Weapon Tests. Since then, in-terest in a test ban treaty has weathered threedecades of debate with a level of intensity thathas fluctuated with the political climate.1 Dur-ing this time, three partial nuclear test limita-tion treaties were signed. Nuclear explosionscompliant with these restrictions are now con-ducted only underground, at specific test sites,and at yield levels no greater than 150 kilo-tons (kt). These three treaties are:

1.1963 Limited Nuclear Test Ban Treaty(LTBT). Bans nuclear explosions in theatmosphere, outer space, and under water.This treaty was signed August 5, 1963.Ratification was advised and consentedto by the United States Senate on Sep-tember 24, 1963 and the treaty has beenin effect since October 10, 1963.

2. Threshold Test Ban Treaty (TTBT). Re-stricts the testing of underground nuclearweapons by the United States and SovietUnion to yields no greater than 150 kt.This treaty was signed July 3,1974. It wassubmitted to the United States Senate foradvice and consent to ratification on July29, 1976 and again on January 13, 1987.It remains unratified, but both nationsconsider themselves obligated to adhereto it.

3. Peaceful Nuclear Explosions Treaty (PNE).This treaty is a complement to the TTBT.It restricts individual peaceful nuclear ex-plosions by the United States and Soviet

‘For an overview of the history of test ban negotiations, thereader is referred to G. Allen Greb, “Comprehensive Test BanNegotiations 1958-1986: An Overview, ” in IVuclear WeaponZ’ests: Prohibition or Lizm”tation?, edited by Jozef Goldblat andDavid Cox, SIPRI, CIIPS, Oxford University Press, London,1987.

Union to yields no greater than 150 kt,and aggregate yields to no greater than1,500 kt. This treaty was signed May 28,1976. It was submitted to the UnitedStates Senate for advice and consent toratification on July 29, 1976 and again onJanuary 13, 1987. It remains unratified,but both nations consider themselves ob-ligated to adhere to it.

Although these treaties have fallen far shortof banning nuclear testing, they have had im-portant environmental and arms control im-pacts. Since 1963, no signatory country com-pliant with these treaties has tested nuclearweapons in the atmosphere, in outer space, orunder water, thus eliminating a major environ-mental hazard. And from an arms control per-spective, testing of warheads over 150 kt hasbeen prohibited since 1974.

While these treaties have had importantpositive impacts, figure 2-1 illustrates that, infact, they have not resulted in any decline inthe amount of testing. The development of newtypes of warheads and bombs has not beenlimited by restricting testing, and so advocatesof test ban treaties continue to push for morerestrictive agreements.

A Comprehensive Test Ban Treaty was thedeclared goal of the past six U.S. Administra-tions, but remained elusive. The Reagan Ad-ministration, however, has viewed limitationson nuclear testing as not in the national secu-rity interests of the United States, both atpresent and in the foreseeable future. Thestated policy of the Reagan Administration isas follows:

A Comprehensive Test Ban (CTB) remainsa long-term objective of the United States.As long as the United States and our friendsand allies must rely upon nuclear weapons todeter aggression, however, some nuclear test-ing will continue to be required. We believesuch a ban must be viewed in the context ofa time when we do not need to depend on nu-clear deterrence to ensure international secu-rity and stability and when we have achievedbroad, deep, and verifiable arms reductions,substantially improved verification capabil-

27

Figure 2-1.— Nuclear Testing, July 16, 1945-December 31, 1987

USA LTBT TTBT100

90

80

70

60

50

40

30

20

10

0

i1IIIIIII

1945

USSR50

40

30

20

10

0 L1945

1950 1955 1960 1965 1970 1975 1980 1985

II I

I: I

- i !I!II— n

1950 1955 1960 I 1965 1970I1975 1980 1985

I

1990

1990

FranceI

2 II

10 iI

o1945 1950 1955 1960 1970

I1975 1980 1985 1990

UK ! I10 I I

Io1945 1950 1955 1960 I 1965 1970

I1975 1980 1985 1990

II i

China II

10

o1945 1950 1955 1960 1965 1970 1975 1980 1985 1990

India i

II

10

1945 1950 1955 1960 1965 1970 1975 1980 1985 1990

Atmospheric Undergroundnuclear explosions nuclear explosions

SOURCE: Data from the Swedish Defense Research Institute,

28

ities, expanded confidence building measures, Recently, a series of events have heightenedand greater balance in conventional forces.2

worldwide interest in test ban treaties and aDespite this declared United States position, number of proposals have been brought before

Congress to further limit the testing of nuclearpublic interest in a test ban remains strong. weapons (see box 2-B). To evaluate these pro-2U.S. Department of State, Bureau of Public Affairs, “U.S.

posals requires an understanding of the desiredPolicy Regarding Limitations on Nuclear Testing, ” Special Re- and available levels of verification needed toport No. 150, Washington, DC, August 1986, monitor compliance.

Box 2-B.—Recent Interest in Nuclear Test LimitationsAugust 6, 1985 to February 26, 1987: The Soviet Union observes a 19-month unilateral test morato-rium. Upon ending its moratorium, the Soviet Union declares that testing would be stopped againas soon as a U.S. testing halt was announced.February 26, 1986: House Joint Resolution 3 (with 207 cosponsors) passes the House with a voteof 268 to 148, requesting President Reagan to resume negotiations with the Soviet Union towardsa comprehensive test ban treaty and to submit to the Senate for ratification the Threshold TestBan Treaty (TTBT) and the Peaceful Nuclear Explosions Treaty (PNE). The wording of the HouseResolution is nearly identical to a similar proposal which previously passed the Senate by a voteof 77 to 22 as an amendment to the 1985 Department of Defense (DoD) Authorization Bill.May 28, 1986: The Natural Resources Defense Council, a private environmental group, signs anagreement with the Soviet Academy of Sciences for the establishment of three independent seismicmonitoring stations near the principal nuclear test sites of each country. The agreement specifiesthat the stations are to be jointly manned by American and Soviet scientists and the data is tobe made openly available.Summer 1986: UN Conference on Disarmament agrees to a global exchange by satellite of sophisti-cated seismic data.August 7, 1986: The Five Continent Peace Initiative formed by the leaders of six nonaligned coun-tries (India, Sweden, Argentina, Greece, Mexico, and Tanzania) urges a fully verifiable suspensionof nuclear testing and offers assistance in monitoring the ban.August 8, 1986: House of Representatives votes 234 to 155 in favor of an amendment to the DefenseAuthorization Bill that would delete funding in calendar year 1987 for all nuclear tests with yieldslarger than 1 kt, provided that the Soviet Union does not test above 1 kt and that the Soviet Unionaccepts a U.S. monitoring program. The House Amendment is dropped prior to the Reykjavik sum-mit when the Administration agrees to submit the TTBT and PNET to the Senate for advice andconsent.May 19, 1987: House of Representatives votes 234 to 187 in favor of an amendment to the fiscalyear 1988 DoD Authorization Bill to delete funding for all nuclear tests with yields larger than1 kt during fiscal year 1988 provided that the Soviet Union does the same and, if reciprocal (in-country) monitoring programs are agreed on and implemented.July 1987: At the expert talks on nuclear testing, Soviets propose calibration of test sites to reducethe uncertainty in yield estimates. The proposal invites U.S. scientists to the Soviet nuclear testingsite to measure Soviet test yields using both the CORRTEX system and seismic methods. In re-turn, Soviet scientists would measure a U.S. test at the Nevada test site using both methods.November 9, 1987: Formal opening of negotiations in Geneva on nuclear test limitations.January 1988: Teams of U.S. and Soviet scientists visit each other’s test site to prepare for jointcalibration experiments to reduce uncertainty in yield estimation.

29

EVALUATING THE CAPABILITY OF AVERIFICATION NETWORK

Making a decision on whether verificationis adequate requires an understanding of thecapability of the verification system, the sig-nificance of the potential violation, and a de-cision as to what is an acceptable level of risk.Developing the basis for the decision is diffi-cult because it involves two different commu-nities: those who can assess the system’s ca-pabilities (a technical question) generally donot form the same community as those whoare officially responsible for assessing the over-all risks and benefits (a policy question).

For policymakers to weigh the benefits ofthe treaty against the risks posed by the pos-sibility of unilateral noncompliance, a clear un-derstanding of the capabilities of the monitor-ing system is necessary. In the frozen riverexample, this understanding would result frommeasurements of how thin the ice is and a tech-nical interpretation of how much weight theice can bear. The decision as to what consti-tutes an acceptable level of risk is a policy de-cision because it is based on an assessment ofthe overall benefits of the treaty weighedagainst the risk. In the frozen river example,this assessment would represent the decisionas to how thin the ice would need to be to de-ter crossing, how deep the river is, and howsignificant a crossing would be.

The burden on the policy-making commu-nity, therefore, is to understand technicaldescriptions of the verification system’s capa-bility and incorporate this knowledge into theirrisk-benefit decision. As we shall see, the dif-ficulty is that monitoring capabilities are notcertain, but rather they can only be describedin probabilistic terms. For example: What arethe chances that a clandestine nuclear testabove a certain yield could go undetected?What are the chances that a detected seismicevent of a certain magnitude could have beena nuclear explosion rather than an earthquake?If an underground nuclear explosion is re-corded, how certain can we be that the yieldof the explosion was below a specific thresh-

old? The answers to these questions can be ob-scured by the manner in which they are por-trayed. In particular:

• differences between verification systemscan be made to look superficially eitherlarge or small,

● opportunities for Soviet cheating can bemisrepresented, and

● the decision of what defines adequate ver-ification can be made through an arbitraryprocess.

The next three sections illustrate the issuesthat arise in assessing a verification capability—and the misrepresentations that are possi-ble–by considering a question that arousedmuch Congressional interest in early 1987:What is our ability to measure seismically theyields of Soviet explosions near the 150 kilotonlimit of the Threshold Test Ban Treaty? The firstsection presents the statistical representationsthat are used to describe yield estimation. Thisincludes the meaning of uncertainty and con-fidence levels, along with a comparative dis-cussion that enables the reader to understandwhat changes in the uncertainty represent. Thenext section examines how these uncertaintiestranslate into opportunities for Soviet cheat-ing. And finally, the third section illustrateshow the policy decision of what constitutes ade-quate yield estimation capability has changedin apparent response to variations in the at-tractiveness of particular monitoring systems.

Uncertainty and Confidence Levels

In determining the verifiability of the 1974Threshold Test Ban Treaty, policymakerswanted to know the capabilities of a seismicmonitoring system for estimating whether So-viet tests are within the treaty’s limits. Thedescription of such capabilities is accomplishedthrough the use of statistics. While the statis-tical calculations are relatively straightfor-ward, difficulties arise in correctly appreciat-ing what the numbers mean. To illustrate how

30

such presentations can be misleading, we willfirst use an example from a common and com-paratively well-understood event:3

At the end of the 1971 baseball season, theSan Francisco Giants were playing the LosAngeles Dodgers in a televised game. In thefirst inning, Willie Mays, approaching the endof his illustrious career, hit a home run. Now,one expects that hitting a home run in thefirst inning should be a rather unusual occur-rence because the pitcher is at his strongestand the batter has not had time to get usedto the pitcher. In any case, Willie Mays hita home run and it triggered what every base-ball fan would recognize as a typical baseballstatistician’s response. The calculations weremade and it was discovered that, of the 646home runs Mays had hit, 122 of them hadbeen hit in the first inning: 19 percent! In themost unlikely one-ninth of the innings, Wil-lie Mays had hit nearly one-fifth of his homeruns. This realization captured the interestof the reporting community and was dis-cussed extensively in the media. In responseto the publicity, the Giants’ publicity direc-tor explained it by saying that”. . . Willie wasalways surprising pitchers in the first inningby going for the long ball before they werereally warmed up. ” The power of statisticalanalysis was able to draw out the hiddentruths about Willie Mays’ performance.Although the data and calculations were cor-

rect, the interpretation could not have beenmore wrong. Throughout Mays’ career, he hadalmost always batted third in the Giants’lineup (occasionally, he batted fourth). Thatmeant he almost always batted in the first in-ning. Because he averaged about four at-batsper game, approximately one-quarter of his at-bats came in the first inning. Therefore, he onlymanaged to hit 19 percent of his home runsduring the first inning which comprised 25 per-cent of the time that he was at bat. Of the mil-lions of people who must have heard and readabout the item, not one pointed out the misin-terpretation of the statistic. This included notjust casual observers, but also experiencedprofessionals who spend their careers interpret-ing just that kind of information.

Whis example is paraphrased from David L. Goodstein, Statesof Matter (Englewood Cliffs, NJ: PrenticeHall) 1975.

The point here is that the interpretation ofnumbers is tricky and statistical presentationscan often be misleading. The real challenge isnot in calculating the numbers, but in correctlyinterpreting what different numbers mean. Inan area with as many technical considerationsand political influences as arms control verifi-cation, one has to be particularly careful thatdifferent numbers represent truly significantdifferences and not just arbitrary distinctions.

As with every real-world measurement, esti-mating the size of a nuclear explosion resultsin variation, or scatter, among the estimates.The use of different instruments at differentlocations, interpretations of the measurementsby different people in slightly different ways,and unknown variations in signals being ob-served result in slightly different estimates.Similarly, if one were to measure the daily tem-perature outside using a number of thermom-eters located in several areas, there would beslight differences in the temperature depend-ing on the particular thermometer, its location(surrounded by buildings and streets, or in apark), how each scale was read, etc.

In seismology, errors come from the instru-mentation, from the interpretation of the data,from our incomplete knowledge about how wellan explosion transmits its energy into seismicwaves (the coupling), and from our limited un-derstanding of how efficiently seismic wavestravel along specific paths (the path bias).Some of these errors are random-they varyunpredictably from one measurement to thenext. Other errors are not random, but are sys-tematic and are the same from one measure-ment to the next.

In our example of measuring temperature,a systematic error would be introduced if eachreading were made using an improper zero onthe thermometer. With such a systematic er-ror, the measurements would continue to bedistributed randomly, but the distributionwould be shifted by the difference between thetrue and incorrect zero. The distance from theincorrect value to the actual value would rep-resent the size of the systematic error. In seis-mology, an example of a systematic error re-

31

suits from the failure to allow correctly for thedifference in seismic transmission between theUnited States and Soviet test sites. The biasterm added to the calculation corrects for thiseffect, although there still remains an uncer-tainty associated with the bias.

The distinction between random and system-atic errors, however, is not a clear boundary.In many cases, random errors turn out to besystematic errors once the reason for the er-ror is understood. However, if the systematicerrors are not understood, or if there are lotsof systematic errors all operating in differentways, then the systematic errors are often ap-proximated as random error. In such cases, therandom uncertainties are inflated to encom-pass the uncertainties in estimating the sys-tematic effect. In monitoring the yields ofSoviet testing near 150 kt, most of the uncer-tainty is associated with estimation of system-atic error because the test site has not beencalibrated. In describing the capability tomeasure Soviet tests, the estimate of the ran-dom error has been inflated to account for theuncertainties in the systematic error.

We will see in the next section that while sys-tematic errors might be exploited if they hap-pen to be to one country’s advantage, randomerrors do not provide opportunities for cheat-ing. Furthermore, the uncertainty in the esti-mates of the systematic errors can often be sig-nificantly reduced by negotiating into thetreaty such provisions as the calibration ofeach test site with explosions of known yield.For this reason, calibration is important andshould certainly be part of any future agree-ment. But, before we discuss how random andsystematic errors affect monitoring, the methodof statistically describing the capabilities needsto be explained. For this, it is useful to returnto our temperature example.

While collecting measurements of the dailytemperature by using many thermometers inmany locations, we would find that some ofthe measurements were high and some werelow, but most of them were somewhere in themiddle. If all of the measurements were plot-ted, they would cluster around one number

with roughly equal scatter distributed to ei-ther side. It would be most likely that the bestactual value for the daily temperature wouldbe near the central number and it would be in-creasingly less likely that the actual valuewould be off towards either end of the scatterdistribution.

In seismology, that central number, meas-ured using several techniques from many seis-mometers located in different locations, is re-ferred to as the “central value yield. ” Nomatter how accurately we could measure theyields of explosions known to be 150 kt, wewould still expect—due to the normal randomscatter of measurements and presuming therewere no systematic error—that roughly halfthe explosions would be recorded as being be-low 150 kt and roughly half would be recordedas being above 150 kt. The width of the distri-bution above and below depends on the capa-bility of the measuring system and can be de-scribed using a “factor of uncertainty. ”

Unless otherwise stated, the factor of uncer-tainty for a given measurement is defined asthat number which, when multiplied by ordivided into an observed yield, bounds therange which has a 95 percent chance of includ-ing the actual (but unknown) value of the yield.There is only a 5 percent chance that the meas-urement would be off by more than this fac-tor. For example, a “factor of 2“ uncertaintywould mean that the measured central valueyield, when multiplied and divided by two,would define a range within which the trueyield exists 95 percent of the time.

Naturally, the more confident one wants tobe that the true value lies within a given range,the larger that range will have to be. The 95percent range is used by convention, but thereis no real reason why this should be the confi-dence level of choice for comparing monitor-ing systems. Using a different confidence levelthan 95 percent to define the factor of uncer-tainty would cause the factor to have a differ-ent value.

As an example, imagine that all Soviet ex-plosions are detonated with an actual yield of150 kt and that these explosions are measured

32

Figure 2-2.–Measurements of 150 kt With Factor-of-2Uncertainty

Measurements of 150 ktw/ factor-of-2 uncertainty

True yield = 150 kt0.010 . /

0.005 “

0 75 150 225 300 375

Yield estimate, in kilotons

Measurements of a 150 kt explosion using a factor-of-2 un-certainty would be expected to have this distribution. Theprobability that the actual yield lies in a particular range isgiven by the area under the curve. Ninety-five percent of thearea lies between 75 and 300 kt.

by a monitoring system described as havinga factor of 2 uncertainty. It would then followthat 95 percent of the yield estimates will bebetween 75 and 300 kt. Therefore, if the Sovietsconduct 100 tests all at 150 kt, we would ex-pect that about 5 of them would be measuredwith central yields either above 300 kt or be-low 75 kt. This is graphically illustrated in fig-ure 2-2, where the probability that the actualyield lies in a particular range is given by thearea under the curve over that range.

Although the range from 75 to 300 contains95 percent of the measurements, we can seethat for the distribution assumed (the normaldistribution) most of the measurements are,in fact, much closer to the actual yield. For ex-ample, figure 2-3 showing the 50 percent con-fidence level for the same distribution illus-trates that over half of the measurements willfall between 118 and 190 kt.

On the other hand, if one wanted to specifya range in which 99 percent of the measure-

Figure 2-3.—Fifty-Percent Confidence Level forMeasurements of 150 kt With Factor-of-2 Uncertainty

Measurements of 150 kt

-W/ factor-of-2 uncertainty

0 75 150 225 300 375


Same distribution as figure 2-2 with 50 percent of the areamarked. Over 50 percent of the measurements would be ex-pected to fail within 118 and 190 kt.

ments fall, the range would have to be extendedto 60 to 372 kt. In real yield estimation, how-ever, there is no meaningful distinction be-tween 95 percent confidence and 99 percentconfidence. The normal distribution is used asa convenience that roughly represents realitynear the center of the distribution. The tailsof the distribution are almost certainly notclose approximations of reality. The generalpoint can still be made: namely that, it takesever greater increases in range for slight im-provements in the confidence level. Table 2-1illustrates this for the case of a normal distri-bution by showing how the yield range variesfor given factors of uncertainty at various con-fidence levels.4

From the table, it is clear that a quoted rangeof values is highly dependent not only on thefactor of uncertainty of the monitoring system,but also on the chosen confidence level. Forexample, the same quoted uncertainty range

4Table 2-1 can be read as follows: “If an explosion is meas-ured at 150 kt using a system with a factor of [1.5] uncertainty,the [70 percent] percent confidence ranges from [121 to 186] kt. ”

Table 2.1.–Confidence Internals for an Explosion Measured at 150 kt

CONFIDENCELEVEL99%95%90%80%70%50%

– – – – – – – – U N C E R T A I N T Y – – – – – – – – – 1Factor of 2 Factor of 1.5 Factor of 1.3

61-372 88-255 106-21175-300 100-225 115-19584-269 106-212 120-18796-236 116-195 126-179104-216 121-186 130-173118-191 130-173 138-164

SOURCE: Office of Technology Assessment, 1988.

33

that can be achieved using a factor of 1.3 mon-itoring system at the 95 percent confidencelevel can be achieved using a factor of 1.5 mon-itoring system at the 80 percent confidencelevel, or even using a factor of 2 monitoringsystem at the 50 percent confidence level.

In selecting an appropriate confidence levelto use for quoting uncertainty, the purpose ofthe comparison must also be recognized. In thecase of monitoring compliance with a thresh-old test ban treaty, one is concerned that in-tentional cheating could be missed or that un-acceptable false alarms could occur. It mustbe remembered, therefore, that the uncertaintydescribes the likelihood that the actual valuewill not fall either above or below the range.For example, the 95 percent confidence levelmeans that given repeated trials there is onlya 5 percent chance the true yield was eitherabove or below the range of the measured yield.In other words, there is a 2.5 percent chancethat it could have been above the range anda 2.5 percent chance that it could have beenbelow the range. The 2.5 percent below therange, however, is not a concern if it is belowthe threshold. The concern is only the 2.5 per-cent chance that it had a yield above the thresh-old. For the purposes of monitoring a thresh-old, this would only be the half of the 5 percentabove the threshold, or in other words 2.5 per-cent. Consequently, the 95 percent confidencelevel really corresponds to a 97.5 percent con-fidence level for monitoring violations of athreshold. Following the same argument, the50 percent confidence level really correspondsto a threshold monitoring at the 75 percent con-fidence level, and so on.

From the previous discussions, it is obviousthat while it maybe convenient to chose a par-ticular confidence level to compare monitor-ing systems (such as the 95 percent confidencelevel), it should only be done with great cau-tion. In particular, it should be kept in mindthat:

● the choice of any particular confidencelevel is arbitrary, in the sense of being onlya convenience to allow comparison of theaccuracies of different yield estimationmethods;

●

●

as seen from table 1-1, differences can bemade to look small or large depending onwhich particular confidence level is cho-sen; andalthough very high confidence levels havelarge ranges of uncertainty, it is with de-creasing likelihood that the actual valuewill be at the extremes of those ranges ofuncertainty.

These considerations are important to ensurethat common mistakes are not made, namelythat:

●

●

the range of uncertainty is not equatedwith a range in which cheating can occur,andthe range of uncertainty chosen for com-parative reasons does not evolve into therange of uncertainty that is used to de-termine what constitutes adequate veri-fication.

The Relationship Between Uncertaintyand Cheating Opportunities

The main reason for designing monitoringsystems with low uncertainty is to reduce theopportunities for cheating. The relationship be-tween uncertainty and opportunities for cheat-ing, however, is not always straightforward.Even if the uncertainty of a particular moni-toring system is large, this does not mean thatthe opportunities for cheating are correspond-ingly large.

As mentioned before, the uncertainty is cre-ated by two types of error: random and sys-tematic. Although we try to estimate the sys-tematic error (such as path bias) as accuratelyas possible, there is a chance that our estimatescould be slightly too high or too low. For ex-ample, if our estimate of the bias5 is too low,we would overestimate the yields of Soviet ex-plosions, and Soviet testing near 150 kt wouldappear as a series of tests distributed arounda yield value that was above 150 kt. If our esti-mate of the bias were too high, Soviet testing

5% chapter 7, “Estirnating the Yields of Nuclear Explosions, ”for an explanation of bias.

34

near 150 kt would appear as a series of testsdistributed around a yield below 150 kt. Thecase where we systematically underestimateSoviet yields and they presume this underes-timation is occurring is the only case that pro-vides opportunity for unrecognized cheating.If this were happening, it could happen onlyto the extent that the systematic effect hasbeen underestimated.

As chapter 7 discusses, the systematic partof the uncertainty can be significantly reducedby restricting testing to specific calibrated testsites. If such calibration were an integral partof any future treaty, the concern over system-atic errors of this kind should be minimal. Themajority of the error that would remain is ran-dom. A country considering violating a treatycould not take advantage of the random errorbecause it would be unable to predict how therandom error would act.

In early 1987, both the Senate Foreign Re-lations Committee and the Senate Armed Serv-ices Committee held hearings on verificationcapabilities in their consideration of advice andconsent to ratification of the 1974 ThresholdTest Ban Treaty and the 1976 Peaceful Nu-clear Explosions Treaty. Members of theseCommittees wanted to know whether seismicmethods could adequately measure the size ofSoviet underground nuclear tests or whethermore intrusive methods were required. The tes-timony was often confusing due to the vari-ous means of representing statistical uncer-tainties. For the time being, we will analyzeonly the use of statistics and take as given theunderlying information. However, that accept-ance is also controversial and is discussedseparately in the chapter on yield estimation(chapter 7).

The Department of Defense presented thecapabilities of seismic monitoring to the Sen-ate Committee on Foreign Relations on Janu-ary 13, 1987 and the Senate Committee onArmed Services on February 26, 1987 in thefollowing manner:6

‘Testimony of Hon. Robert B. Barker, Assistant to the Sec-retary of Defense (Atomic Energy) and leader of formal nego-tiations on Nuclear Test Limitations.

The seismic methods that we currentlymust rely onto estimate yields of Soviet nu-clear detonations are assessed to have abouta factor-of-two uncertainty for nuclear tests,and an even greater uncertainty level for So-viet peaceful nuclear explosions.

This uncertainty was then explained asfollows:

This uncertainty factor means, for exam-ple, that a Soviet test for which we estimatea yield of 150 kilotons may have, with 95 per-cent probability, an actual yield as high as300 kilotons-twice the legal limit-or as lowas 75 kilotons. 7

These statements are misleading in that theycreate the impression that there is a high prob-ability, in fact, almost a certainty, that theSoviets could test at twice the treaty’s limitbut we would measure the explosions as be-ing within the 150 kt limit. They imply thata factor of 2 uncertainty means that there isa high probability that an explosion measuredat 150 kt could, in actuality, have been 300 kt.Yet as we have seen in the discussion of un-certainty, given a factor of two uncertainty,the likelihood of an explosion with a yield of300 kt actually being measured (with 95 per-cent probability) as 150 kt or below is less than1 chance in 40.

The chances decrease even further if morethan a single explosion is attempted. For ex-ample, the chance of two explosions at 300 ktboth being recorded as 150 kt or less is about1 in 1,600; and the chance of three explosionsat 300 kt or greater being recorded as 150 ktor less would be roughly 1 in 64,000. Thus, itis highly unlikely that explosions could berepeatedly conducted at 300 kt and systemat-ically recorded as being 150 kt or less.

So far we have been looking only at the likeli-hood that a test will appear as 150 kt or less.

‘This statement is nearly identical to the wording in the U.S.Department of State, “Veri~g Nuclear Test Limitations: Pos-sible U.S.-Soviet Cooperation, ” Special Report No. 152, Aug.14, 1986, which states “A factor of two uncertainty means, forexample, that a Soviet test for which we derive a ‘central yield’value of 150 kt may have, with a 95 percent probability, a yieldas high as 300 kt or as low as 75 kt. ”

35

From a practical point of view, it must be rec-ognized that the test would not need to looklike 150 kt or less; it would only have to ap-pear as though it were within the error of a150 kt measurement in order to avoid credibleassertions of non-compliance. Some could mis-interpret this as meaning that a test well abovethe threshold might have enough uncertaintyassociated with it so that its estimate mightappear to be within the expected uncertaintyof a test at 150 kt. They might then concludethat the opportunity to test well above thethreshold cannot be denied to the Soviets.

Such a one-sided assessment of the uncer-tainty is extremely misleading because it as-sumes all of the errors are systematic and canbe manipulated to the evader’s advantage. Acountry considering violating the thresholdwould also have to consider that even if partof the systematic uncertainty could be con-trolled by the evader, the random part of theuncertainty could just as likely work to its dis-advantage. This point was briefly recognizedin the following exchange during a Senate Com-mittee on Foreign Relations hearing:

“ . . . knowing these probabilities, if youreally started to cheat, as a matter of fact youwould take the risk of being out at the far tail.That would really show up fast. If you set outto do a 300 kt, you could show up on our seis-mographs as 450, right?”

Senator Daniel P. Moynihan

“The problem is, if you fired an explosionat 300 or 350 . . . it could very well look 450or 500 and the evader has to take that intoconsideration in his judgment. ”

Dr. Milo Nordyke,Director of Verification

Lawrence LivermoreNational Laboratory

Also, this analysis assumes that only onemethod of yield estimation will be used. Othermethods of yield estimation are also availableand their errors have been shown to be onlypartially correlated. The evader would have totake into account that even if the uncertaintyis known and can be manipulated for onemethod of yield estimation, other methodsmight not behave in the same manner. Such

considerations would severely diminish the ap-peal of any such opportunity.

In conclusion, it can be seen that althoughthe statistical descriptions of the capabilitiesof various methods of yield estimation havebeen debated extensively, the differences theyrepresent are often insignificant. There is bothsystematic and random uncertainty in themeasurements of Soviet yields. The system-atic error would provide only a limited oppor-tunity for cheating, and then only if it was inthe advantage of the cheater. Even in such acase, only the portion of the error that is sys-tematic can be exploited for cheating. Further-more, much of the systematic error would beremoved through such treaty provisions ascalibrating the test site. Once the systematicerror had been nearly eliminated, the remain-ing uncertainty would be random. The randomuncertainty does not provide opportunity forcheating. In fact, if a country were consider-ing undertaking a testing series above thethreshold, it would have to realize that the ran-dom uncertainty would work against it. Witheach additional test, there would be a lesserchance that it would be recorded within thelimit and a greater chance that at least one ofthe tests would appear to be unambiguouslyoutside the limit.

What Constitutes AdequateVerification?

After the accuracies and uncertainties ofvarious verification systems have been under-stood, a decision must be made as to what con-stitutes an acceptable level of uncertainty. In1974, when the Threshold Test Ban Treaty wasfirst negotiated, a factor of 2 uncertainty wasconsidered to be the capability of seismic meth-ods. At that time, a factor of 2 uncertainty wasalso determined to constitute adequate verifi-cation.8 Presently, the level of accuracy claimed

80riginally, the factor of 2 uncertainty was established forthe 90 percent confidence level, whereas today it refers to the95 percent confidence level. It should be noted that a factorof 2 at the 90 percent confidence level corresponds to about afactor of 2.5 at the 95 percent confidence level. Thus the ac-cepted level of uncertainty in 1974 was really about a factorof 2.5 using the present confidence level. The insistence on a

(continued on next page)

36

for the on-site CORRTEX method is a factorof 1.3.9 This level of 1.3 has subsequently beendefined as the new acceptable level of uncer-tainty, although many believe it was definedas such only because it corresponds to the ca-pabilities of this newly proposed system.

It appears that the determination of ade-quate compliance is a subjective process thathas been influenced by the capabilities of spe-cific monitoring systems. A decision as to whatconstitutes adequate verification should notbe determined by the political attractiveness

● of any particular monitoring system, but ratherit should represent a fair assessment of the pro-tection required against non-compliance. In thefrozen river analogy, this would be a fair assess-ment of how thin the ice must be to deter some-one from crossing. Monitoring capability cer-

{continued from previous page)higher confidence level occurred simply because it was moreconvenient to use the 95 percent confidence level which cor-responds to 2 standard deviations.

‘See appendix, Hydrodynamic Methods of Yield Estimation.

tainly influences our decision as to whether atreaty is worthwhile, but it should not influ-ence the standards we set to make that deci-sion. Also, the capability of a monitoring sys-tem is just one aspect to be considered, alongwith other important issues such as negotia-bility and intrusiveness.

What constitutes adequate verification mayalso vary for different treaty threshold levels.For example, a factor of 2 uncertainty for mon-itoring a 100 kt threshold would mean that 95percent of the measurements at the thresholdlimit would be expected to fall within 50 and200 kt (a total range of 150 kt), while a factorof 2 uncertainty for monitoring a 1 kt thresh-old would mean that 95 percent of the meas-urements at the threshold limit would be ex-pected to fall within% and 2 kt (a total rangeof 1.5 kt). A range of uncertainty of 1.5 kt maynot provide the same opportunities or incen-tives for cheating as a range of uncertainty of150 kt. Consequently, at lower treaty thresholds,the significance of a given yield uncertaintywill almost certainly diminish.

THE QUESTION OF DETERMINING COMPLIANCEIn addition to understanding the accuracy

and uncertainty of the verification system, anddeciding on an acceptable level of uncertainty,a decision will also have to be made as to whatwould constitute compliance and non-compli-ance. Violations of the treaty must be distin-

. guished from errors in the measurements (bothsystematic and random) and errors in the test.This is of particular concern in light of find-ings by the administration that:

Soviet nuclear testing activities for a num-ber of tests constitute a likely violation of le-gal obligations under the Threshold Test BanTreaty .’”

To examine the context in which this mustbe viewed, we can once again return to the fro-zen river analogy and imagine a situation

‘“’’ The President’s Unclassified Report on Soviet Noncom-pliance with Arms Control Agreements,” transmitted to theCongress Mar. 10, 1987.

where we come by and see marks on the ice.We must then determine whether the marksindicate that someone successfully crossed theice. This could be misleading because all thatwe are doing is looking in isolation at the prob-ability that a certain mark could have beenmade by someone crossing the ice. Thus thelikelihood that a mark was made by a personbecomes the likelihood that someone crossedthe ice. This, however, is only part of the is-sue. If we knew for example that the ice wereso thin that there was only a 1 in 10 chanceit could have been successfully crossed, thatthe water was deep, and that there was no rea-son to get to the other side, these factors mightweigh in our determination of whether a markwas mane-made or not. (Why would a personhave taken such risks to gain no value?) Onthe other hand, if we knew the ice were thickand could be crossed with high confidence, thatthe water was shallow, and that real value wasto be obtained by crossing, then we might

37

make a different judgment as to whether themarks were man-made because there wouldreally be understandable motivation. Thus thequestion of compliance is also dependent ona judgment reflecting one’s perception of theadvantages that could be obtained through aviolation.

In the case of test ban treaties, there are also“gray areas” due to the associated error of themeasurements. For example, it must be as-sumed that a country will test up to the limitof the treaty, and therefore, some of the esti-mates would be expected to fall above 150 ktsimply due to random error.11

Assuming that the errors are known and thatapparent violations of the treaty due to sucherrors are recognized, there may also be otherviolations that cause concern but do not ne-gate the benefits of the treaty. These includeaccidental violations, technical violations, andviolations of the “spirit” of the treaty.

Accidental violations are violations of thetreaty that may occur unintentionally due tothe inexact nature of a nuclear explosion. Itis possible that the explosion of a device witha yield that was intended to be within the limitof the treaty would produce an unexpectedlyhigher yield instead. This possibility was rec-ognized during negotiations of the TTBT. Thetransmittal documents which accompanied theTTBT and the PNE Treaty when they weresubmitted to the Senate for advice and con-sent to ratification on July 29, 1976 includedthe following understanding recognized byboth the United States and Soviet Union:

Both Parties will make every effort to com-ply fully with all the provisions of the TTBTreaty. However, there are technical uncer-tainties associated with predicting the pre-cise yields of nuclear weapon tests. These un-certainties may result in slight, unintendedbreaches of the 150 kt threshold. Therefore,the two sides have discussed the problem andagreed that: (1) One or two slight, unintended

“For example, if the Soviet Union tested 20 devices at 150kt and we estimated the yields using a system that was describedas having a factor of 2 uncertainty, the probability of measur-ing at least one of them as being 225 kt or greater is 92 percent.

breaches per year would not be considered aviolation of the Treaty; (2) such breacheswould be a cause for concern, however, and,at the request of either Party, would be sub-ject for consultations.Technical violations are violations of the

treaty that do not result in any sort of strate-gic advantage. An example would be a techni-cal violation of the 1963 Limited Test BanTreaty (LTBT) which prohibits any explosionthat:

. . . causes radioactive debris to be presentoutside the territorial limits of the State un-der whose jurisdiction or control such explo-sion is conducted.12

This prohibition includes the venting ofradioactive debris from underground explo-sions. Both the United States and the SovietUnion have accused each other of releasingradioactive material across borders and of vio-lations of the 1963 LTBT. These violations are“technical” if the treaty is viewed as an armscontrol measure. However, they are materialviolations if the treaty is viewed as an envi-ronmental protection measure.

Violations of the “spirit” of the treaty arealso of concern. These include, for example, ac-tions which are contrary to the treaty’s pream-ble. The treaty’s preamble declares the inten-tions and provides a context for the treaty.Such declarations, however, are nonbinding.

Another area concerns treaties that havebeen signed but never ratified. Both the 1974TTBT and the 1976 PNE Treaty remain un-ratified, although they were signed over 10years ago. Because neither the United Statesnor the Soviet Union have indicated an inten-tion not to ratify the treaties, both parties areobligated under international law (Article 18,the 1969 Vienna Convention on the Law ofTreaties) to refrain from acts which would de-feat their objectives and purposes.

All of these types of violations contributeto the gray area of compliance versus noncom-pliance and illustrate why determining com-

‘2 Article I,b.

38

pliance is a political as well as a technical deci-sion. In the case of monitoring undergroundnuclear tests, the actual measured yield thatwould constitute clear evidence of a violationwould always be higher than the yield limitof the treaty. Perhaps an analogy for uncer-tainties in yield estimates and Soviet compli-ance under the TTBT is in monitoring a speedlimit of 55 mph. Under the present 150 kt limit,an observed yield of 160 kt is like comparing58.7 mph to 55 mph. The police do not givetickets when their radar shows a speed of 58.7mph because most speedometers are not thataccurate or well calibrated, and because curvesand other factors can lead to small uncertain-ties in radar estimates of speed. Similarly, al-though a 160 kt measurement maybe regardedby some as a legal lack of compliance, such anumber can well arise from uncertainties inseismic estimates. At radar measurementsover 65 mph the police do not question thatthe 55 mph limit has been exceeded, and thespeeder gets a ticket. With this standard, itwould take a calculated yield of about 180 to190 kt to conclude that a violation had likelytaken place. As mentioned before, however,this argument does not mean that the Sovietscould test up to 180 to 190 with confidence,because the uncertainty could just as likelywork against them. A 180 to 190 kt test mightproduce an observed yield well over 200 kt just

as likely as it might produce a yield within theexpected error range of a treaty compliant test.

It must be recognized, however, that the cal-culated yield for declaring a treaty violationwill always be higher than the limit of thetreaty. Consequently, one or two small breachesof the treaty could occur within the expecteduncertainty of the measurements. A countryintent on cheating might try to take advan-tage of this by risking one or two tests withinthe limits of the uncertainty range. Even if de-tected, a rare violation slightly above the per-mitted threshold could be explained away asan accidental violation due to an incorrectprediction of the precise yield of the nucleartest. This should be kept in mind when choos-ing a threshold so that small violations of thelimit (whether apparent or real) do not fall ina range that is perceived to be particularly sen-sitive.

What is done about violations is an addi-tional problem. In domestic law there are vari-ous kinds of violations. Traffic tickets, mis-demeanors, felonies, capital crime–all aredifferent levels. Similarly, in monitoring com-pliance, there are some things that amount totraffic tickets and some that amount to felo-nies. We must decide in which cases violationsor noncompliance are at the heart of a treatyand in which cases they area marginal problem.

Chapter 3

The Role of Seismology

CONTENTS

PageIntroduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41The Creation of Seismic Waves, . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41Types of Seismic Waves . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41Teleseismic Waves . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42

Body Waves... . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42Surface Waves . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44

Regional Waves. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ... 45Recording Seismic Waves . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48Seismic Arrays . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50

FiguresFigure No.

3-1.3-2.3-3.3-4.3-5.3-6.

3-7.3-8.3-9.

3-1o.3-11.3-12.3-13.

PageSeismic Waves Propagating Through the Earth . . . . . . . . . . . . . . . . . . 42Body Waves . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43Seismic Radiation Patterns . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44Reflected and Refracted Waves . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44Surface Waves . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45Seismograph Recording of P, S, and Surface Waves From a DistantEarthquake . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46Regional and Teleseismic Signals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47Pn Wave . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47Pg Wave . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47Lg Wave . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47Seismic Instrumentation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49Distribution of RSTN Stations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50Beamforming . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50

Chapter 3

The Role of Seismology

Seismology provides a technical means for monitoringunderground nuclear testing

INTRODUCTION

Verifying a ban on nuclear testing requiresglobal monitoring systems capable of detect-ing explosions in the atmosphere, underwater,and below ground. Tests in the atmosphere andunder water can be readily detected with highdegrees of confidence. The atmosphere is mon-itored effectively with satellites containing sen-sors that can detect the visible and near-infrared light emitted by a nuclear explosion.The oceans can also be monitored very effec-tively because water transmits acoustic waves

efficiently. Underwater explosions would bedetected by the acoustic sensors already inplace as part of anti-submarine warfare sys-tems. The most uncertain part of the globalverification system is the monitoring of under-ground nuclear explosions that might be con-ducted within the Soviet Union. The main tech-nical tools for monitoring underground nuclearexplosions come from the field of seismology,which is the study of earthquakes and relatedphenomena.

THE CREATION OF SEISMIC WAVES

A nuclear explosion releases its energy in lessthan 1/1,000,000 of a second (a microsecond).The explosion initially produces a gas bubblecontaining vaporized rock and explosive ma-terial. Within a microsecond, the temperaturewithin the bubble reaches about 1 milliondegrees and the pressure increases to severalmillion atmospheres. A spherical shock waveexpands outward from the bubble into the sur-rounding rock, crushing the rock as it travels.As the hot gases expand, rock is vaporized nearthe point of the explosion and a cavity iscreated.

ing medium. Eventually, the shock waveweakens to the point where the rock is nolonger crushed, but is merely compressed andthen returns (after a few oscillations) to itsoriginal state. These cycles of compression andrelaxation become seismic waves that travelthrough the Earth and are recorded by seis-mometers. By looking at seismic records (seis-mograms) and knowing the general propertiesof the travel paths of various waves, seismol-ogists are able to calculate the distance to theseismic event and what type of motion causedthe wave.

While the cavity is forming, the shockwavecontinues to travel outward into the surround-

TYPES OF SEISMIC WAVESAs with earthquakes, seismic waves result- eling along the Earth’s surface (surface waves).

ing from an explosion radiate outward in all The body and surface waves that can bedirections (figure 3-l). These waves travel long recorded at a considerable distance (over 2,000distances either by passing deep through the km) from the earthquake or explosion that crebody of the Earth (body waves), or else by trav- ated them are referred to by seismologists as

41

42

Figure 3-1 .—Seismic Waves Propagating Through

S o u r c e= .

An earthquake or underground explosion generates seismicwaves that propagate through the Earth.

SOURCE: Office of Technology Assessment, 1988.

teleseismic waves. Teleseismic body and sur-face waves are important to monitoring be-cause to be recorded they do not usually re-quire a network of seismometers near thesource. This allows for seismometers to be lo-cated outside the region that is being mon-itored.

Teleseismic waves contrast with what seis-mologists call regional waves. Regional wavesare seismic waves that travel at relatively highfrequencies within the Earth’s crust or outerlayers and typically are observed only at dis-tances of less than 2,000 km. Regional wavesare therefore recorded at seismic stations that,for application to monitoring Soviet nuclearexplosions, would have to be located within theterritory of the Soviet Union. Such stationsare called in-country or internal stations.

When seismic waves reach a seismic station,the motion of the ground that they cause isrecorded by seismometers. Plots of the waveforms are called seismograms. Because thedifferent waves travel at different speeds andalong different routes, they arrive at seismicstations at different times. The farther awaythe seismic station is from the source of thewaves, in general, the more dispersed in timethe different arrivals will be. By studying seis-mograms, seismologists are able to recognizethe various types of waves produced by eventssuch as earthquakes and nuclear explosions.

TELESEISMIC WAVES

The various types of teleseismic waves aredifferentiated by both the paths along whichthey travel and the type of motion that ena-bles them to propagate. The two main subdi-visions (body waves and surface waves) are dis-tinguished by the areas through which theytravel.

Body Waves

The waves traveling through the body of theEarth are of two main types: compressionalwaves (also called P waves) and shear waves(also called S waves). The designations P andS are abbreviations originally referring toprimary and secondary arrivals; compressionalwaves travel faster than shear waves and, con-sequently, arrive first at a recording station.Apart from their different speeds, P waves andS waves also have different characteristics.

P waves travel in a manner similar to soundwaves, that is, by molecules “bumping” intoeach other resulting in compression and dila-tion of the material in which they propagate.A cycle of compression moves through theEarth followed by expansion. If one imagineda particle within the wave, its motion wouldbe one of shaking back and forth in the direc-tion of propagation in response to the alter-nating compressions and expansions as thewave trains move through. The particle mo-tion is in the direction of travel and the wavecan propagate through both solids and liquids.A sudden push or pull in the direction of wavepropagation will create P waves (figure 3-2a).

In contrast, S waves propagate by moleculestrying to “slide” past each other much as hap-pens when one shakes a rope from side to sideand the disturbance passes down the rope. The

43

Figure 3-2.—Body Waves

P wave I C o m p r e s s i o n s -1 Undisturbed medium

– D i l a t a t i o n s

S wave

IA m p l i t u d e I

1 — - - - - - - - - Wavelength

Waves traveling through the body of the Earth are of two main types: compressional waves (P waves) and shear waves (S waves).

SOURCE B A Bolt, Nuclear Explosions and Earthquakes, W H Freeman & Co , 1976

wave motion is at right angles to the directionof travel. Any particle affected by the wavewould experience a shearing motion as thewave passed through. Because liquids have lit-tle resistance to shearing motion, S waves canonly pass through solid materials. (Liquids canbe compressed, which is why P waves cantravel through both solids and liquids.) A sud-den shearing motion of a solid will result inS waves that will travel at right angles to thedirection of propagation (figure 3-2 b).

In a P wave, particle motion is longitudinalin the direction of propagation, thus there canbe no polarization of a P wave. S waves,however, being transverse, are polarized andcan be distinguished as either horizontal or ver-tical. Particle motion in an SH wave is horizon-tal. In an SV wave, particles oscillate in a ver-tical plane perpendicular to the path.

An underground explosion creates a uniformpressure outward in all directions. Therefore,

explosions should be a source of nearly pureP waves. In practice, however, some S wavesare also observed. These waves are usually dueto asymmetries of the cavity or the pressurewithin the cavity created by the explosion,structural inconsistencies in the surroundingrock, or the presence of pre-existing stressesin the host rock that are released by the explo-sion. An earthquake, on the other hand, isgenerally thought of in terms of blocks of theEarth’s crust breaking or slipping past oneanother along a fault region. Because of theshearing motion, an earthquake creates mostlyS waves. P waves are also generated from anearthquake, but only in a four-lobed patternthat reflects the opposite areas of compressionand expansion caused by the shearing motion(figure 3-3). As discussed in chapter 5, thisdifference in source geometry can be used asa means of distinguishing the seismic signalscaused by explosions from the seismic signalsgenerated by earthquakes.

44

Figure 3.3. —Seismic Radiation Patterns

Explosion Earthquake

The pattern of seismic body waves generatedby a given source becomes complicated whenthe waves interact with the structures of theEarth. In general, when a wave hits a bound-ary within the Earth, such as where two differ-ent rock types meet, both reflected and trans-mitted P and S waves are generated at severaldifferent angles. (figure 3-4)

Because of all these possibilities, P wavesand S waves break into many types as theytravel through the Earth. At a depth of 30 to60 kilometers in the Earth, the velocity withwhich sound passes through the Earth in-creases markedly. This discontinuity, calledthe Mohorovicic Discontinuity (Moho), servesas a wave guide to trap seismic energy in theupper crust of the Earth.

Surface Waves

At the Earth’s surface, two additional seis-mic wave types are found. These surface waves,called Rayleigh waves and Love waves,1 areproduced by constructive interference of bodywave energy in the upper layers of the crust.

The particle motion of Rayleigh waves issomewhat analogous to that of ripples spread-ing over the surface of a lake. The analogy isnot exact, however, because unlike a waterwave, the orbital motion of a particle in a Ray-

‘These waves are named after the mathematicians who firstdeveloped the theory to describe their motion (Lord Rayleighand A.E.H. Love).

Figure 3-4.– Reflected and Refracted Waves

ReflectedI n c i d e n t ss

Reflected

Medium 1

M e d i u m 2

Refracted S

When a wave hits a boundary within the Earth both reflectedand refracted waves can be generated.

leigh wave is retrograde to the path of wavepropagation (figure 3-5a). The energy of theRayleigh wave is trapped near the ground sur-face, so as depth increases, the rock particledisplacements decrease until, ultimately, nomotion occurs.

The second type of surface wave whichtravels around the Earth is called a Love wave.Love waves have a horizontal motion that isshear, or transverse, to the direction of propa-gation (figure 3-5b). It is built up of trappedSH waves and has no vertical motion.

While Rayleigh waves can be detected byseismometers sensitive either to vertical orhorizontal motion of the Earth’s surface, Lovewaves are only detected by seismometers sens-ing horizontal motions.

Surface waves of either type, with a periodof 10 seconds, travel with a velocity of about3 kilometers per second. This corresponds toa wavelength of about 60 kilometers. Becauseof this long wavelength, surface waves aremuch less likely to be affected by small-scalevariations in Earth structure than short periodbody waves. This makes the use of surfacewaves attractive for measuring yields of So-viet tests. Body waves (P waves and S waves)are faster than surface waves and therefore ar-rive first at a recording station. This is fortu-nate because the surface waves have larger am-

45

Figure 3-5. -Surface Waves

Rayleigh wave

Love wave

Waves traveling along the surface of the Earth are of two main types: Rayleigh waves and Love waves.SOURCE: B.A Bolt, Nuclear Explosions and Earthquakes, W.H. Freeman & Co., 1976.

plitudes than the body waves. If they all unless the records were filtered to take advan-arrived simultaneously, the smaller amplitude tage of the different frequency content of theP waves would be hidden by the surface waves two types of waves (figure 3-6).

REGIONAL WAVES

In contrast with the teleseismic waves de-scribed above, regional waves are usually ob-served only at distances of less than 2,000 km.In general they have larger amplitudes andhigher frequency content then waves from thesame source recorded at teleseismic distances.Depending on their propagation characteris-tics, such regional waves are denoted by Pn,Pg, Sn, and Lg. A comparison of teleseismicand regional seismic waves is illustrated in fig-ure 3-7. Note the different time scales. The am-plitude scales are also different.

At regional distances, Pn is usually the firstwave to arrive at any given station. It is a wave

that goes down through the crust of the Earth,then travels mostly horizontally near the topof the upper mantle (in contrast to the usualbody, which goes deeply within the mantle),and finally travels upward through the crust,where the receiver is located. This path isshown in the upper part of figure 3-8. In thelower part are shown some of the multiplebounces that also contribute to the Pn wave.The Moho marks the boundary between thecrust and the upper mantle.

Pg is a wave that comes in later than Pn, andtravels the path between source and receiverwholly within the Earth’s crust. As shown in

46

Figure 3-6.—Seismograph Recording of P, S, and Surface Waves From a Distant Earthquake

P S u r f a c es w a v e s

A

Horizontal ground motion

SOURCE: Modified from F. Press and R. Seiver, Earth, W.H. Freeman & Co., San Francisco, 1974.

figure 3-9, Pg is thought to be guided by aboundary layer within the crust. It does notpropagate to as great a distance as Pn.

Sn is a shear wave that travels a path verysimilar to that of Pn, but arrives later. Sn ar-rives later because it is composed of shearwaves, which propagate slower than the Pwaves that make up Pn.

For purposes of explosion monitoring, Lcan be the most important of the regionalwaves because it is typically the largest waveobserved on a seismogram at regional dis-tances. Lg is a type of guided shear wave thathas most of its energy trapped in the Earth’scrust. In fact, this wave can be so strong that(contrary to what is implied by its inclusionin the class of “regional waves”) L for a large

explosion can be observed even at teleseismicdistances, i.e. well in excess of 2,000 km. The

observation of Lg out to teleseismic distances,however, can only occur across continents. Be-neath an ocean, where the crust is much thin-ner, Lg fails to propagate even short dis-tances.

Figure 3-10 illustrates how the Lg wavepropagates. At the top is shown a seismicsource within the Earth’s crust. Energy de-parting downward is partially reflected in thecrust and partially transmitted down into themantle. However, for waves that travel in themore horizontal direction, as shown in the mid-dle part of the figure, energy cannot get intothe mantle and is wholly reflected back intothe crust. The type of reflection occurring hereis the total internal reflection that is similarto that occurring within the prisms of a pairof binoculars. For a fixed source and receiver,there may be many reflection paths, all totally

47

Figure 3-7.—Regional and Teleseismic Signals

Regional signalP n P g

S n

1 1 Io 50 100

Time (seconds)

I Io 200 400

Time (seconds)

Upper seismogram is of a “regional distance” from the source. Signals here are associated with waves that propagate in thecrust and upper most mantle. Lower seismogram is of a “teleseismic distance. ” This wave has propagated through the deepinterior of the Earth.SOURCE” Modified from Defense Advanced Research Projects Agency.

Figure 3-8.—Pn Wave Figure 3-10.—Lg Wave

Source ReceiverSource Crust

(slow; low attenuation)Moho

\

Pn - wave

Figure 3-9.-Pg Wave

Source Receiver

Lower crustMoho

Mantle(faster; higher attenuation)

Total reflection

Wave guide. Many “trapped” rays in continental

Receiver

All totally reflected

Lg - wave

crust.

Moho

Pg - wave

48

reflected and thus trapped within the crust.This is illustrated in the bottom of the figure.The Lg wave is composed of the entire familyof trapped waves, with the crest acting as awave guide. For regions in which the crust iscomposed of material that transmits seismicwaves efficiently, Lg may be recorded at dis-tances of thousands of kilometers.

For most of the last 30 years, regional waveshave been thought of mainly in the context ofmonitoring small explosions which may not bedetected teleseismically. In recent years, how-ever, Lg has come to be recognized as usefulfor estimating yields for large explosions, in-dependent of the more conventional body waveand surface wave measurements (see ch. 7).

RECORDING SEISMIC WAVES

Seismic waves are measured at observato-ries around the world by recording the groundmotion. Most observatories have “triaxial”seismometers, meaning that they recordground motion in three directions at right an-gles to each other. Typically they are orientednorth-south, east-west, and vertical. By hav-ing all three components, seismologists canreconstruct the complete three-dimensionalground motion from the seismograms.

The principle by which the seismometerswork can bethought of as a heavy mass freelysupported by a spring from a frame fixed tothe Earth. When an earthquake or explosionoccurs, seismic waves traveling through theEarth reach the seismometer. The frame isshaken in response to the motion of the wave.Although the frame is displaced by the groundmotion, the heavy mass tends to remain sta-tionary because of its inertia. The displacementof the grounded frame relative to the station-ary mass is therefore a measure of the groundmotion. This movement is then electronicallymagnified so that displacements as small as0.00000001 centimeters (the same order asatomic spacings) can be detected.

If the Earth were perfectly still, recordingsmall earthquakes and underground explo-sions would be easy. However, processes suchas the winds, ocean waves, tides, man’s activ-ity, and seismic waves generated by earth-quakes in other regions continually cause mo-tions of the Earth. All of this motion is sensed

by seismometers and recorded. Although seis-mic instruments are sensitive enough to de-tect seismic waves generated by even thesmallest explosion, it will be seen in the fol-lowing chapters that naturally occurring back-ground noise levels are the limiting factor indetecting small earthquakes and explosions.

To reduce the background noise caused bywind and other surface effects, seismometershave been designed to fit into torpedo-shapedcasings and placed in narrow boreholes at adepth of about 100 meters. These types of sta-tions are currently being used as part of theRegional Seismic Test Network (RSTN). TheRSTN is a prototype system designed to evalu-ate the utility of in-country stations that couldbe placed within the Soviet Union to monitora low-yield or comprehensive test ban treaty.A typical high quality installation containsthree primary seismometers mounted in mutu-ally perpendicular orientations along withthree back-up seismometers (figure 3-11).

The seismometer installation is protectedagainst tampering both electronically andthrough the inherent ability of the instrumentto sense disturbances in the ground. The bore-hole package sends data to a surface stationthat contains transmitting and receivingequipment and in some cases to an antennafor satellite communications. Also containedin surface stations that might be used for mon-itoring are power sources, environmental con-

49

Figure 3-11 .—Seismic Instrumentation

Photo credits: Sandia National Laboratories

Seismic station downhole package containingseismometers, authentication circuits, and

processing electronics

Three seismometers mountedin mutually perpendicular

orientations

Seismometer

50

Photo credit: Sand/a National Laboratories

An example of what an internal seismic stationmight look like.

trol apparatus, and tamper-protection equip-ment. The distribution of six of these stationsin North America is designed to simulate thedistribution of 10 internal stations within theSoviet Union (figure 3-12).

Figure 3-12.— Distribution of RSTN Stations

Six Regional Seismic Test Network (RSTN) stations are dis-tributed throughout North American to simulate the distri-bution of 10 internal stations within the Soviet Union.

SOURCE: Modified from Sandia National Laboratories.

SEISMIC ARRAYSThe use of seismic arrays has been proposed

as an alternative and/or supplement to singlethree-component seismic stations. A seismicarray is a cluster of seismometers distributedover a fairly small area, usually on the orderof a few kilometers. The signals from the vari-ous instruments are then combined to improvethe detection and identification of any smallor weak signals that maybe recorded. Seismicarrays can function like phased array radarreceivers, sensitive to waves from a particu-lar direction while excluding waves from otherdirections. By doing so, arrays can pull smallsignals out from the surrounding backgroundnoise.

Figure 3-13.—Beamforming

To test the application of arrays for moni-toring regional seismic events, the UnitedStates and Norway installed the NorwegianRegional Seismic Array (NORESS) in 1984.NORESS is located north of Oslo and consistsof 25 individual sensors arranged in 4 concen-tric rings with a maximum diameter of 3 km.

Beamforming is a process of shifting and adding recordedwaveforms to emphasize signals and reduce noise. The origi-nal waveforms (a) can be shifted to remove propagation de-lays (b). The signal waveforms are now aligned but the noiseis not. The average of the shifted waveforms in (b) would bethe beam.

SOURCE: Energy & Technology Review, Lawrence Livermore National Labora-tory, August 1986, p. 13.

51

To imagine how an array works, consider anexample where a seismic wave is coming froma known location and with a known speed. Thetime it takes for the wave to travel to each sen-sor in the array can be predicted from theknown direction and speed. In contrast, theseismic background noise will vary randomlyfrom sensor to sensor. The recorded signalsfrom each sensor in the array can then beshifted in time to allow for propagation acrossthe array and combined (figure 3-13). The sen-sors are close enough so that the signal fromthe coherent source is nearly the same fromeach seismometer and remains unchanged. Thebackground noise varies more rapidly and can-cels out when the records are added together.Thus, it is possible to suppress the noise rela-tive to the signal and make the signal easier

to detect. This process of shifting and addingthe recorded waveforms is called beamform-ing. The combination of the shifted waveformsis called the beam.

Arrays do, however, have some drawbackscompared to three-component single stations:

1.

2.3.

A

they are more expensive because they re-quire more hardware and higher instal-lation costs;they require access to a larger area; andthey have more sensors that, in turn, gen-erate more data that must be trans-mitted, stored and processed.

comparison of the advantages and dis-advantages of arrays versus three-componentsingle stations is presented in chapter 4.

Chapter 4


CONTENTS

PageIntroduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55The Meaning of ’’Detection” . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55Locating Seismic Events . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57Defining Monitoring Capability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58

Seismic Magnitude . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58Converting Magnitude to Explosive Yield. . . . . . . . . . . . . . . . . . . . . . . . . . 60Seismic Monitoring in Probabilistic Terms . . . . . . . . . . . . . . . . . . . . . . . . . 60

Limitations to Seismic Monitoring Capability . . . . . . . . . . . . . . . . . . . . . . . . 61Seismic Noise.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61Reduction of Signals at Source or Sensor . . . . . . . . . . . . . . . . . . . . . . . . . . 61Seismic Instrumentation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62Seismic Magnitude Estimation Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . 63

Seismic Networks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64Existing Networks and Arrays . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64Planned Networks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67Hypothetical Networks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67

Seismic Monitoring Capability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68Calculating Seismic Monitoring Capability . . . . . . . . . . . . . . . . . . . . . . . . . 68Global Detection Capability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68Detection Capability Within the U.S.S.R. Using No Internal Stations . . 69Detection Capability Within the U.S.S.R. Using Internal Stations . . . . . 69Considerations in Choosing Monitoring Thresholds . . . . . . . . . . . . . . . . . . 70Use of High-Frequency Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70Arrays v. Three-Component Stations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74

BoxBox Page4-A. Locating A Seismic Event. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58

FiguresFigure No. Page4-1. Seismic Signals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 564-2. Seismic Noise . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 624-3. Effect of Noise on Event Magnitude Computation. . . . . . . . . . . . ., . . . 634-4. Contributing Seismograph Stations. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 654-5. Worldwide Standardized Network Station Distribution . . . . . . . . . . . . . 664-6. Detection Capability of a Seismic Station . . . . . . . . . . . . . . . . . . . . . . . . 684-7. Example of Calculated Detection Capability . . . . . . . . . . . . . . . . . . . . . . 71

Chapter 4


The first requirement for a seismic monitoring network is to detect and locateseismic events that could have been caused by an underground nuclear

explosion.

INTRODUCTIONThe first requirement for a seismic monitor-

ing network is that it be capable of detectingseismic events. If the Earth were perfectlyquiet, this would be easy. Modern seismome-ters are highly sophisticated and can detectremarkably small motions. However, processessuch as the winds, ocean waves, and even rushhour traffic continually shake the Earth. Allof this ground movement is sensed by seismo-meters and creates a background from whichsignals must be recognized. Seismic networks,consisting of groups of instruments, are de-

signed to detect events like earthquakes anddistinguish them from normal backgroundnoise. The extent to which a seismic networkis capable of detecting events, referred to asthe network’s detection threshold, is depen-dent on many factors. Of particular importanceare the types of seismic stations used, the num-ber and distribution of the stations, and theamount of background noise at the station loca-tions. This chapter reviews these factors anddiscusses the capability of networks to detectseismic events within the Soviet Union.

THE MEANING OF “DETECTION”

In practical terms, detecting a seismic eventmeans more than observing a signal above thenoise level at one station. There must be enoughobservations (generally from more than onestation) to estimate the location of the eventthat created the detected signal. Measure-ments of the seismic waves’ amplitudes andarrival times must be combined according tostandard analytical techniques to give the loca-tion and origin time of the event that gener-ated the signal. The event could be any of sev-eral natural or man-made phenomena, such asearthquakes, nuclear or chemical explosions,meteorite impacts, volcanic eruptions, or rockbursts.

The seismic signals from these events travelfrom their source to individual seismic record-ing stations along different pathways throughanon-uniform Earth. Consequently, no two sig-nals recorded at separate stations will be iden-tical. Even if they came from the same source,the amplitudes, shape, and transit times of the

signals would vary according to the path theytook through the Earth. This fact has an im-portant impact on the results of the calcula-tions used to determine magnitude and location.The results will never have perfect precision.They will be based on averages and will haveassociated with them statements of what thepossible errors in the most probable solutionmight be. Origin times and locations of seis-mic events, the parameters that make up a “de-tection, ” are always based on some averagingof the data from individual stations. To deter-mine the origin time, and location of a seismicevent, the data from several stations must bebrought together at a single place to carry outthe required analysis. Seismic stations thatroutinely send their data to a central locationfor analysis are said to form a network.

The energy radiating from an undergroundnuclear explosion expands outward. Whilemost of the explosive energy is dissipated bycrushing and melting the surrounding rock, a

55

76-584 0 - 88 : QL 3 - 3

56

small fraction is transformed into seismicwaves. These seismic waves propagate out-ward and, in so doing, encounter various bound-aries and rock layers both at the surface anddeep within the Earth. The boundaries causea separation of the seismic waves into a vari-ety of wave types, some of which travel deepthrough the interior of the Earth (body waves)and some of which travel along the surface (sur-face waves). See chapter 3 for a full discussionof wave types and travel paths.

Seismic signals are detected when they aresufficiently above the background noise insome frequency band. Figure 4-1 shows seis-mograms with standard filters designed to en-hance the detection of distant events. In thiscase, the signal can be seen because it is largerthan the noise at high frequency. When dataare in digital form (as they are in the latest

generation of seismic instrumentation), filtersare applied in various frequency bands beforedetection processing.

Relatively sophisticated techniques can nowbe used to detect a seismic signal in the pres-ence of noise. In those cases where three per-pendicular components of ground motion arerecorded using a vertical and two horizontalcomponent seismometers, use can also be madeof the known particle motion of P waves todifferentiate a P wave signal from backgroundnoise. Another technique which can enhancethe probability of detecting a signal requiresa number of closely spaced seismic sensorsknown as an array. The data recorded by thesesensors can be summed together in a mannerwhich takes account of the expected signalpropagation time across the array. The arrayenhances signals from great distance that prop-

Figure 4-1.—Seismic Signals

Signal

Small seismic signals in the presence of seismic noise at four different stations.

SOURCE: U.S. Geological Survey.

57

agate vertically through the array and can re-ject noise that travels horizontally. Therefore,the array summation process tends to enhancethe signal and to reduce the noise.

For many years, most seismic verificationefforts in the United States concentrated onthe use of teleseismic signals. Teleseismic sig-nals are seismic waves which travel to dis-tances greater than 2,000 km and go deepthrough the interior of the Earth. Teleseismicwaves are used because all seismometers mon-itoring Soviet testing are located outside theU.S.S.R. and generally at distances which are

teleseismic to the Soviet test sites. The possi-bility of establishing U.S. seismic stationswithin the U. S. S. R., close to Soviet nucleartesting areas, was discussed seriously as partof negotiations for a Comprehensive Test BanTreaty. Because techniques for monitoring ex-plosions with regional signals were not as wellunderstood as techniques for monitoring withteleseismic data, research efforts to improveregional monitoring greatly increased duringthe late 1970s. Regional distances are definedto be less than 2,000 kilometers and wavespropagating to this distance travel almost en-tirely through the Earth’s outer crustal layers.

LOCATING SEISMIC EVENTS

The procedure for estimating the location ofa seismic event, using seismic data, involvesdetermining four numbers: the latitude and lon-gitude of the event location, the event depth, andthe event origin time. Determination of thesefour numbers requires at least four separatemeasurements from the observed seismic sig-nals. These values are usually taken as the ar-rival time of the P wave at four or more differ-ent seismic stations. In some cases, however,determination of the numbers can be accom-plished with only two stations by using thearrival times of two separate seismic wavesat each station and by using a measure of thedirection of the arriving signal at each station.A relatively poor estimate of location can alsobe obtained using data from only a single array.

The event location process is an iterative onein which one compares calculated arrival times(based on empirical travel-time curves) withthe observed arrival times. The differences be-tween calculated and observed arrival timesare minimized to the extent possible for each

station in the process of determining the loca-tion of the seismic event.

All seismically determined event locationshave some error associated with them. The er-ror results from differences between the ex-pected travel time and the actual travel timeof the waves being measured and from impre-cision in the actual measurements. Generally,these errors are smallest for the P wave (thefirst signal to arrive), and this is why the ca-pability to detect P waves is emphasized. Loca-tion errors are computed as part of the loca-tion estimation process, and they are usuallyrepresented as an ellipse within which theevent is expected to be. Generally, this com-putation is made in such a way that there isa 95 percent probability that the seismic eventoccurred within the area of the error ellipse.Thus, location capability estimates are usuallygiven by specifying the size of this error ellipse.Similarly, there is always an uncertainty asso-ciated with the measured depth of the eventbeneath the Earth’s surface.

58

Box 4-A.—Locating A Seismic EventEstimating the location of a seismic event can be compared to deducing the distance to a light-

ning bolt by timing the interval between the arrival of the flash and the arrival of the sound ofthe thunder. As an example, consider the use of the P wave and S wave, where the flash is the first-arriving P wave and the thunder is the slower traveling S wave. The time interval between thearrival of the P wave and the arrival of the S wave (which travels at about half the speed of theP wave), increases with distance. By measuring the time between these two arrivals and knowingthe different speeds the two waves travel, the distance from the event to the seismometer couldbe determined (figure A). Knowing the distance from several stations allows the location to be pin-pointed (figure B).

2000 4000 6000 6000 10,000

Distance from earthquake (km)

The time required for P, S, and surface waves to travel a given dis-tance can be represented by curves on a graph of travel time againstdistance over the surface. To locate an earthquake the time intervalobserved at a given station is matched against the travel-time curvesfor P and S waves until the distance is found at which the separationbetween the curves agrees with the observed P-S time difference.Knowing the distance from the three stations, A, B, and C, one canlocate the epicenter as in figure B.

Knowing the distance, say X A , of an earthquake from a givenstation, as by the method Figure A, one can say only that theearthquake lies on a circle of radius A, centered on the station. If,however, one also knows the distances from two additionalstations B and C, the three circles centered on the three stations,with radii XA, XB, Xc, intersect uniquely at the point Q, thelocation of the seismic event.

SOURCE: This example is taken from Frank Press and Raymond Seiver, Earth (San Francisco, CA: W. H. Freeman and Company, San Fran-cisco, CA, 1974).

DEFINING MONITORING CAPABILITY

Seismic Magnitude of these tasks is generally described as a func-

A seismic monitoring system must be abletion of a measure-called the seismic magnitude

to detect the occurrence of an explosion, toof an event.

locate the explosion, to identify the explosion, Seismic magnitude was first developed asand to estimate the yield of the explosion. The a means for describing the strength of an earth-capability of a seismic network to perform each quake by measuring the motion recorded on

59

a seismometer. To make sure that the meas-urement was uniform, a standard method wasneeded. The original calculation procedure wasdeveloped in the 1930s by Charles F. Richter,who defined local magnitude, ML, as the log-arithm (to the base 10) of the maximum ampli-tude (in micrometers) of seismic waves observedon a Wood-Andersen torsion seismograph ata distance of 100 kilometers (60 miles) fromthe earthquake.

Subsequently, the definition of seismic mag-nitude has been extended, so that the meas-urement can be made using different types ofseismic waves and at any distance. For bodywaves, the equation for seismic magnitude mb

is:

mb = log (A/T) + B(d,h),

where A is the maximurn vertical displacementof the ground during the first few seconds ofthe P wave, and T is the period of the P wave.The B term is a correction term used to com-pensate for variations in the distance (d) be-tween the seismic event and the recording sta-tion and the depth (h) of the seismic event. Fora seismic event at the surface of the Earth,h=0. The B correction term has been deter-mined as a function of d and h by observingseismic signals from a large number of earth-quakes.

For surface waves, seismic magnitude Ms

can be calculated by a similar equation:

M s = log (A/T) + b log d + C,

where b and c are numbers determined fromexperience. A number of formulas, involvingslightly different values of b and c for MS,have been proposed.

The terms in both the body wave and sur-face wave magnitude equations that are usedto compensate for distance reflect an impor-tant physical phenomenon associated withseismic wave propagate. This phenomenon, re-ferred to as attenuation, can be simply statedas follows: the greater the distance any par-ticular seismic wave travels, the smaller thewave amplitude generally becomes. Attenua-

tion of wave amplitude occurs for a numberof reasons including:

1. the spreading of the wave front over agreater area, thereby reducing the energyat any one point on the wavefront;

2. the dissipation of energy through natu-ral absorption processes; and

3. energy redirection through diffraction,refraction, reflection and scattering of thewave at various boundaries and layerswithin the Earth.

As a consequence, a correction term is neededto obtain the same magnitude measurementfor a given seismic event from data taken atany seismic station. The correction term in-creases the amplitude measurement to com-pensate exactly for the amplitude decreasecaused by the different attenuation factors.

Therefore, if the amplitude of the seismicwave is to be used to estimate the size of theseismic event (whether it is an explosion or anaturally occurring earthquake), a good under-standing of how amplitude decreases with dis-tance is needed for both body and surfacewaves. The distance-dependent numbers in thebody and surface wave equations representaverage corrections which have been developedfrom many observations. In general, these cor-rections will not be exact for any one particu-lar path from a particular seismic event to aparticular sensor. It is important, therefore,either to calibrate the site to receiver path orto compute the magnitude of the event usingseismic signals recorded at a number of well-calibrated stations. If multiple stations areused, the event magnitude is calculated byaveraging the individual station magnitudesfor an event. From this procedure, an averagebody wave magnitude (rob) and an averagesurface wave magnitude (Ms are determinedfor the event.

Obviously, the distance the wave has trav-eled must be known to determine the attenua-tion in amplitude and correct for it. Thereforethe seismic event must be located before itsmagnitude can be determined.

60

Converting Magnitudeto Explosive Yield

The detection and identification capabilitiesof seismic networks are described most con-veniently in terms of seismic magnitudes, typi-cally mb. This measure is used because mb isdirectly related to seismic signal strength.When interpreting capabilities in terms of ex-plosive yield, however, an additional step isrequired to translate mb to kilotons. The samemagnitude value can correspond to yields thatrange over a factor of about 10. Variations inthe magnitude-yield relationship are caused byvariations in the structure of the Earth in thevicinity of the test site (low signal attenuationversus high signal attenuation areas), the ma-terial in which the explosion is emplaced (hard,water-saturated rock versus dry, porous ma-terials), and the way in which the explosion isemplaced (tamped versus detonated in a largecavity designed to muffle the signal).

For example, if an explosion is “well-coupled,”that is, if the energy is well transmitted fromthe explosion to the surrounding rock, an mb

of 4.0 corresponds to an explosion of about 1kt. This relationship is true only for explosionsin hard rock and may vary considerably de-pending on how well the seismic waves aretransmitted through the area’s geology. Inareas that are geologically old and stable, seis-mic waves are transmitted more efficiently. Anmb of 4.0 produced by a well-coupled explo-sion in an area of good transmission might cor-respond to an explosion much smaller than akt. In areas that are geologically young andactive, seismic waves are not transmitted asefficiently and an mb of 4.0 may correspondto a well-coupled explosion larger than 1 kt.

Even greater changes in the relationship be-tween mb and yield can occur if the explosionis intentionally “de-coupled” from the sur-rounding rock in a deliberate attempt to muf-fle the seismic signal. As we will see in chap-ter 6, decoupling can be accomplished at lowyields under some situations by detonating theblast in a large underground cavity. Throughsuch evasion methods, the same 1 kt explosion

that produced a magnitude mb of 4.0 when“well-coupled” might be muffled down to aseismic signal of around mb 2.() at low fre-quencies. Lesser reductions can be accom-plished by detonating the explosion in dry po-rous material.

Because the yield that corresponds to a spe-cific mb depends so much on the scenario thatis being discussed, seismologists generally useseismic magnitude to describe monitoring ca-pabilities. In translating seismic magnitudesto yields, the reader must consider the contextin which the comparison is made. In particular,it should be considered whether the explosionis being recorded in an area of good transmis-sion and whether the explosion is well-coupled.Unless specifically stated, this report trans-lates seismic magnitudes to yields correspond-ing to “tamped” conditions, that is, a well-coupled explosion in hard rock. Situationswhere decoupling is feasible, and the effectsof such decoupling, are discussed in chapter 6.

Seismic Monitoring in ProbabilisticTerms

Whether seismic measurements are made byhand or by computer, some error is involved.Even greater additional errors arise from theimperfect estimates of how well seismic wavestravel through different parts of the Earth andhow well seismic energy is coupled to the Earthduring the explosion. All of these errors resultin some uncertainty in the final determined pa-rameters. This is true whether these parame-ters are event magnitude, location, identifica-tion characteristics, or yield. In all cases,however, it is possible to estimate a confidencefactor in probability terms for the determinedparameter. It is important to realize, therefore,that while the numbers are not presented with100 percent certainty, estimates of the uncer-tainty are known. In general, this uncertaintyis greatest for the small events and decreasesfor the larger events. A discussion of the un-certainty and what it means in terms of na-tional security concerns is presented in chap-ter 2.

61

LIMITATIONS TO SEISMIC

The strength of a seismic signal diminisheswith distance. In general, the closer the seis-mic station is to the source, the stronger thesignal will be. Hence, a principal element ofmonitoring strategy is to get close to areas ofconcern. It follows that the more high qualitystations distributed throughout a given area,the greater the capability will be to detect smallevents.

Seismic Noise

As noted previously, if the Earth were per-fectly still, detecting even the smallest seis-mic event would be easy. However, the Earth’ssurface is in constant motion. This motion isthe result of many different energy sources.Major storms over ocean areas and the result-ant wave action on continental shores causesignificant noise in the 2- to 8-second band.Wind noise and noise from atmospheric pres-sure fronts are particularly prominent onhorizontal-component seismic recordings. Thesemore or less continuous motions of the Earthare referred to as seismic noise or microseisms(figure 4-2). For purposes of siting a seismicstation, it is highly desirable to find an areathat has a low background level of seismicnoise. Generally, the lower the backgroundseismic noise at any station, the smaller willbe the seismic signal which can be detected atthat station.

Cultural activities can also generate seismicnoise that appears in the frequency range usedto monitor nuclear explosions. Generally, thisman-made noise has frequencies higher than1 Hz. Heavy machinery, motors, pumping sta-tions, and mills can all generate observableseismic noise. However, careful siting of seis-mic stations can minimize the problem of mostman-made seismic noise. From a monitoringpoint of view, it is important that noise sur-veys be made prior to the final selection of sitesfor seismic stations. If such sites are negoti-ated within other countries, provisions shouldbe made for relocating the sites should seis-mic noise conditions change.

MONITORING CAPABILITY

Seismic signals and noise are concentratedin various frequency bands. Only the noisewithin the frequency bands in which seismicsignals are observed is a problem. Even strongnoise can be eliminated by filtering as long asthe noise is outside the detection bands of in-terest.

The possibility also exists that a seismic sig-nal from one event can be masked by the seis-mic signal from another event. While this doesindeed happen, it is only a problem for moni-toring at yields around 1 kt or less without in-ternal stations. For events of interest in theU.S.S.R. above a magnitude of about mb 4.0,there are a sufficient number of stations de-tecting the event so that masking of the sig-nal at a couple of stations generally poses noserious problem. For events much below mb

4.0, a number of stations at regional distances(distances less than about 2,000 km) wouldhave to be used to avoid the masking problem.Such stations would be available if the UnitedStates obtains access to data from seismic sta-tions placed within the U.S.S.R.negotiated agreement betweenStates and the Soviet Union.

Reduction of SignalsSource or Sensor

as part of athe United

at

Poor coupling to the geologic media, eitherat the explosion source or at the seismic sen-sor, will act to reduce the amplitude of the seis-mic signal received. If the explosion is in densehard rock or in water saturated rock, the sourcecoupling will be good. If the explosion is in al-luvium, dry porous rock, or within a cavity,the coupling will not be as good. Decouplingan explosion by detonation within a cavity isan important evasion scenario which will bediscussed in chapter 6.

At the seismic sensor, signal reduction canoccur if the sensor is not placed on hard rock.In particular, if there is a layer of soil uponwhich the sensor sits, the signal-to-noise ratioat the sensor can be far less than if the sensor

62

Figure 4-2. —Seismic Noise

-

Background seismic noise at three different stations.

SOURCE: U.S. Geological Survey.

were placed on or within hard rock. Sensorsplaced in boreholes in hard rock provide su-perior coupling to the Earth and also providea more stable environment for the instrumentpackages, with a concomitant reduction innoise.

Seismic Instrumentation

Until recently, the instrumentation that wasavailable for detecting, digitizing, and record-

ing seismic signals did not have the capabilityto record all the signal frequencies of interestwith sufficient range. Further, the mechani-cal and electronic components comprising seis-mic recording systems generate internal noise,which is recorded along with true ground noiseand seismic event signals, and this internalnoise was the limiting factor in recording seis-mic signals in the high frequency range. Spe-cifically, the internal noise of the older designsof high frequency seismic detectors was higher

63

than ground noise at frequencies above about5 Hz. Thus, while ground noise is now knownto decrease with increasing frequency, the sys-tem noise remained constant or increased withincreasing frequency in the older systems.Therefore, trade-offs were made, and the en-tire frequency range of the signal was gener-ally not recorded. Specifically, in the high fre-quency range data was generally not recordedabove 10 Hz and even then was highly con-taminated in the 5-10 Hz range by internal sys-tem noise. Consequently, small seismic events,particularly small explosions, with expectedmaximum signal energy in the high frequencyrange above 5 Hz were not detected becausetheir high frequency signals were below the in-ternal system noise levels. Most existing seis-mic stations are of this type and so are limitedfor nuclear test monitoring.

Today, broadband systems capable of re-cording the entire frequency range with a largedynamic range and with low internal noise areavailable. However, the best high performancesystems are not widely distributed. To estab-lish confident detection-identification capabil-ities using high frequency seismic signals atlow event magnitudes, it will be necessary toexpand the number of high performance sta-tions and to place them in diverse geologic envi-

ronments in order to simulate the requirementsof in-country monitoring.

Seismic Magnitude EstimationProblems

As discussed earlier, the estimate of anevent’s seismic magnitude is made by combin-ing the estimates obtained from many singlestations in an averaging procedure to reducerandom errors. This procedure works well foran event which is neither too small nor toolarge.

For a small event, however, the averagingprocedure can result in a network magnitudevalue which is biased high. This follows fromthe fact that for a small event, the signals willbe small. At those stations where signals fallbelow the noise, the small signal amplitudeswill not be seen. Consequently, only higher am-plitude values from other stations are avail-able for use in computing the network aver-age. With the low values missing, this networkaverage is biased high unless a statistical cor-rection is made.

Figure 4-3 illustrates this effect. All six sta-tions (A through F) record the same magni-tude 4 event. Stations A, B, and C record the

Figure 4-3.—Effect of Noise on Event Magnitude Computation

4.8

4.24.5

3.2

STA-A STA-B STA-C

STA-D

STA-E

I

seismicnoiselevel

Average event magnitude = 4.0 (6 stations) STA-F

Computed average magnitude = 4.5 (3 stations)

SOURCE: Modified from Air Force Technical Applications Center.

.

64

event below average. Stations D, E, and F rec-ord the event above average. Normally, the sta-tions would all average out to a magnitude 4.0.For a small event, however, the stations thatrecord low (A, B, &C) do not record the signalbecause it is below the noise level. Only thestations that record higher (D, E, & F) showthe event. Without the values from the low sta-tions, the calculation of the average is madeusing only the stations that record high. Theresulting calculation biases the average to thevalues of the higher stations, giving a falseaverage magnitude value of (in this example)4.5 for a 4.0 seismic event.

In the past, computed magnitudes of smallevents were systematically biased high in thismanner. As a result, for most of the last 25years, the U.S. capability to detect seismicevents within the U.S.S.R. has, in fact, beensignificantly better than the estimates of thiscapability. Not until the late 1970s was it dem-

onstrated that the small events being detectedwere 0.2 to 0.4 magnitude units smaller thanpreviously thought. In terms of yield, thismeans that the networks were actually capa-ble of detecting events down to half as largeas previously thought possible. Within the lastfew years, analysis procedures have been em-ployed to correct for most of this bias usinga procedure called maximum likelihood esti-mation.

For large events, a similar bias problem usedto exist occasionally, but for a different rea-son. Old seismometers could not record verylarge signals without clipping the signal. Largeramplitude signals were either not available orwere under estimated. The resulting bias oflarge events, however, did not affect estimatesof detection capability and only became a prob-lem in the determination of the size of verylarge events.

SEISMIC NETWORKS

Existing Networks and Arrays

Although many thousands of seismic sta-tions exist around the world, the actual num-ber of stations which routinely report data tonational and international data centers is a fewthousand. For example, in figure 4-4 the 3,500stations are shown that routinely report datato the National Earthquake Information Cen-ter (NEIC), a center in Colorado operated bythe United States Geological Survey. Some ofthese stations report much more often thando others. The instrumentation at these sta-tions is diverse and the quality of the data var-ied. While these stations are very useful forseismic signal detection, they are less usefulfor purposes of magnitude estimation and forresearch requiring stations evenly distributedaround the world.

For purposes of treaty monitoring opera-tions and research, a well distributed networkwith a common set of instrumentation at allstations is most useful. To obtain such a stand-ard network, the United States funded the de-

velopment and deployment of the WorldwideStandardized Seismograph Network (WWSSN)in the early 1960s (figure 4-5). The WWSSNis maintained by the United States Geologi-cal Survey (USGS). The quality and perform-ance of the WWSSN is generally very good,but the recording system is limited in dynamicrange and resolution because of the use of whatis now obsolete analog equipment and also be-cause of high internal noise in the amplifyingequipment. (For example, the data are cur-rently recorded only on photographic paperrecords.)

Beginning in the early 1970s, digital record-ing seismic stations were developed by theUnited States and other countries. The datafrom these stations can be easily processed bydigital computers to enhance the signal-to-noise for signal detection and to analyze thedata for seismic source determination and forresearch purposes. These stations are includedin such networks as:

. The Regional Seismic Test Network(RSTN). These are high quality stations

●

●

● ✎●

●●

✎

.

/’-7’-.

o●

●

●

●

z

● I

●

●

● 9●

●

. ●

. “

66

● ’

67

●

●

●

designed by the Department of Energyand now operated by the USGS. They wereintended to be prototypes of in-countrystations. There are five RSTN stations dis-tributed over North America at inter-station distances that represent monitor-ing in the Soviet Union with 10 internalstations.The NORESS seismic array in Norway.This seismic array is funded by the De-fense Advanced Research Projects Agency(DARPA) and the Department of Energy(DOE). The prototype array is located insouthern Norway, an area thought to begeologically similar to the western partof the Soviet Union.The recently installed China Digital Seis-mic Network (CDSN), which is a coopera-tive program between the People’s Repub-lic of China and the USGS.The Atomic Energy Detection System(AEDS) seismic network. This network isoperated by the Air Force Technical Ap-plications Center (AFTAC). The purposeof the AEDS network is to monitor treatycompliance of the Soviet testing program.Consequently, the stations are located soas to provide coverage primarily of theU.S.S.R. The capabilities of the AEDSnetwork are described in a classified annexto this report. The USGS and the AEDSstations are the main sources of routineinformation on Soviet testing.

In addition to these networks, there is alsoa jointly operated NRDC-Soviet Academy testsite monitoring network in the United Statesand the Soviet Union. The network consistsof three stations in each country around theKazakh and Nevada test sites at distances ofabout zOO kilometers from the boundaries ofeach test site. These stations are supplyinghigh-quality seismic signal data, in the highand intermediate frequency range from 0.1 Hzto about 80 Hz. The stations are designed tobe modern prototypes of the in-country seis-mic stations required for monitoring a lowthreshold test ban treaty and they are notlimited by system noise in the high frequencyrange. Plans call for the addition of five more

such stations distributed across the SovietUnion and for several more to be similarly dis-tributed across the United States.

Planned Networks

There are a number of planned new networksthat will provide increased capability to detect,locate, and characterize seismic events aroundthe world. These networks are being developedby the United States and other countries.

Hypothetical Networks

Existing unclassified networks external tothe U.S.S.R. have an excellent capability formonitoring events with seismic magnitudesgreater than 4.0 within the U.S.S.R. However,for explosions less than a few kt, the possibil-ity exists that the seismic signals from suchexplosions could be reduced through an eva-sion method. To demonstrate a capability todefeat credible evasion scenarios that could beapplied to explosions with yields less than afew kt, seismic stations within the SovietUnion would be necessary.

Obviously, there are a number of require-ments for such internal stations and their data.Among these are the following: the data mustbe provided in an uninterrupted manner; thedata must be of high quality; the seismic noiseat the stations should be low; the operatingparameters of the stations and the character-istics of the data should be completely knownat all times; the data should be available tothe United States within a reasonable timeframe; the stations should be located for ef-fective monitoring of the U. S. S. R.; and any in-terruption or tampering with the operation ofthe station should be detectable by the UnitedStates. Obviously, these requirements canmost easily be achieved by deploying U.S. de-signed and built seismic stations within theU.S.S.R. at sites chosen by the United States.

The number of stations required within theU.S.S.R. is a function of a number of factorsincluding: the threshold level down to whichmonitoring is desired, the seismic noise at thestations, the signal-to-noise enhancement ca-

68

pability of the stations, the signal propagation posed distributions of internal stations capa-characteristics within the U. S. S. R., and the ble of detecting seismic events down to vari-possibilities for various evasion scenarios ous thresholds. The number of internal stationsthought to be effective within different areas proposed for the various distributions rangesof the U.S.S.R. Many seismologists have pro- between 10 and 50.

SEISMIC MONITORING CAPABILITY

Calculating Seismic MonitoringCapability

Calculating the detection capability of ex-isting seismic stations is straightforward. Forhypothetical stations, however, the detectioncapability must be estimated by adopting anumber of assumptions. Because there existsa range of possible assumptions which can beargued to have validity, there are also a rangeof possible capabilities for a network of hypo-thetical internal stations.

For existing stations, the average detectioncapability is easily determined by observingthe number of seismic events detected as afunction of log amplitude (figure 4-6). The sta-tion can be expected to detect all events withina given region down to some magnitude level.A cumulative plot of the detections, such asillustrated in figure 4-6, will show that a straightline can fit these values down to this magni-tude threshold level. This threshold marks the

Figure 4-6.—Detection Capability of a Seismic Station

1,000

o 3

Log amplitude

SOURCE: Modified from Air Force Technical Applications Center

point where the station fails to detect allevents. The 90-percent detection threshold (orany other threshold) can be determined fromthis plot. If the event magnitude rather thanthe observed amplitude is used, it is importantthat all magnitudes used in such plots be cor-rected for low-magnitude bias as previouslydiscussed. By examining all stations of a net-work in this manner, the station detection pa-rameters can be determined and used to com-pute overall network performance for the givenregion.

For hypothetical internal stations, no detec-tion statistics are generally available for com-puting the cumulative detection curve as afunction of magnitude. Therefore, the detec-tion capability must be estimated by assum-ing the following factors: the seismic noise atthe station, the propagation and attenuationcharacteristics of the region through which thesignals will travel, the efficiency of the seis-mic source, and the signal-to-noise ratio re-quired for the signal to be detected. All theabove factors must be evaluated in the fre-quency range assumed to be best for detect-ing a signal. The result of all these factors,when considered together, is to provide an esti-mate of the probability that a signal from asource of a given size and at a given locationwill be detected at a certain station. Individ-ual station detection probabilities are thencombined to determine the overall probabilitythat four or more stations will detect the event.Translating this capability to situations whereevasion might take place requires additionalconsiderations (see ch. 6).

Global Detection Capability

There are many seismic stations that existaround the world from which data can be ob-

69

tained. While no attempt has ever been madeto determine the global detection capabilityof all these stations, a rough estimate couldbe made by reviewing the various reportingbulletins and lists. However, the current globaldetection capability does not really matter be-cause it will soon change as various plannednetworks become installed. Consequently, anaccurate assessment of global capability is bestaddressed by discussing planned networks.

For regions external to the U. S. S. R., and par-ticularly for regions of the Southern Hemi-sphere, the greatest detection and location ca-pability will reside not with the AEDS, butwith a number of existing and planned seis-mic networks which are unclassified. This isa logical consequence of the AEDS being tar-geted primarily at events within the U.S.S.R.In particular, national networks such as thoseof Australia, China, the United States, Italy,and Canada will provide significant global ca-pability.

Given all the national and global data sources,a cautious estimate of the global detection ca-pability by the year 1991 (assuming 90 per-cent confidence of four or more stations detect-ing an event using only open unclassifiedstations) is mb 4.2. For many regions, such asthe Northern Hemisphere, the detection capa-bility will be, of course, much better. There-fore, by 1991, any explosion with a magnitudecorresponding to 1-2 kt well-coupled that is det-onated anywhere on Earth will have a highprobability of being detected and located bynetworks external to the U.S.S.R. Opportuni-ties to evade the seismic network outside theSoviet Union are limited. Because evasionscenarios require large amounts of clandestinework, they are most feasible within the bordersof a closed country such as the Soviet Union.Consequently, monitoring networks are de-signed to target principally the Soviet Union.

Detection Capability Within theU.S.S.R. Using No Internal Stations

Given that a large range of possible networksexists, a few type examples are useful to con-vey a sense of what can be accomplished. For

example, the capability of a hypothetical net-work consisting of a dozen or so seismic ar-rays that are all outside the borders of theSoviet Union can be calculated. A cautious esti-mate is that if such a network were operatedas a high-quality system, it would have 90 per-cent probability of detecting at four or morestations all seismic events within the SovietUnion with a magnitude at least as low as 3.5.This corresponds to an explosion having a yieldbelow 1 kt unless the explosion is decoupled.

The hypothetical detection threshold of mb

3.5 is considered cautious because it is knownthat a greater detection capability might ex-ist at least for parts of the U.S.S.R. For exam-ple, the single large NORSAR array in Nor-way has the potential to achieve detectionthresholds equivalent to an event of mb 2.5 orlower (corresponding to a well-coupled explo-sion between 0.1 and 0.01 kt) overlarge regionsof the Soviet Union.1

Also, fewer stations (fewer than the fourneeded above) may detect much smaller events,and this can provide useful information. How-ever, detection by one or a few stations maynot be adequate to, with high confidence, locateor identify events. Also, reductions of the de-tection threshold must be accompanied by acomparable capability to locate the events (forfocusing other intelligence resources) and toseparate nuclear explosions from earthquakesand legitimate industrial explosions.

Detection Capability Within theU.S.S.R. Using Internal Stations

Seismic stations located internalto a coun-try for the purpose of monitoring have a num-ber of important advantages: improved detec-tion capability, improved location capability,and improved identification capability.

The potential instantaneous detection thresholds for the largeNORSAR array (42 seismometers spread over an area of about3,000 km’), as described in “Teleseismic Detection at High Fre-quencies Using NORSAR Data” by F. Ringdal in IVORSARSemiannual ?’ecfmical Summary, Apr. I-Sept. 30, 1984, are:

West of Ural Mountains-mb 2.0-2.5 (possibly better)Caspian Area–rob 2.0-2.5Semipalatinsk-mb 2.5-3.0Siberia–mb 2.5-3.5

70

Although much debate is associated with thepredicted detection capabilities of internal sta-tions, improved detection capability alone isprobably not of the greatest significance at thistime because the current detection capabilityis already very good. The improvement thatinternal stations will provide to identificationcapability (differentiating explosions from nat-ural events) is by far the most important rea-son for requiring internal stations and shouldbe considered the basic requirement for inter-nal stations. The problem of detecting andidentifying seismic events in the face of vari-ous evasion scenarios will be discussed in thenext two chapters.

Based on cautious assumptions for a net-work of 30 internal arrays or about 50 three-component internal stations, it appears likelythat a detection threshold of mb 2.5 (90 per-cent probability of detection at four or morestations) could be reached. This correspondsto a well-coupled explosion of 0.1-0.01 kt, ora fully decoupled nuclear explosion with a yieldof about 1 kt. Based on more optimistic as-sumptions about the conditions to be encoun-tered and prospective improvements in dataprocessing capability, this same network couldhave a detection capability as low as mb 2.0.Detection capability contours for one such pro-posed 30-array internal network are shown infigure 4-7.

Considerations in ChoosingMonitoring Thresholds

Depending on the number of internal sta-tions, detection capabilities could either in-crease or decrease. The point, however, is thatvery low detection thresholds, down to mag-nitude 2.0, can be achieved. In fact, almost anydesired signal detection level can theoreticallybe obtained by deploying a sufficient numberof internal stations; although there maybe dis-agreements over the number and types of sta-tions needed to achieve a given threshold. Thedisagreements could be resolved as part of alearning process if the internal network is builtup in stages.

Another consideration is that all detectionestimates used in this report are based on a90-percent probability of four or more stationsdetecting an event. While this maybe a pru-dent estimation procedure from the monitor-ing point of view, an evader who did not wishto be caught might adopt a considerably morecautious point of view. (See chapter 2 for a dis-cussion of the relationship between uncertaintyand cheating opportunities.) Such concernsmight be increased by the realization that formany seismic events, there will beat least onestation that will receive the signal from theevent with a large signal-to-noise ratio. The sig-nal will be so large with respect to the noiseat this station that the validity of the signalwill be obvious and will cause a search for otherassociated signals from neighboring seismicstations. The possible occurrence of such a sit-uation would be of concern to a country con-templating a clandestine test.

Throughout all of this discussion it must bekept in mind that an improvement in detec-tion capability does not necessarily corresponddirectly to an improvement in our monitoringcapability. Although a reduced detection thresh-old must be accompanied by a reduced iden-tification threshold; occurrences such as indus-trial explosions might ultimately limit theidentification threshold. For example, the esti-mate of detection capability for internal sta-tions which is given above, mb 2.0 to 2.5, cor-responds to a decoupled explosion of about 1kt. A decoupled 1 kt explosion produces thesame mb signal as a 1/70 kt (15) ton chemicalexplosion. At such small magnitude levels,there are hundreds of chemical explosions det-onated in any given month in the U.S.S.R. andin the United States.

Use of High-Frequency Data

It has long been known to seismologists thatseismic signals of moderate to high frequen-cies can indeed be detected at large distancesfrom the source under favorable geological con-ditions. Such is the case in the eastern UnitedStates, and it is generally thought that thisis also true of most of the Soviet Union. In con-

71

72

trast, tectonic regions such as the westernUnited States and the southern fringe of theU.S.S.R. are generally characterized by strongerattenuation of high frequencies, which are there-fore lost beyond relatively short distances.

The design and development of a nuclearmonitoring strategy based on high-frequencyseismic signals calls for:

● A determination of Earth structure withinand around the region to be monitored,and an evaluation of its signal transmis-sion characteristics at high frequencies.

● A methodology for identifying and select-ing sites with low ground noise, and equip-ping such sites with high performance sen-sors and recording systems, so as to achievethe largest possible signal-to-noise ratio.

● A reliable understanding of the high-fre-quency radiation of natural (earthquakes)and man-made (explosions) seismic sources.Although empirical evidence based on di-rect observations is sufficient in principle,a predictive capability, based on theory,is required to assess properly new and un-tested monitoring conditions.

Such requirements parallel in every respectthe usual constraints placed on standard mon-itoring systems. However, direct experimen-tation pertinent to high-frequency monitoringhas been rather limited so far, and the relevantdata available today are neither abundant nordiverse. Consequently, an assessment of whetherthese requirements can be met relies of neces-sity on some degree of extrapolation from ourpresent experience, based on theoretical argu-ments and models. This situation leaves roomfor debate and even controversy.2

Recently, it has been argued that the capa-bility to detect and identify low-yield nuclearexplosions could be greatly improved by using

‘High quality seismic data is now becoming available fromthe NRDC-Soviet Academy of Sciences stations in the SovietUnion and the United States, with more widely distributed sta-tions to be added in 1988 in both countries. This data may helpreduce the necessity for extrapolation and decrease the uncer-tainties that foster the debate.

high-frequency (30 -40 Hz) seismic data.3Themajor points of the argument are:

●

●

●

that natural seismic ground noise levelsare very low at high frequencies, and thatlarge seasonal fluctuations are not antic-ipated;that careful station selection could makeit possible to emplace seismic sensors inparticularly quiet sites; andthat present seismic recording technologyallows high-fidelity recording; by suppres-sion of system-generated noise to levelsbelow ground noise even at high frequen-cies and at quiet sites.

Advocates of high-frequency monitoring ex-plain the efficiency of high-frequency wavepropagation observed in the North Americanshield in terms of a simple model for attenua-tion of seismic waves in stable continental re-gions and argue that the model applies as wellto stable continental Eurasia. Finally, they ar-

retical models of earthquakes and undergroundexplosions provide adequate predictive esti-mates of the relative production of high-fre-quency energy by various seismic sources andjustify their choice by comparison with limitedobservations.

Based on these arguments and a systematicmodeling procedure, some seismologists rea-son that the most favorable signal-to-noise ra-tio for detection of low yield (i.e. small magni-tude) events in stable continental areas will befound at moderate to high frequencies (about30 Hz). They further infer that a well-designednetwork of 25 internal and 15 external high-quality stations using high frequencies wouldyield multi-station, high signal-to-noise detec-tion of fully decoupled 1-kt explosions from anyof the potential decoupling sites within theU. S. S. R., and result in a monitoring capabil-ity at the l-kt level.

On the other hand, the inference that suchsignificant benefits would necessarily accrue

‘For example, J. F. Evemden, C. B. Archarnbeau, and E. Cran-swick, “An Evaluation of seismic Decoupling and UndergroundNuclear Test Monitoring Using High-Frequency Seismic Data, ”Reviews of Geophysics, vol. 24, No. 2, 1986, pp. 143-215.

73

by relying on high-frequency recordings hasbeen strongly questioned in the seismologicalcommunity. Indeed many scientists feel thatthe case is currently unproven. Major pointsof disagreement include:

● The concern that the theoretical seismicsource models used so far in the analysisdescribed above are too simple. Studiesaimed at constructing more realistic modelsindicate that the high-frequency wavesgenerated by seismic sources are stronglyaffected by complexities of source be-havior that the simple models do not takeinto account. On the other hand, advo-cates of high frequency seismic monitor-ing believe that the models they have usedhave successfully predicted a number ofcharacteristics of seismic sources thatwere subsequently verified and that noneof the many well-documented observa-tions of seismic wave characteristics fromlarge events are in conflict with their theo-retical model predictions. Thus they ar-gue that the model predictions, for some-what smaller events at somewhat higherfrequencies than are ordinarily studied,are reasonable extrapolations.

● The concern that it may be difficult toidentify candidate station sites where thehigh-frequency noise is sufficiently low topermit actual realization of the desiredbenefits. Experience to date is limited, andone does not really know whether a givensite is suitable until it has been occupiedand studied for at least a year. On theother hand, advocates of high frequencymonitoring feel that suitable low-noisesites are not at all rare and can be rathereasily found in most, if not all, geologicenvironments within the continents. Theyargue that stations selected so far havehad adequately low high-frequency noisecharacteristics and the selection processwas neither difficult nor time consuming.They conjecture that doubts are based onmisidentification of high-frequency inter-nal seismic recording system noise asground noise, and that once high-perfor-

mance systems with low system noise be-come more wide-spread, this concern willdisappear.

● The concern that observations which canbe employed to test directly the validityof the proposed use of high frequencies areas yet quite scant, and their interpreta-tion is not free of ambiguities. For exam-ple, the characterization of source spec-tra and the propagation and attenuationof high-frequency waves remain issueswhich are not resolved unequivocally byobservations, and yet are critical to theformulation of a high-frequency monitor-ing strategy. Similarly, available dataoften exhibit an optimal signal-to-noise ra-tio at frequencies near 10 Hz, in apparentdisagreement with the arguments enun-ciated earlier. In response to these con-cerns, proponents argue that the NRDC-Soviet observations of signal-to-noise ra-tios greater than 1 and out to frequenciesabove 20 Hz provides evidence for thepotential of high frequency monitoring.While proponents of high frequency mon-itoring agree that the observations of thelargest signal to noise ratios for signalsfrom seismic events often occur near 10Hz, this is not in disagreement with thepredictions or technical arguments ad-vanced for high-frequency monitoring.They argue that these observations are ob-tained from seismic receivers with systemnoise that is greater than ground noise atfrequencies above 10 Hz, and that manyof the observations are from industrial ex-plosions that are ripple-fired and so areexpected to have lower high-frequencycontent than a small nuclear test.

● The concern stated earlier that the moti-vation for the proposed high-frequencymonitoring approach uses contested theo-retical arguments and models to extrapo-late from our present experience and thusattempt to guide further steps towards asignificantly improved capability. In thepresent case, models are used to extrapo-late both toward high frequencies andtoward small yields, and just how far one

74

may extrapolate safely remains a matterof debate. The proponents of high fre-quency monitoring agree that the data re-lating to high frequency monitoring islimited with respect to the geologic re-gions to which it pertains. Furthermore,it is clear that experience in the system-atic detection and identification of verysmall seismic events using high-frequencydata is absent and that, as a consequence,it has been necessary to extrapolate fromexperience with larger seismic eventswhere lower frequency data is used. Propo-nents believe, however, that what limiteddata is available does support the mostcritically important predictions; these be-ing the apparent availability of low-noisesites and the efficiency of high-frequencywave propagation to large distances.

These controversial aspects notwithstand-ing, there is general agreement among seismol-ogists that good signal-to-noise ratios persistto higher frequencies than those used routinelytoday for nuclear monitoring. In particular,data in the 10-20 Hz band show clear signalswhich are undoubtedly not used optimally.Given the fact that recording of even higherfrequencies is demonstrably feasible in somesituations, and given the potential advantagesfor low-yield monitoring, the augmentation ofour experience with such data, the concomi-tant continued development of appropriateanalysis techniques to deal with them, and thevalidation of the models used in their interpre-tation are goals to be pursued aggressively.Not until a sufficient body of well-documented

observations of this nature has been collectedcan we expect to achieve a broad consensusabout the performance of high-frequency mon-itoring systems.

Arrays v. Three-Component Stations

Both small-aperture, vertical-component ar-rays such as NORESS, and three-component,single-site stations such as the RSTN stationhave been considered for use as internal sta-tions. In choosing which to use, it should berealized that many combinations of arrays andsingle stations will provide the same capabil-ity. For example, for any array network, thereis a single station network with comparablecapability; but the network of single stationsprobably requires about twice as many sites.Although a single array has advantages overa single three-component station (see chapter3), for monitoring purposes it is preferable tohave a large number of station sites with three-component stations rather than to have a smallnumber of sites with arrays. This is true be-cause it permits better accommodation to de-tails of regional geology, and better protectionagainst noise sources temporarily reducing ca-pability of the network as a whole. However,if the number of sites is limited by negotiatedagreement, but the instrumentation can in-clude either arrays or three-component sta-tions, then arrays are preferable. This is trueboth because of the inherent redundancy of ar-rays and their somewhat better signal-to-noiseenhancement capability over single three-com-ponent stations.

Chapter 5

Identifying Seismic Events

CONTENTS

Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Basis for Seismic Identification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Earthquakes . . . . . . . . . .Chemical Explosions . . .Rockbursts . . . . . . . . . . .

Methods of IdentificationLocation . . . . . . . . . . . . .Depth . . . . . . . . . . . . . . .Ms:mb . . . . . . . . . . . . . . .Other Simple Methods .Spectral Methods. . . . . .High-Frequency Signals

Identification Capability .Identification Capability

Stations . . . . . . . . . . . .Identification Capability

Box5-A. Theory and5-B. Progress in

Figure No.5-i.5-2.5-3.

5-4.5-5.5-6.5-7.

5-8.

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . ....!! . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Within the U.S.S.R. Using No Internal

.Within the U.S.S.R Using Internal

Boxes

Stations::

77787884858585858688888990

9191

Observation. . . . . . . . . . . . . . . . . . . . . . . . . . . . .Seismic Monitoring . . . . . . . . . . . . . . . . . . . . . .

Figures

. . . . . . . . . . 82

. . . . . . . . . . 92

Seismicity of the World, 1971 -1986 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Seismicity of the U.S.S.R. and the Surrounding Areas, 1971-1986 . . . .Cross-Sectional View of Seismicity Along the Kurile-KamchatkaCoast . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Earthquakes v. Explosions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Ms v. mb for a Suite of Earthquakes and Explosions.. . . . . . . . . . . . . .Msv. mb . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .The mb:MsDiscriminant for Populations of NTS Explosions andWestern U.S. Earthquakes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .The VFM Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

7980

81838686

8789

Table No. Page5-l. Approximate Numbers of Earthquakes Each Year, Above Different

Magnitude Levels . . . . . . ........ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78

Chapter 5

Identifying Seismic Events

Once a seismic event has been detected, the next step is to determine whetherit was created by an underground nuclear explosion.

INTRODUCTION

Once the signals from a seismic event havebeen detected, and the event located, the nextstep in seismic monitoring is that of identifi-cation. Was the event definitely, or possibly,an underground nuclear explosion, or can thesignals be identified unambiguously as hav-ing another cause? As discussed in the previ-ous chapter, seismic signals are generated bynatural earthquakes and by natural rockburstsin mines, as well as by chemical and nuclearexplosions. The identification problem in seis-mic monitoring, called the discrimination p r o b -lem, is to distinguish underground nuclear ex-plosions from other seismic sources.

In the case of a located seismic event forwhich signals are large, that is, larger thancould be ascribed to a chemical explosion ora rockburst, the only candidates for the sourceof the signals are an earthquake or a nuclearexplosion. Physical differences between earth-quakes and nuclear explosions cause their seis-mic signals to differ, and these differences canbe used to identify the events.

Identification becomes more complicated,however, for small events. When identifyingsmall events (comparable in size to an explo-sion of less than 10 kilotons), the identifica-tion procedures encounter four types of diffi-culty that do not arise for larger events:

1.

2.

3.

4.

the quality of available signals is typicallylower;the number of natural events that mustbe discriminated against is larger;the possibilities now include chemical ex-plosions and rockbursts;the event, if nuclear, is of a size where at-

tempts may be made to muffle or hide theseismic signal.

Each of these difficulties becomes more severeat lower yields.

The last difficulty listed above brings up thesubject of evasion which is discussed in thenext chapter. For the purposes of understand-ing this chapter, however, the reader shouldrecognize that identification capabilities mustalways be considered against the feasibility ofvarious evasion scenarios. The successful useof a muffling or decoupling evasion techniquecould cause a 1-10 kt decoupled nuclear explo-sion to produce seismic signals comparable toeither a chemical explosion of 10-100 tons re-spectively, or an earthquake ranging frommagnitude 2-3 respectively. The similarity insize and, in some cases, the properties of thesignals for these three types of events pose themost serious monitoring challenges. The mon-itoring system must be able to demonstratea capability to identify with high confidenceseismic events whose signals might be inten-tionally reduced or hidden through credibleevasion techniques. It is the need to demon-strate such a capability, as signals becomesmaller, that ultimately sets the threshold forseismic identification.

The present chapter describes the differentkinds of seismic sources and the basis for solv-ing the identification problem. It then describescapabilities of current seismic networks, andhow these capabilities might be improved bythe addition of new seismic stations, includ-ing stations within the U.S.S.R.

77

78

BASIS FOR SEISMIC IDENTIFICATION

The seismic signals from a nuclear explosionmust be distinguished from seismic signals cre-ated by other events, particularly earthquakes,chemical explosions, and rockbursts. Mostseismic signals from located events are causedby earthquakes, although chemical explosionsare also present in large numbers. Significantrockbursts and other sources of seismic sig-nals are rare. A monitoring network, however,will encounter seismic signals created by allof these sources and must be able to identifythem. Consequently, the burden on any moni-toring network will be to demonstrate a capa-bility to detect and identify with high confi-dence a clandestine nuclear test against thebackground of large numbers of earthquakesand industrial explosions and infrequent rock-bursts. This section reviews basic propertiesof these other seismic sources and discussesthe physical basis for discriminating them fromnuclear explosions.

Earthquakes

Earthquake activity is a global phenomenon,though most of the larger earthquakes are con-centrated inactive tectonic regions along edgesof the Earth’s continental and oceanic plates(figure 5-l). In the U. S. S. R., most large earth-quakes are located in a few active regions alongthe southern and eastern borders of the coun-try (figure 5-2). For the many earthquakesoccurring along the Kurile Islands region onthe Pacific border of the Soviet Union, the shal-low earthquakes generally occur on the oceanside of the islands, while the deeper earth-quakes occur beneath the islands and towardthe U.S.S.R. landmass (figure 5-3). Elsewhere,earthquakes can occur as deep as 700 kilome-ters in the Earth’s mantle.

In figure 5-2, large areas of the U.S.S.R. areshown for which there is no significant earth-quake activity. Given presently available in-formation, such areas are referred to as aseis-mic; although some activity may occur in theseregions at magnitudes below mb 3.0.

As table 1 illustrates, there are about 7,500earthquakes that occur with mb 4.0 or aboveeach year, and about 7 percent of these occurin the Soviet Union. However, approximatelytwo-thirds of the Soviet seismicity occurs inoceanic areas, mainly off the Kurile-Kamchatkacoast. Seismicity in oceanic areas does not raisean identification problem because acoustic sen-sors provide excellent identification capabil-ity for nuclear explosions in water. Conse-quently, it is earthquakes on Soviet land areasfrom which nuclear explosions need to be dif-ferentiated. For example, at the magnitudelevel of mb 4.0 and above, where there are7,500 earthquakes each year, approximately183 occur on Soviet land areas and would needto be identified. The number of such earth-quakes that occur each year above variousmagnitudes is shown in the third column oftable 1. The smaller the magnitude, the moreearthquakes there are.

Earthquakes and their associated seismo-grams have been studied on a quantitative ba-sis for about 100 years. Since 1959, seismolo-gists have engaged in a substantial researcheffort to discriminate between earthquakes andexplosions by analyzing seismic signals. Thou-sands of studies have been conducted andreports written. As will be discussed in the fol-lowing section, this identification problem isnow considered to be solved for events abovea certain magnitude.

Table 5-1.—Approximate Numbers of EarthquakesEach Year, Above Different Magnitude Levels

m b Globally a Soviet Unionb Soviet land areasc

5.0 950 70 234.5 2,700 200 674.0 7,500 550 183

21,000 1,500 5003.0 59,000 4,200 1,4002.5 170,000 12,000 4,000

aBased On F. Ringdal’s global study of a 10-year period, for which he finds thestatistical fit log N = 7.47-O.9m . F. Ringdal, “Studyof Magnitudes, Seismicity,

Jand Earthquake Detectability sing a Global Network,” in The Vela Program,Ann U. Kerr, (cd.), Defense Advanced Research Projects Agency, 1985.

bBa9ed on 7 percent of global earthquakes with numbers rounded uP slightlycBa9ed on removing two-thirds of Soviet Union earthquakes. Two-thirds of theearthquakes in the Soviet Union occur in oceanic areas, e.g., off the Kurile-Kamchatka coast and, therefore, do not present an identification problem.

79

b

●

●

●

80

● ●

81

Figure 5-3.—Cross-Sectional View of Seismicity Along the Kurile-Kamchatka Coast

Kurile Islands Kurile-Kamchatka Trench

/ Pacific Ocean

Two-thirds of Soviet seismicity at magnitudes greater than mb 3.5 is off the Kurile-Kamchatka coast. This sectional view showsthat almost all these events are deep, below land or shallow, below ocean. Neither case is a candidate explosion location.

SOURCE: L.R. Sykes, J.F. Evernden, I.L. Cifuentes, American Institute of Physics, Conference Proceed/r@ 704, pp. 85-133, 1983.

From these studies, a number of identifica-tion methods have evolved. These methods allhave as their basis a few fundamental differ-ences between nuclear explosions and earth-quakes. These differences are in:

●

●

●

location,geometrical differences between the pointsource of an explosion and the much largerrupture surface of an earthquake, andthe relative efficiencies of seismic wavegeneration at different wavelengths.

With regard to the first difference, manyearthquakes occur deeper than 10 km, whereasthe deepest underground nuclear explosionsto date have been around 2.5 km and routineweapons tests under the 150 kt threshold ap-pear all to be conducted at depths less than

1 km. Almost no holes, for any purpose, havebeen drilled to more than 10 km deep. The ex-ceptions are few, well known, and of a scalethat would be difficult to hide. Because nuclearexplosions are restricted to shallow depths (lessthan 2.5 km), discrimination between manyearthquakes and explosions on the basis ofdepth is possible in principle. In practice, theuncertainty in depth determination makes thedivision less clear. To compensate for the pos-sibility of very deep emplacement of an explo-sion, the depth must be determined to be be-low 15 km with high confidence before theevent is to be identified unequivocality as anearthquake. Events with shallower depths thatshow large uncertainties in depth determina-tion might be considered as unidentified on thebasis of depth.

82

With regard to source geometry, the funda-mental differences are due to how the seismicsignals are generated (see box 5-A). As dis-cussed in chapter 3, an underground nuclearexplosion is a highly concentrated source ofcompressional seismic waves (P waves), sentout with approximately the same strength inall directions. This type of signal occurs be-cause the explosion forces apply a fairly uni-form pressure to the walls of the cavity cre-ated by the explosion. The simple model of anuclear explosion is a spherically symmetricsource of P waves only. This contrasts withan earthquake, which is generated as a resultof massive rock failure that typically produces

Box 5-A.—Theory and Observation

One of the points on which early efforts atdiscrimination stumbled was the claim thatexplosions, because they are concentratedpressure sources, would generate insignifi-cant SH waves and Love waves. (These wavesentail types of shearing motion, in which theground moves in a horizontal direction.) Butit was soon found, in the early 1960s, thatsome explosions generated quite strong SHand Love waves, so this discriminant cameto be seen as unreliable.

A salutary further effect was that discri-minants based purely on theoretical predic-tions came to carry little weight until theyhad passed stringent tests with actual datafrom explosion and earthquake sources. Oc-casionally, some useful discriminants are dis-covered empirically and a theoretical under-standing is not immediately available. Arecent example of this is the observation thatregional phases Pn, Pg, and Lg for smallearthquakes in the western U.S. typically ex-hibit more high-frequency energy (at 6-8 Hz)than do small NTS nuclear explosions of com-parable seismic magnitude.* However, in thesymbiotic relation between theory and obser-vation, there is generally a framework of un-derstanding in which progress in one fieldguides workers to new results in the other.*J-R. Murphy and T.J. “A Discrimination Analysis ofShort-Period Regional Seismic Data Recorded at Tonto ForestObservatory, ” BuLletin of the Seismological Society of Amer-ica, vol. 72, No. 4, August 1982, pp. 1351-1366.

a net shearing motion. It is common here tothink in terms of two blocks of material (withinthe crust, for shallow events) on opposite sidesof a fault. The stress release of an earthquakeis expressed in part by the two blocks movingrapidly (on a time scale of seconds) with respectto each other, sliding in frictional contact overthe plane of the fault as a result of the spon-taneous process of stress release of rock—stress that has accumulated over time throughgeologic processes. Because of this shearingmotion, an earthquake radiates predominantlytransverse motions–i.e., S waves, from allparts of the fault that rupture. Though P wavesfrom an earthquake are generated at about 20percent of the S-wave level, they have a four-lobed radiation pattern of alternating compres-sions and rarefactions in the radiated first mo-tions, rather than the relatively uniform pat-tern of P-wave compressions radiated in alldirections from an explosion source. Theseidealized radiation patterns,. P waves from anearthquake and an explosion, are shown sche-matically at the top of figure 5-4. These twotypes of source geometries result in the follow-ing differences:

1.

2.

Energy partitioning. The energy of a seis-mic event is partitioned into compres-sional, shear, and surface waves. Earth-quakes tend to emit more energy in theform of shear waves and surface wavesthan explosions with comparable compres-sional waves (figure 5-4). At lower magni-tudes, however, differences of this typemay be difficult to distinguish if the sur-face waves from either source become toosmall to detect.Dimensions of the source. Earthquakestend to have larger source dimensionsthan explosions because they involve alarger volume of rock. The bigger thesource volume, the longer the wavelengthof seismic waves setup by the source. Asa result, earthquakes usually emit seismicwaves with longer wavelengths than ex-plosions. Instead of describing the signalin terms of wavelength, an equivalentdescription can be given in terms of seis-mic wave frequency. Explosions, which

83

Figure 5-4.—Earthquakes v. Explosions

Earthquake

p +

Short period I Long period(P waves) (surface waves)

II

Earthquake

I

I

I

+

IExplosion

— .

II

5 seconds I 5 minutesI

(Upper) Different radiation patterns for earthquakes and explosions. Earthquakes involve shear motion along a fault plane.Explosions are compressional sources of energy and radiate P waves in all directions.(Lower) Recorded signals (i.e., seismograms) of earthquakes and explosions generally have different characteristic features.Note the much stronger P-wave: surface-wave ratio, for explosions as compared to the earthquake.SOURCE: F. Ringdal, “Seismological Verification of Comprehensive Test Ban Treaty,” paper presented in “Workshop on Seismological Verification of a Comprehensive

Test Ban Treaty,” June 4-7, 1985, Oslo, Norway.

are small, intense sources, send out stronger been observed and the distinction maysignals at high frequencies; and earth- diminish for small events. The differencequakes generally have more low-frequency in frequency content of the respective sig-(long-wavelength) energy than explosions nals, particularly at high frequencies, is(figure 5-4). However, exceptions have a topic of active research.

84

These basic differences in seismic signal areunderstood on theoretical grounds; and thistheoretical framework guides the search fornew empirical methods of identification. Asnew types of seismic data become available (forexample, data from within the Soviet Union),it maybe expected that methods of identifica-tion will be found that work at smaller magni-tude levels. At low magnitudes, however, wehave not only the problem of distinguishingnuclear explosions from earthquakes, but alsoof distinguishing nuclear explosions from thelarge chemical explosions that are commonlyused for industrial purposes such as mining.This is a much more difficult problem becausein theory the two source types should producesimilar signals.

Chemical Explosions

Chemical explosions are used routinely in themining and construction industries, and theyoccur also in military programs and on nucleartest sites. In general, from the seismic moni-toring perspective, a chemical explosion is asmall spherical source of energy very similarto a nuclear explosion. The magnitudes causedby chemical explosions are generally below mb

4.0, although events with magnitudes up tomb 4.5 occasionally occur. The fact that largenumbers of chemical explosions in the yieldrange from 0.001 to 0.01 kt are detected andlocated seismically is a testament to the capa-bility of local and regional seismic networksto work with signals below mb 3.0.

There is little summary information avail-able on the number and location of chemicalexplosions in the U. S. S. R., though it appearsthat useful summaries could be prepared fromcurrently available seismic data. In the UnitedStates, chemical explosions around 0.2 kt arecommon at about 20 mines. At each of thesespecial locations, tens of such explosions mayoccur each year. Presuming that similar oper-ations are mounted in the U. S. S. R., this activ-ity is clearly a challenge when monitoring nu-clear explosions with yields below about 1 kt.It is also a challenge when considering the pos-sibility that a nuclear testing nation may seek

to muffle a larger nuclear explosion (say, around5 kt), so that its seismic signals resemble thoseof a much smaller (say, around 0.1 kt) chemi-cal explosion.

The problem of identifying industrial explo-sions can be partially constrained in threeways:

1.

2.

In the United States, almost all chemicalexplosions detonated with yields 0.1 kt orgreater for industrial purposes are in facta series of more than a hundred small ex-plosions spaced a few meters apart andfired at shallow depth with smalltime de-lays between individual detonations. Oneeffect of such “ripple-firing” is to gener-ate seismic signals rather like that of asmall earthquake, in that both these typesof sources (ripple-fired chemical explo-sions, and small earthquakes) occur overa large area and thus lose some of the char-acteristics of a highly concentrated sourcesuch as a nuclear explosion. If individualsalvos are large enough, however, theymight be able to mask the signals froma sub-kiloton nuclear explosion. Recent re-search on this problem has shown the im-portance of acquiring seismic data at highfrequencies (30-50 Hz); and indeed, suchdata suggest the existence of a distinctivesignature for ripple-fired explosions.1 23

Many of the rare chemical explosionsabove about 0.5 kt that are not ripple firedcan be expected to result in substantialground deformation (e.g., cratering).4 Suchsurface effects, together with absence ofa radiochemical signal, would indicate achemical rather than a nuclear explosionsource. The basis for this discriminant is

IA.T. Smith and R.D. Grose, “High-Frequency Observationsof Signals and Noise Near RSON: Implications for Discrimina-tion of Ripple-fired Mining Blasts,” LLNL UCID-20945, 1987.

2A.T. Smith, “Seismic Site Selection at High Frequencies: ACase Study, ” LLNL UCID-21047, 1987.

9D.R. Baumgardt, “Spectral Evidence for Source Differencesbetween Earthquakes and Mining Explosions in Norway,” Seis-mologi”cal Researcb Letters, vol. 38, January-March 1987, p. 17.

41t is also relevant that chemical explosions in the 0.1 -0.5kt range are commonly observed to be quite efficient in gener-ating seismic waves. For example, 0.5 kt chemical explosionsare known to have m~ around 4.5 for shield regions.

85

3.

that chemical explosions are typicallyvery shallow.To the extent that remote methods ofmonitoring chemical explosions are deemedinadequate (for example, in a low yieldthreshold or comprehensive test ban re-gime), solutions could besought by requir-ing such constraints as prior announcementof certain types of chemical explosions, in-spections, shot-time monitoring, and in-country radiochemical monitoring. Notethat in the United States (and so perhapsalso in the U. S. S. R.) chemical explosionsabove 0.1 kt occur routinely at only alimited number of sites.

Rockbursts

In underground mining involving tunnelingactivities, the rock face in the deeper tunnels

may occasionally rupture suddenly into thetunnel. This is referred to as a rockburst andresults from the difference between the lowpressure existing within the tunnel and thegreat pressure that exists within the surround-ing rock. To prevent such mine rockbursts,bracing structures are used in the deepertunnels.

In terms of magnitude, rockbursts are allsmall. They occur over very restricted regionsof the Earth and generally have a seismic mag-nitude of less than 4.0 mb.

The source mechanisms of rockbursts arevery similar to those of small earthquakes. Inparticular, the direction of the first seismic mo-tion from a rockburst will have a pattern simi-lar to that for earthquakes. Therefore, for theseismic identification problem, rockbursts canbe considered small earthquakes which occurat very shallow depths.

METHODS OF IDENTIFICATION

Over the years, a number of identificationmethods have been shown to be fairly robust.Some of these methods perform the identifi-cation process by identifying certain earth-quakes as being earthquakes (but not identify-ing explosions as being explosions). Otheridentification methods identify certain earth-quakes as being earthquakes and certain ex-plosions as being explosions. The identificationprocess is therefore a winnowing process.

Location

The principal identification method is basedon the location of a detected seismic source.If the epicenter (the point on the Earth’s sur-face above the location) is determined to be inan oceanic area, but no hydroacoustic signalswere recorded, then the event is identified asan earthquake. Large numbers of seismic eventscan routinely be identified in this way, becauseso much of the Earth’s seismic activity occursbeneath the ocean. If the location is determinedto be land, then in certain cases the event canstill be identified as an earthquake on the ba-

sis of location alone, e.g. if the site is clearlynot suitable for nuclear explosions (such asnear population centers) or if there is no evi-dence of human activity in the area.

Depth

With the exception of epicenter locations,seismic source depth is the most useful dis-criminant for identifying large numbers ofearthquakes. A seismic event can be identifiedwith high confidence as an earthquake if itsdepth is determined to be below 15 kilometers.

The procedure for determining source depthis part of finding the event location using thearrival time of four or more P-wave signals.Also, certain seismic signals, caused by energythat has traveled upward from the source,reflected off the Earth’s surface above thesource region, and then traveled down into theEarth, are similar to the P wave out to greatdistances. A depth estimate can be obtainedby measuring the time-difference between thefirst arriving P-wave energy and the arrivaltime of these reflected signals. The analysis

86

of broadband data through wave-form model-ing is particularly useful for detecting thesereflected signals. Empirical methods can alsobe used, based on comparison with previouslyinterpreted seismic events in the same generalregion as the event under study.

In principle, an advantage of depth as anidentification method is that it is not depen-dent on magnitude: it will work for small eventsas well as large ones, provided the basic dataare of adequate signal quality and the signalsare detected at a sufficient number of stations.It will not alone, however, distinguish betweenchemical and nuclear explosions unless the nu-clear explosion is relatively large and deep; orbetween underground nuclear explosions andearthquakes unless the earthquakes are suffi-ciently deep.

M s:m b

Underground nuclear explosions generatesignals which tend to have surface wave mag-nitude (Ms) and body wave magnitudes (rob)that differ from those of earthquake signals.This is basically a result of explosions emit-ting more energy in the form of body waves(high-frequency seismic radiation), and earth-quakes emitting more energy in the form ofsurface waves (low frequency seismic radia-tion). The phenomenon is often apparent in theoriginal seismograms (figure 5-4), and exam-ples are shown in terms of Ms:mb diagrams infigures 5-5, 5-6, and 5-7. These diagrams canbe thought of as separating the population ofexplosions from that of earthquakes. For anyevent which is clearly in one population oranother, the event is identified. For any eventwhich is between the two populations (as oc-casionally happens for explosions at the Ne-vada Test Site), this method does not providereliable identification. The method, therefore,has the potential of identifying certain earth-quakes as being earthquakes and certain ex-plosions as being explosions.

To use this identification method, both mb

and MS values are required for the event. Thisis no problem for the larger events, but forsmaller events (below mb 4.5) it can be verydifficult to detect low-frequency surface waves

Figure 5-5.—Ms v. mb for a Suite ofEarthquakes and Explosions

7.0

6.0

4.0

2.0 4.0 5.0 6.0

M s

SOURCE: P.D. Marshall and P.W. Basham, Geophysical Journal of the Royal As.tronomical Society, vol. 28, pp. 431-458, 1972.

Figure 5-6.—MS v. mb

9

8

7

6

o = Earthquakes

o

0

0 0 0

. 0 . o 0 0

oA

I A

A

5 6 7 8mb

SOURCE: L.R. Sykes, J.F. Evernden, I.L. Cifuentes, American Institute of Phys-ics, Conference Proceeding 104, pp. 85-133, 1983. (All of the events(both earthquakes and explosions) were corrected for bias.]

using external stations alone.5 This difficultyis present particularly for explosions, with the

$rn ~Pi~ studies of redon~ data that evaluate the Ms:mbdiscriminant, it has been found that between 10 percent and20 percent of the surface waves of small events are masked bysignals caused by other events. This, however, has little neteffect on the discrimin ation of events for which more than onestation is close enough to receive surface waves.

87

Figure 5-7.—The mb:Ms Discriminant for Populations of NTS Explosions and Western U.S. Earthquakes

6

2o

x “x

- +

.1

1 I I I 13 4 5 6 7 8

M b

X ExplosionSOURCE: Modified from S.R. Taylor, M.D. Denny, and E.S. Vergino, Regional mb:Ms Discrimination of NTS Explosions and Western United States Earthquakes A Progress

Report, Lawrence Livermore National Laboratory -January - 1986. “ -

result that the Ms:mb method works intrinsi-cally better for identifying small earthquakesthan it does for identifying small explosions.Internal stations would provide important ad-ditional capability for obtaining Ms values forevents down to mb 3.0 and perhaps below.

It is known that Ms:mb diagrams, with theirseparate populations as shown in figures 5-5,5-6, and 5-7, tend in detail to exhibit somewhatdifferent properties for sources occurring indifferent geophysical provinces. That is, thebest identification capabilities are obtained,for a particular event of interest, if there is al-ready a data base and an associated Ms:mb

diagram tailored for the general region of theEarth in which that event occurred.

The Ms:mb method has proven to be themost robust identification technique availablefor shallow events. For events below mb 4.0,the separation between the two Ms:mb popu-lations has been found to decrease in somestudies. In the opinion of many seismologists,however, a useful separation (at low magni-tudes) may be possible if good quality data forsuch small events can be found. Again, in thecontext of monitoring for small undergroundexplosions in the U. S. S. R., obtaining such datawould require in-country stations and empiri-cal confirmation.

76-584 0 - 88 : QL 3 - 4

88

Other Simple Methods

Several other methods, with applications inparticular circumstances, are used to provideidentification. Several are still in the researchstage, and are having an impact on the designof new instruments and the development ofnew procedures in data analysis. Some of thesemethods, based on the whole spectrum of fre-quencies contained in a seismic signal, are de-scribed in the following section. Other meth-ods have been used for decades, and are stilloccasionally important for use in identifyingcertain problem events. In order of decreas-ing importance, the remaining three simplemethods can be listed as follows:

1. The use of “first motion.” By this term ismeant an identification method based ondifferences in the direction of the initialmotion of P waves. As illustrated in fig-ure 5-4, an explosion is expected to createinitial compressive motions in all direc-tions away from the source, whereas anearthquake typically creates initial com-pressive motions in some directions andrarefactional motions in others. In a com-pressive wave, the ground first movesaway from the source. In a rarefaction, theground first moves toward the source.From a good seismometer recording, it isoften possible to observe the direction ofthis initial motion (for example, from ob-servation of whether the ground firstmoves up or down at the seismometer).This identification method can be power-ful if the signal-to-noise ratio is large. Butfor small events in the presence of seis-mic noise (as discussed in chapter 4) it canbe difficult or impossible to determine thedirection of the first motion. Because itis never clear that rarefactional motionsmay not exist in directions for which net-work coverage is poor, the method at bestidentifies an earthquake as an earthquake,but cannot unequivocally identify an ex-plosion as an explosion.

2. The observation of S waves. Because of thecompressive nature of explosion sources,explosions typically generate less shearwave energy than do most earthquakes.

3.

Therefore, the observation of significantshear wave energy is indicative that theevent is probably an earthquake.Complexity. Many explosion-generatedbody wave signals tend to be relativelysimple, consisting of just a few cycles.Many earthquake-generated body wavesignals tend to be relatively complex, con-sisting of a long series (known as the coda)following the initial few P-wave cycles.The concept of complexity was developedin an attempt to quantify this differencein signal duration. Complexity is the com-parison of amplitudes of the initial partof the short-period signal with those of thesucceeding coda. There are cases, however,where an explosion signal is complex atsome stations and an earthquake signalis simple. Therefore, the complexitymethod is regarded as not as reliable asMs:mb. In practice, whenever the com-plexity method works, other identificationmethods also work very well.

Spectral Methods

The basis of the success of Ms:mb diagramsis, in part, the fact that the ratio of low fre-quency waves to high frequency waves is typi-cally different for earthquakes and explosions.Thus, Ms is a measure of signal strength ataround a frequency of 0.05 Hz, and mb is ameasure of signal strength at around a fre-quency of 1 Hz. As a way of exploiting suchdifferences, Ms:mb diagrams are quite crudecompared to methods that use a more completecharacterization of the frequency content ofseismic signals. The analysis of earthquake andexplosion signals across their entire spectrumof frequencies is an important component ofthe current research and development effortin seismic monitoring.

Because Ms:mb diagrams work well whensurface waves are large enough to be measured,the main contribution required of more sophis-ticated analysis is to make better use of theinformation contained in the P-wave signalsand other large amplitude, high frequency sig-nals. This offers the prospect of improved iden-tification capabilities just where they are most

89

needed, namely for smaller events. Thus, in-stead of boiling down the P-wave signals atall stations simply to a network mb value, onecan measure the amplitude at a variety of fre-quencies and seek a discriminant which workson systematic differences in the way earth-quake and explosion signals vary with fre-quency. One such procedure is the variablefrequency-magnitude (VFM) method, in whichthe short-period body wave signal strength ismeasured from seismograms filtered to passenergy at different frequencies, say, fl and f2.6

Discrimination based on a comparison of mb(f1)and mb(f2) in many cases shows a clear sepa-ration of earthquake and explosion populations(figure 5-8).7 89

An analogy for the VFM discriminant wouldbe a person’s ability to tell by ear the differ-ence between a choir with loud sopranos andweak contraltos, and a choir with loud con-traltos and weak sopranos. Ms:mb is like com-paring basses and contraltos. For small events,VFM is complementary to Ms:mb in severalways: VFM is improved by interference of themain P wave with waves reflected from the sur-face above the explosion, while Ms:mb isdegraded; VFM is more effective for hard rockexplosions, while Ms:mb will preferentially dis-criminate explosions in low-velocity materials;and VFM is insensitive to fault orientation.Further, the Ms:mb method requires averag-ing over observations from a network of sta-tions surrounding the event, while the VFMmethod may be applied at a single station atany distance and direction from the source. Itis found in practice, however, that the VFMmethod is most reliable when several high per-formance stations are used. (Obviously if sev-

‘W.B. Archambeau, D.G. Harkrider and D.V. Hehnberger, U.S.Arms Control Agency Report, “Study of Multiple SeismicEvents,” California Institute of Technology, ACDA/ST-220,1974.

7J.F. Evernden, “Spectral Characteristics of the P-codas ofEurasian Earthquakes and Explosions,” Bulletin of the Seis-mological Society of America, vol. 67, 1977, pp. 1153-1171.

‘J.M. Savino, C.B. Archambeau, and J.F. Masse, Discri~”-nation Results From a Ten Station Network, DARPA ReportSSS-CR-79-4566, S-CUBED, La Jolla, CA, 1980.

‘J.L. Stevens and S.M. Day, “The Physical Basis for m~:M,and Variable Frequency Magnitude Methods for Earth-quake/Explosion Discrimination, ’ Journal of Geophysical Re-search, vol. 90, March 1985, pp. 3009-3020.

eral stations provide VFM data independentlyidentifying an event, as an explosion for ex-ample, then assigning a numerical probabilitythat the event observed was indeed an explo-sion can be accomplished with more reliabil-ity and confidence.) VFM is more sensitive tonoise and regional attenuation differences thanMs:mb, but the main point is that simultane-ous use of both methods should allow improveddiscrimination of small events.

High-Frequency Signals

The use of high-frequency signals in seismicmonitoring is strongly linked to the questionof what can be learned from seismic stationsin the U.S.S.R. whose data is made availableto the United States. Such “in-country” or “in-ternal” stations were considered in CTBT ne-gotiations in the late 1950s and early 1960s,and the technical community concerned withmonitoring in those years was well aware thatseismic signals up to several tens of Hz couldpropagate from explosions out to distances of

Figure 5-8.—The VFM Method

8.0 I I I Earthquakes

Presumed explosions o7.0

t 00 80

mb [2.25 Hz]

The body-wave magnitude is determined at two different fre-quencies. Earthquakes and explosions fall into separatepopulations because explosions are relatively efficient atgenerating higher frequencies.SOURCE: J.L. Stevens and S.M. Day, Journal of Geophysical Research, voI 90,

pp. 3009-3020, 1985.

90

several hundred kilometers. The seismic wavesmost prominent at these so-called “regionaldistances” (by convention, this means dis-tances less than about 2,000 km) are knownas Pg,Pn,Sn, and L (see ch. 3 for a discus -sion of these regional waves). Pg propagateswholly within the crust; Pn and Sn travel atthe top of the mantle just below the base ofthe crust; and Lg, guided largely by the crus-tal layer, is often the strongest signal and issometimes observed across continents even atdistances of several thousand kilometers.

However, this early interest in regionalwaves lessened once the Limited Test BanTreaty of 1963 was signed and it was recog-nized that subsequent programs of large un-derground nuclear explosions could be moni-tored teleseismically (i.e., at distances beyond2,000 km), so that internal stations were notneeded. At teleseismic distances, seismic bodywave signals are usually simpler and have in-deed proved adequate for most purposes of cur-rent seismic monitoring, even under treaty re-strictions on underground nuclear testing thathave developed since 1963. Thus, it is in thecontext of considering further restraints ontesting, such as a comprehensive or low-yieldtest ban, that the seismic monitoring commu-nity needs to evaluate high-frequency regionalseismic signals. The basis for the role of high-frequency monitoring, in such a hypotheticalnew testing regime, is that the greatest chal-lenge will come from decoupled nuclear explo-sions of a few kilotons, and for them the mostfavorable signal-to-noise ratios may beat highfrequency. Internal stations, of high qualityand at quiet sites, will be essential to addressthe monitoring issues associated with suchevents.

It is clear that a dedicated program to makeoptimal use of high-frequency seismic data

from internal stations should lead to signifi-cant improvements in detection capability.10

The daily recording of many very small seis-mic events, together with the possibilities forevasion at low-yields, focuses attention on thediscrimination problem for events in the mag-nitude range mb 2.0 to mb 4.0.

For an explosion that is well-coupled to theground, an mb = 2.0 can be caused by onlya few thousandths of a kiloton, that is, a fewtons of TNT equivalent. At this level, a moni-toring program would confront perhaps 1,000to 10,000 chemical explosions per year. Giventhe difficulty of discriminating between chem-ical and nuclear explosions, it is clear that be-low some level events will still be detected butidentification will not be possible with high con-fidence. However, recognizing that relativelyfew chemical explosions occur at the upper endof this mb range (2.0 - 4.0), where many chem-ical explosions can be identified by character-istics of their signals over a broad frequencyband (see the earlier discussion of ripple-firedchemical explosions), high frequency datarecorded within the U.S.S.R. can clearly con-tribute substantially to an improved monitor-ing capability. It is difficult to reach precisebut general conclusions on what future yieldlevels could be monitored, because much willdepend on the degree of effort put into newseismic networks and data analysis. Conclu-sions would depend too on judgments aboutthe level of effort that might be put into clan-destine nuclear testing. The discussion of eva-sion scenarios is taken up in the next chapter,and high-frequency seismic data are clearlyuseful for defeating some attempts to hidesmall nuclear explosions.

IOFor a more complete discussion of the use of high-frequencydata for detection, see ch. 4, “Detecting Seismic Events. ”

IDENTIFICATION CAPABILITY

By applying one or more of the methods dis- set of different identification methods, takencussed above, many seismic events can be iden- as a whole and applied in a systematic fash-tified with high confidence. Note, however, that ion, that must be assessed when giving sum-no one method is completely reliable. It is the maries on capability.

91

Identification Capability Within theU.S.S.R. Using No Internal Stations

The identification capability of any networkof seismic stations will always be poorer thanthe detection capability for that network. Ingeneral, a rough estimate of the 90 percentidentification threshold will be about 0.5 mb

above the 90 percent detection threshold formagnitudes above mb 3.5. Below a magnitudeof about 3.5 mb, this difference may be larerthan 0.5 mb.

In the previous chapter, a network consist-ing of a dozen or so arrays all outside the So-viet Union was discussed as a type exampleto convey a sense of what capabilities can beachieved. A cautious calculation of the detec-tion capability for such a network was that itwould have a 90 percent probability of detect-ing at four or more stations all seismic eventswithin the Soviet Union with a magnitude(rob) of 3.5 or greater.” Therefore, for seismicevents in the U. S. S. R., a cautious estimate ofthe identification capability of such a networkexternal to the Soviet Union is that the 90 iden-tification threshold will be mb 4.0. From ta-ble 5-1, it can be seen that approximately 180earthquakes occur at magnitude 4.0 or aboveeach year on Soviet land areas. This wouldmean that approximately 18 of these earth-quakes would not be identified with high con-fidence by routine seismic means alone. This,however, does not translate into an opportu-nity to cheat, because from the cheater’s per-spective there is only a one-in-ten chance thatany given event above this magnitude will notbe identified through seismic means.

Events larger than this will be identified seis-mically with even higher confidence. A largerpercentage of the events that are smaller willnot be identified, and the number of events willalso increase with decreasing seismic mag-nitude.

“See ch. 4, “Detecting Seismic Events. ”

Identification Capability Within theU.S.S.R. Using Internal Stations

With a number of internal stations, it isclearly possible to attain a much improved de-tection and location capability within the So-viet Union. From estimates described in theprevious chapter, there is consensus that de-tection capability (90 percent probability of de-tection at four or more stations) of mb 2.() -2.5can be realized, although the debate is not yetresolved on the number of internal stationsthat would be required, nor on whether thevalue is closer to mb 2.0 or mb 2.5.

A detection capability down to somewherein the range mb 2.0-2.5, however, cannot beeasily translated into a statement about iden-tification capability. Estimation of the capa-bility of hypothetical networks, using regionalseismic waves, is difficult because data on noiselevels and signal propagation efficiency are notusually available and assumptions must bemade that turn out to have a strong influenceon conclusions. Although there is now generalagreement on the detection capability ofhypothetical networks, there are some signif-icant differences in opinion on the identifica-tion capability such networks provide.

Discussion of identification capability forsmall events (mb in the range 2.5- 4.0), usinginternal stations, is one of the main areas oftechnical debate in seismic monitoring. Inter-nal stations will significantly improve the ca-pability to identify many small earthquakesas earthquakes, because such stations extendthe discriminants related to the MS:mb

method down to lower magnitudes. (Internalstations will also permit more accurate deter-mination of epicenters and source depths, thussupplying for small events the most basic in-formation upon which most sources are iden-tified.) But certain types of shallow earth-quakes, below some observable magnitudelevel, are recognized as difficult if not impos-sible to identify. Also, preliminary use of highfrequencies in the United States, to dis-criminate the explosion signals of small nucleartests in Nevada from signals of small earth-quakes in the western United States, has had

92

some discouraging results. While some saythat this is indicative merely of difficultiesassociated with high frequency seismic wavepropagation in the western United States (andthat high frequencies propagate with highersignal-to-noise ratios in the U. S. S. R.) or thatthe instrumentation used in the study pre-vented data of sufficiently high quality frombeing brought to bear on the problem, othersclaim that these preliminary results are validand are indicative of a fundamental lack of ca-pability in the seismic method, at very smallmagnitudes.

In the absence of access to extensive Sovietdata (particularly, explosion data), some guid-ance in what might be possible is given bydetailed studies of experience with U.S. explo-sions. Here, it is recognized that use of morethan one discriminant results in improvementover use of the single best discriminant (whichis usually Ms:mb). In one multi-discriminantexperiment for U.S. explosions and earthquakeswhich drew heavily on data such as that pre-sented in the last diagram of figure 5-5 (inwhich mb for some earthquakes is less than3.0), seismic discrimination was achieved forall events that had both mb and MS values.12 13

The discriminants tested were Ms:mb, com-bined with a list that included relative excita-tion of short-period SH waves; relative signalamplitudes for Pn, Pg, Lg, and the largest partof the P wave; generation of higher mode sur-face waves; long-period surface wave energydensity; relative amplitudes of crustal Loveand Rayleigh waves; excitation of Sn; spectralratios of Pn, Pg and L ; various other spectral

methods; and a dept discriminant (see box5-B). Not included, was what in practice in theU.S.S.R. would be a key but not definitive dis-criminant, namely an interpretation of the epi-center location.

From this body of experience, there appearsto be agreement that, with internal stations

‘*M.D. Denny, S.R. Taylor, and E.S. Vergino, “Investigationsof m~ and MS Formulas for the Western U.S. and Their Impacton the m~:M, Discriminant, ” UCRL-95103 LLNL, August1986.

19R.E. Glaser, S.R. Taylor, M.D. Denny, et.al., “Regional Dis-crirninants of NTS Explosions and Western U.S. Earthquakes;Multivariate Discriminants, ” UCID-20930 LLNL, November1986.

Box 5-B.—Progress in Seismic Monitoring

Progress in seismic monitoring has beencharacterized by research results that, whenfirst offered, seemed optimistic but which inseveral key areas have withstood detailedsubsequent study and thus have become ac-cepted. Occasionally there have been setbacks,as noted elsewhere in describing explosion-induced S-waves, and signal complexity.

The current situation is still one of activeresearch, in that spectral discriminants havebeen proposed that (if corroborated by em-pirical data, which is now lacking) would per-mit monitoring down to fractions of a kilo-ton. A key problem in estimating what futurecapability is possible is that current researchsuggestions often entail data analysis signif-icantly more sophisticated than that requiredfor conventional discriminants. Databasemanagement would also have to be improved.The requirement, for operational purposes,that a discriminant be simple to apply, is thusin conflict with the requirement that the max-imum amount of information be extractedfrom seismic signals.

that detect down tomb 2.0-2.5, identificationcan be accomplished in the U.S.S.R. down toat least as low as mb 3.5. This cautious iden-tification threshold is currently set by the un-certainty associated with identifying routinechemical explosions that occur below this level.Many experts claim that this identificationthreshold is too cautious and that with an in-ternal network, identification could be donewith high confidence down to mb 3.0. Atpresent, however, this has not been acceptedas a consensus view, partly because some ex-perts are on principle unwilling to extrapolatefrom experience with limited U.S. data to ahypothetical situation that relies on internalstations in the U.S.S.R. On the other hand, ad-vocates of high-frequency monitoring maintainthat identification can be routine at thresholdswell below mb 3.0. The acceptance of an iden-tification capability below mb 3.0, however,would probably require practical experiencewith data from a monitoring network through-out the Soviet Union.

Chapter 6

Methods of Evadinga Monitoring Network

CONTENTS

PageIntroduction. , . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95Evasion Schemes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95

Testing Behind the Sun or in Deep Space . . . . . . . . . . . ., . . . . . . . . . . . . . 96Simulating an Earthquake . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96Testing During an Earthquake . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96Testing in a Large Underground Cavity-Decoupling . . . . . . . . . . . . . . . . 97Testing in a Nonspherical Cavity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97Testing in Low-Coupling Materials. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97Masking a Test With a Large Chemical Explosion. . . . . . . . . . . . . . . . . . . 97

Physics of Cavity Decoupling . . . . . . . . . . . . . . . . . . . ., . . . . . . . . . . . . . . . . . 98Efficiency of Cavity Decoupling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98

Limits on Cavity Construction in Salt Deposits . . . . . . . . . . . . . . . . . . . . . 98Limits on Cavity Construction in Hard Rock . . . . . . . . . . . . . . . . . . . . . . . 99Cavity Size Requirement for Decoupling . . . . . . . . . . . . . . , . ..........,100

Constraints on Decoupling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ........101Decoupling Factors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. ...101Low Frequencies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ......101High Frequencies. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. ....102

Decoupling Opportunities in the Soviet Union . . . . . . . . . . . . . . . . . . . . ....103Partial Decoupling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ., ..., ...105Monitoring Capabilities Considering Evasion . . . . . . . . . . . . . . . . . . . .. ....106

Monitoring Capability Within The U.S.S.R. Using No InternalStations. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .........107

Monitoring Capabilities Within The U.S.S.R. With Internal Stations ...108

FiguresFigure No. Page

6-l. Minimum Cavity Size Required To Decouple a 5 kt NuclearExplosion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ..............100

6-2. Decoupling Factor as a Function of Frequency. . ..................1026-3. Salt Deposits in the Soviet Union . . . . . . . . . . . . . . . . . . . . ., . ........1046-4. Partial Decoupling Effect . . . . . . . . . . . . . . . . . . . . . . . . . . . . ..........105

TableTable No. Page6-1. Numbers of Decoupled Explosions of Yield Greater Than 1 ktThat

May Be Possible in Cavities Created by Contained U.S.S.R.Underground Explosions, 1961-86 . . . . . . . . . . . . . . . . . . . . . . . . . .. ....103

Chapter 6

Methods of Evading a Monitoring Network

Seismic monitoring when combined with treaty constraints and othermonitoring methods must demonstrate a capability to defeat any plausible

scenario for evading the monitoring network

INTRODUCTION

The previous chapters have discussed the ca-pability of various networks to detect and iden-tify seismic events. From this discussion it isclear that well-coupled nuclear explosions withinthe Soviet Union could be detected and iden-tified with high confidence down to yields well-below 1 kiloton using a high-quality seismicnetwork. Yet, in deciding what limits on un-derground nuclear testing could be verified,further considerations are necessary. A coun-try attempting to conduct a clandestine testwould presumably use every practical meansto evade the monitoring network by reducing,masking, and disguising the seismic signal cre-ated by the explosion. Consequently, detectionand identification thresholds cannot be directlytranslated into monitoring capabilities with-out considering the various possibilities forevasion.

As we will see in this chapter, certain eva-sion scenarios could create serious problemsfor a seismic monitoring system under certainconditions. The need to demonstrate that theseevasion scenarios can be defeated (i.e., the ex-plosions in question identified) with high con-fidence is what limits our monitoring capabil-ity. The problem of evasion must be dealt withby a combination of seismic methods, treatyconstraints, and other monitoring methodsthat reduce the difficulties and uncertaintiesof applying seismic monitoring methods toevery conceivable test situation. In short, seis-mic monitoring needs some help and the obvi-ous approach is to require the structuring ofany treaty or agreement to create a testingenvironment that makes it much more likelythat a combination of prohibitions, inspections,and seismic methods will provide the desiredhigh levels of verification capability.

EVASION SCHEMESOver the past three decades, researchers

have conceived a number of theoretical sce-narios by which a low-threshold test ban treatymight be evaded. These include: testing behindthe sun, testing in deep space, detonating aseries of explosions to simulate an earthquake,testing during or soon after an earthquake,testing in large underground cavities, testingin nonspherical cavities, testing in low-couplingmaterial such as deposits of dry alluvium, andmasking a test with a large, legitimate indus-trial explosion. While some of these scenarioswarrant genuine attention from a monitoringperspective, others can be dismissed because

of extreme difficulty of execution or even in-feasibility. To determine which are which,standards of credibility need to be applied.

For an evasion scenario to be credible it mustbe technically feasible and it must create aworthwhile advantage for the country consid-ering cheating. As discussed in chapter 2, acountry considering cheating would have toevaluate the risks and costs of being caughtagainst the benefits if not caught. The coun-try concerned about preventing cheating hasto guess the other country’s values for mak-ing such decisions. While a slight probability

95

of detection might be sufficient to deter cheat-ing, a much more stringent standard is usu-ally needed to achieve high confidence that anycheating would be detected. Thus, the degreeof confidence needed to satisfy the concernedcountry is often higher than what is neededin practice to prevent cheating.

Although the majority of proposed evasionscenarios have been shown to be readily defeatedby a good seismic monitoring network, a fewconcepts have evoked serious concern and anal-ysis on the part of seismologists and other sci-entists. The remainder of this first section pro-tides a brief listing of the various evasionscenarios. The following three sections discussin detail evasion scenarios involving cavitydecoupling and how opportunities for decouplingcould be reduced. The final section assessesthe extent to which the most threatening eva-sion scenarios limit our capability to monitorseismically underground nuclear explosions.

Testing Behind the Sunor in Deep Space

It has been suggested that the Soviet Unioncould cheat on all test limitation treaties sim-ply by testing in deep space or behind the sun.The idea is that one or two space vehicles wouldgo behind the sun or into deep space. A nu-clear device would be detonated and an instru-ment package would record the testing infor-mation and at a later time transmit the databack to Earth. Some feel that such a testingscenario is both technically and economicallyfeasible. Others feel that the technical sophis-tication, risk, and uncertainty of such a testexceeds the utility of any information thatcould be obtained in such a manner. Such atest would bean unambiguous violation of sev-eral treaties, and hence, discovery would becostly to the tester. In any case, if clandestinetesting behind the sun or in deep space is dem-onstrated on technical grounds to be a concern,the risk could addressed, albeit at considerableexpense, by deploying satellites to orbit aroundthe sun. Such satellites equipped with detec-tors for thermal photons could monitor explo-sions in deep space. Alternatively, the risk

could be addressed politically by negotiatingan agreement to conduct simple inspectionsof the rare vehicles that go into deep space.

Simulating an Earthquake

It has been suggested that a series of nu-clear explosions could be sequentially deto-nated over the period of a few seconds to mimicthe seismic signal created by a naturally occur-ring earthquake. The purpose of such a sequen-tial detonation would be to create a P-waveamplitude that would indicate an earthquakewhen using the Ms:mb discrhninant.1 This eva-sion method has been dismissed, however, be-cause it only works if the P-wave amplitudeis measured over just one cycle. If the P-waveamplitude is measured over several cycles, theM s: mb discrimin ant will indicate an explosion.Furthermore, the sequence of waves simulatedby the explosion will only appear as an earth-quake over a particular distance range. Con-sequently, a well-distributed network thatrecords over a variety of ranges would not befooled. In addition, such an evasion attemptwould create large seismic signals that otherdiscriminants might recognize as being createdby an explosion.

Testing During an Earthquake

The hide-in-earthquake scenario posits thata small explosive test can be conducted with-out detection by detonating it shortly after anearby naturally occurring earthquake. If theearthquake is sufficiently large and the explo-sion is properly timed, the seismic signal ofthe explosion will be partially or completelyhidden by the larger seismic signal of the earth-quake. This evasion method was at one timeconsidered a challenge to seismic monitoringeven though the technical difficulties associ-ated with the execution have long been knownto be great. For example, seismologists cur-rently have no reliable techniques for the short-term prediction of the time, location, and sizeof earthquakes, and this limitation is unlikelyto be overcome in the near future.

‘A discussion of M,:m~ and other discriminants is presentedin chapter 5, “Identifying Seismic Events. ”

97

Recent developments in seismic instrumen-tation and data handling have further reducedthe feasibility of this evasion scenario. Newseismic instrumentation is now capable of fil-tering so as to pass only high-frequency seis-mic waves. Because nuclear explosions producehigher frequency seismic waves than earth-quakes, it is often possible to remove the ef-fects of distant earthquakes and see the wavescreated by the explosion. For this reason andthe difficulty of detonating an explosion at theright location and time, the hide-in-earthquakescenario is no longer considered a credible eva-sion threat. However, because high-frequencyseismic waves may not always be detectableat great distances, it maybe necessary to haveseismic stations within the Soviet Union to ob-viate the hide-in-earthquake scenario at yieldlimits as small as a few kilotons.

Testing in a Large UndergroundCavity—Decoupling

If a nuclear explosion is set off in a suffi-ciently large underground cavity, it will emitseismic waves that are much smaller thanthose from the same size explosion detonatedin a conventional underground test. This scheme,called cavity decoupling, has been experimen-tally verified at small yields. It is the consensusof geologists that significant opportunities ex-ist within the Soviet Union to construct un-derground cavities suitable for decoupling lowyield explosions. Furthermore, it is the con-sensus of seismologists that seismic waves canbe muffled by this technique. Consequently,the technical capability to conduct clandestinedecoupled nuclear tests determines the yieldthreshold below which treaty verification byseismic means alone is no longer possible withhigh confidence. The later sections of this chap-ter discuss cavity decoupling scenarios in detail.

Testing in a Nonspherical Cavity

This evasion scenario suggests that the det-onation of an explosion in an nonsphericalcavity could be used to focus the resulting seis-mic waves away from monitoring stations.This evasion scenario has been dismissed for

two reasons. First, a nonspherical cavity wouldhave no better and perhaps worse decouplingthan a spherical cavity of the same volume.2

Second, a monitoring network would have seis-mic stations in many directions, not just one.The presence of such stations would increasethe risk of detection by at least one station,possibly at an enhanced level.

Testing in Low-Coupling Materials

As discussed in the previous chapters, theproportion of explosive energy converted intoseismic waves depends on the type of rock inwhich the explosion occurs. Low-coupling ma-terials such as dry porous alluvium have air-filled pore spaces that absorb much of the ex-plosive energy. This has led to the concern thata monitoring network could be evaded by det-onating an explosion in low-coupling material.

The opportunities for such evasion are thoughtto be limited in the Soviet Union because nogreat thicknesses of dry alluvium are knownto exist there. In fact, large areas of the So-viet Union are covered with permafrost thatwould produce well-coupled seismic signals.Estimates of the maximum thickness of allu-vium in the Soviet Union indicate that it wouldonly be sufficient to muffle explosions up to1 or 2 kt. Even if such an opportunity doesexist, alluvium is a risky medium for testingbecause it is easily disturbed. An explosion inalluvium could create a subsidence crater orother surface expression. Consequently, clan-destine testing in low-coupling material is con-sidered feasible only for explosions below 1 or2 kt.

Masking a Test With a LargeChemical Explosion

As discussed in chapter 5, chemical explo-sions are used routinely in the mining and con-struction industries. In monitoring a low-yieldor comprehensive test ban treaty, there wouldbe concern that large chemical explosions could

2L.A. Glenn and J.A. Rial, “Blast-Wave Effects on DecouplingWith Axis-Symmetric Cavities, ” Geophysical Journal of theRoyal Astronomical Society, October 1987, pp. 229-239.

98

be used to mask the signals from a nuclear test. decoupled in a large underground cavity andUnlike an earthquake, such explosions could the reduced seismic signal either masked withbe timed to coincide with a clandestine nuclear the simultaneous detonation of a very largetest. If done in combination with cavity chemical explosion or attributed to a chemi-decoupling, this evasion scenario would be a cal explosion. The combination of decouplingchallenge to a monitoring network. For exam- and masking is discussed further in the de-ple, a nuclear explosion of a few kt could be tailed sections on decoupling scenarios.

PHYSICS OF CAVITY DECOUPLINGAn underground nuclear explosion creates

seismic waves with a broad range of frequen-cies. For purposes of seismic detection andidentification of small events, frequencies fromroughly 1 Hz to perhaps as high as 30 or 50Hz may be important.

For the lower end of this frequency range,the amplitude or size of the seismic waves cre-ated by an explosion is approximately propor-tional to the total amount of new cavity vol-ume created by the explosion. A conventional,or tamped, test is detonated in a hole whoseinitial volume is negligible compared to itspost-test volume. Because the initial hole issmall, the rock surrounding the explosion isdriven beyond its elastic limit by the explo-sion and flows plastically. This flow results inlarge displacements of the surrounding rockmass, and therefore leads to a large cavity-volume increase around the explosion, and ef-ficient generation of seismic waves.

If, on the other hand, the explosion occursin a hole of much greater initial volume, theexplosive stresses at the cavity wall will besmaller. This results in less flow of the rock,hence less cavity expansion and reduced coup-ling to seismic waves. If the initial hole is suffi-ciently large that the stresses in the surround-ing rock never exceed the elastic limit, theseismic couplings minimized. Further increaseof the emplacement hole size will not furtherreduce coupling at low frequencies, and the ex-plosion is said to be fullly decoupled.

Cavity construction on a scale required forexplosion decoupling is possible in either saltof sufficient thickness or in hard rock such asgranite. In either case, the cavity volume re-quired for full decoupling increases in propor-tion to the explosion yield and decreases as thestrength of the rock increases.

EFFICIENCY OF CAVITY DECOUPLING

Limits on Cavity Constructionin Salt Deposits

Large cavities suitable for decoupled nucleartesting above 1 kt can be constructed in saltdeposits either by detonating a nuclear explo-sion of several tens of kilotons, or by solutionmining. For example, a stable, free-standingcavity was created by the U.S. “Salmon” test,a 5.3 kt explosion in a salt dome.3 This cavity

‘U.S. Department of Energy, Office of Public Affairs, NevadaOperations Office, Anncmnced United States Nuclear Tests, July1945 Through December 1983, NVO-209 (rev.4), January 1984.

was sufficiently large to decouple the subse-quent 0.38 kt “Sterling” nuclear test, whichwas detonated in the Salmon explosion cavity.4

Nuclear explosions create cavity volume ap-proximately in proportion to their yield. Thus,applying the yield ratio given in the Salmon/Sterling experiment, an explosion greater than14 kt would be required to create a cavity suffi-cient to fully decouple a 1 kt test; similarly,an explosion greater than 140 kt would be re-quired to create a cavity sufficiently large to

41bid.

99

fully decouple a 10 kt test. Explosive construc-tion of cavities adequate to decouple shotsabove 1 kt would obviously be impossible toaccomplish clandestinely. Past nuclear testsin the Soviet Union have produced many cavi-ties suitable for decoupling, but the locationand approximate size of most of these is known;and thus evasion opportunities at these sitescould be limited if activity at them is mon-itored.

Solution mining on the required scale wouldalso be difficult to conceal. For example, withpresent techniques it would take many monthsor perhaps even a year of continuous opera-tion at high circulation rates to solution minea cavity adequate to fully decouple a 5 kt ex-plosion. The technology also requires enormousamounts of water and the disposal of enormousamounts of brine, further hindering conceal-ment. However, given that such an operationwould be detected and monitored, it might bedifficult to distinguish legitimate and evasion-related activity. Concealment of the site wouldnot be necessary if appropriate activity (min-ing of salt) exists. Consequently, areas of saltdeposits might require treaty provisions deal-ing with chemical explosions with magnitudescomparable to decoupled nuclear explosions.

Apart from the significant problems of con-cealment and resource application, there do notappear to be constraints preventing the con-struction, by solution mining, of cavities largeenough to fully decouple explosions up to 10kt. In fact, the Soviet literature reports solu-tion-mined cavities with volumes up to one mil-lion cubic meters. If such a cavity could be con-structed with a spherical shape, it would be60 meters in radius. A spherical cavity witha 60 meter radius would have sufficient vol-ume to decouple an explosion up to 14 kt, basedon cube root scaling of the U.S. Salmon/Ster-ling salt dome decoupling experiment. However,these existing large, solution-mined cavitiesare not spherical. They are highly elongated,irregular in shape, and filled with brine. Thesefeatures reduce the size of the explosion thatcould be decoupled in the cavity. Furthermore,the brine in the cavity supports through itsown hydrostatic pressure a considerable por-

tion of the overburden (i.e. the weight of theoverlying rocks). If the cavity was empty, theoverburden pressure would not be supportedand the stability of the cavity would be un-certain.

Both salt domes and bedded salt regionshave to be considered as candidate locationsfor construction of decoupling cavities in salt,although the mining procedure would be morecomplex in bedded salt deposits. To create acavity in bedded salt, solution mining of solu-ble layers would have to be combined with ex-plosive mining of insoluble interbeds.

The creep strength of natural rock salt con-trols the maximum depth at which a stablecavity can be maintained. Cavity collapse ormajor changes in cavity shape occur over timescales of a few months when the overburdenpressure at cavity depth exceeds the internalpressure in the cavity by more than about 20MPa (200 times atmospheric pressure). Thiscorresponds to a maximum depth of about 1km for a stable, empty cavity. A brine-filledor gas-pressurized cavity might be stable toabout 2 km depth. If a cavity is made by anexplosion or by solution mining, the salt willbe weakened. This will be a consideration be-cause for weak salt a larger cavity is needed,than predicted for strong salt, to fully decou-ple a given explosion.

Limits on Cavity Constructionin Hard Rock

No cavities have been constructed in hardrock on the scale of those known in salt. Thereis agreement among verification experts thatdecoupling cavities with radii up to about 25meters, suitable for repeated testing up toabout 1 or 2 kt, can probably be constructedwith existing technology. Repeated testingcould likely be detected well enough to get goodlocations; and the detection of repeated eventsat the same location would be suspicious. Thereappear to be no known technological limita-tions preventing construction of cavities upto perhaps 45 meters radius, suitable for de-coupling explosions up to about 10 kt. How-

ever, the long-term stability under repeated ex-plosive loading is questionable.

In constructing a cavity of radius larger thanabout 25 meters, a very extensive network oflong cables would be needed to strengthen andpre-stress a large region of the rock surround-ing the cavity. Such construction would requirean elaborate network of additional tunnels andshafts in the surrounding rock. The technol-ogy is untested on this scale and constructionmay be severely complicated in many areas bythe presence of high compressive stresses inthe rock and by joints, fractures, and other rockinhomogeneities that are present in even themost uniform granites.

Concealment of such a massive excavationoperation from satellite reconnaissance orother National Technical Means would be ex-tremely difficult, and thus some plausible coveroperation would probably be necessary. A po-tential evader would also have to consider thepossible leakage along joints of radioactiveproducts such as bomb-produced noble gases.Finally, the evader would also have to be con-cerned that explosions might result in theunexpected collapse of the cavity and the for-mation of a crater on the surface. Such a pos-sibility is not without precedent: in the 1984“Midas Myth” test in Nevada, 14 people werehurt and one man killed during the unexpectedformation of a crater above a tiny collapsedcavern at a depth of 1,400 feet.

Cavity Size Requirement forDecoupling

The minimum cavity radius required for fulldecoupling is proportional to the cube root ofthe explosion yield and inversely proportionalto the cube root of the maximum pressurewhich the overlying rock can sustain withoutblowing out or collapsing. In salt, the maxi-mum sustainable cavity pressure increases ap-proximately in proportion to depth. Therefore,the minimum cavity size for decoupling is in-versely proportional to the cube root of depth(i.e. smaller cavities will work at deeper depths).

As noted above, however, there is a limit tohow deep a cavity can be maintained in salt,due to salt’s low creep strength. Thus, thereare two separate issues regarding the depthof cavities: 1) the deeper the cavity, the smallerthe size required to decouple an explosion ofa given yield, and 2) the deeper the cavity, thelower the strength of the salt and the more dif-ficult it is to maintain an open cavity. The lowstrength of salt eventually limits the depth atwhich a cavity can be created. Even the smallesthole below 1-2 km will squeeze shut.

As discussed earlier, the limiting depth forstability of a large, empty cavity in salt isabout 1 km. This depth implies that the Salmon/Sterling cavity (at 0.82 km depth), was verynear the maximum depth for stability of anempty cavity. Consequently, the Salmon cavitysize approximately sets the lower bound on

Figure 6-1 .—Minimum Cavity Size Required ToDecouple a 5 kt Nuclear Explosion

To fully decouple a 5 kt explosion in salt, a spherical cavitywith a radius of at least 43 meters would be required. Theheight of the Statue of Liberty with pedestal (240 ft) is 85°/0of the required diameter (282 ft).

SOURCE: Office of Technology Assessment, 1988

101

cavity size for full decoupling (at the yield ofthe Sterling test). This implies a minimumcavity radius of at least 25 meters for fulldecoupling of a 1 kt test in salt. Cavity require-ments are expected to scale as the cube rootof the explosion yield. For example, to fullydecouple a 5 kt explosion in salt, a sphericalcavity with a radius of at least 43 meters (25times the cube root of 5) would be required (fig-ure 6-l).

In granite, a smaller cavity, perhaps around20 meters in radius, might be expected to suc-cessfully decouple a 1 kt explosion, while a 34

meter cavity would be needed to decouple 5kt. This number is a rough estimate becauseit does not take into account the joints andfractures that would be present in even themost uniform granites. The difference betweenthe salt and granite estimates is due to thegreater strength of granite, which might there-fore sustain a somewhat higher pressure. Suchestimates, however, remain uncertain andsome experts doubt that the effective strengthof granites would be greater than salt and be-lieve that a radius comparable to that of saltwould be needed to decouple the same size ex-plosion in granite.

CONSTRAINTS ON DECOUPLINGDecoupling Factors

The reduction of seismic wave amplitudesachievable by full decoupling is called thedecoupling factor. On theoretical grounds, thedecoupling factor is expected to be smaller athigh frequencies than at low frequencies.5 Thisexpectation has been confirmed experimentally.The transition from low-frequency decouplingto high-frequency decoupling occurs over arange of frequencies rather than abruptly, andthe transition frequency range depends onyield. For a 1 kt explosion, seismic waves ofabout 6 Hz and below can be assumed to becontrolled by the full low-frequency decouplingfactor, whereas seismic waves above 6 Hz willexhibit much less decoupling.

Low Frequencies

Several decoupling experiments have beencarried out by the United States. Taken to-gether, these experiments permit us to esti-mate the low-frequency decoupling factor withconsiderable confidence. In the 1966 Salmon/Sterling experiment, a smaller nuclear explo-sion was detonated in the cavity created by

‘Donald B. Larson (cd.), Lawrence Livermore National Lab-oratory, Proc&”ngs of the Department of Energy SponsoredCavity Decoupling Workshop, Pajaro Dunes, CaZiforn;a, July29-31, 1985.

a larger explosion. Analysis of the seismicwaves from these events led to the conclusionthat the low-frequency decoupling factor is ap-proximately 70. That is, a fully decoupled ex-plosion in salt has its low-frequency seismicamplitude reduced by a factor of 70 comparedto a “tamped,” or “well-coupled,” explosionof the same yield in salt. The 1985 DiamondBeech/Mill Yard experiment compared decou-pled and tamped nuclear explosions in tuff. Inthis case, the observed decoupling factor wasagain 70. The 1959 Cowboy series of tests indome salt used conventional explosives insteadof nuclear explosives. While initial estimatesof the decoupling factor from Cowboy rangedfrom 100 to 150, it was subsequently deter-mined that conventional explosives are signif-icantly less efficient when detonated in a largecavity than when detonated under tamped con-ditions. When a correction was made for thiseffect, the Cowboy data yielded an estimateof the full low-frequency decoupling factor ofapproximately 70, in close agreement with theresults obtained in the nuclear experiments.

Earlier theoretical estimates that the low-frequency decoupling factor could be as highas 200 or greater were based on several sim-plifying assumptions. Seismologists are now inagreement that the experimentally determineddecoupling factor of 70 is appropriate at low fre-quencies.

.

102

High Frequencies

Roughly speaking, if two explosions excitelow-frequency seismic waves whose amplitudesdiffer by a factor F, their high-frequency seis-mic waves are expected to have amplitudeswhose ratio is approximately the cube root ofF. Thus, seismic theory predicts that the de-coupling factor will be much reduced at highfrequencies. The Salmon/Sterling experimentaldata corroborate this prediction. Figure 6-2shows the decoupling factor as a function offrequency inferred from the Salmon/Sterlingexperiment, with both explosions scaled to 1kt. As already discussed, the low-frequencydecoupling factor averages about 70. However,the experimentally observed decoupling fac-tor begins to drop at about 6 to 8 Hz, and thedrop is quite sharp above about 10 Hz. At 20Hz, the decoupling factor is down to approxi-mately 7. This result can reasonably be extrap-olated to estimate decoupling for other yieldsby scaling the frequency axis in figure 6-2 bythe inverse of the cube root of yield. For ex-ample, a 5 kt explosion would be expected tobe decoupled by a factor of approximately 7at a frequency of about 12 Hz (20 divided bythe cube root of 5). These scaling considera-tions provide an additional argument in favorof using high-frequency recordings to extendmonitoring capabilities down to lower yieldlevels.

At this time, the exact value of the high-frequency decoupling factor is considered lesscertain than the low-frequency factor becausethe instruments recording the Salmon explo-sion lacked sufficient dynamic range to pro-vide reliable data above 20 Hz. High-frequencydata from the Diamond Beech/Mill Yard ex-periment in tuff also show high-frequencydecoupling factors less than 10, consistent withthe Salmon/Sterling experience in salt. How-ever, interpretation of the Diamond Beech/MillYard data in terms of decoupling is compli-cated by the facts that the decoupled eventwas a factor of 100 smaller than the tampedevent, the decoupling cavity was hemispheri-cal, the measurements were made at short dis-

tances, and the events were not co-located.Data from a better experiment could reducethe uncertainty in the high-frequency decouplingfactor.

The evidence for reduced decoupling at highfrequencies comes from experiments with spher-ical (or hemispherical) cavities. However, theo-retical calculations show that this conclusionis not altered even when highly elongated

Figure 6-2.—Decoupling Factorof Frequency

103

102

10’

10°

as a Function

10° 10’ 102

Frequency, Hz

--- Theoretical _ Average Salmon/Sterling spectralratio scaled to 1 kt

● . ● ● 95% Confidence interval

The experimentally observed decoupling factor decreases athigher frequencies.

SOURCE: Modified from J.R. Murphy, Summary of presentation at the ARPADecoupling Conference, Feb. 7, 1979.

103

cavity geometry is considered, that is, it does ated, elongated decoupling cavity may enhancenot appear to be possible to increase the high- high-frequency decoupling in certain preferredfrequency decoupling factor by constructing directions, but will decrease high-frequencyspecially shaped, air-filled cavities.6 An evacu- decoupling in other directions.

6Glenn and Rial, op. cit., footnote 2.

DECOUPLING OPPORTUNITIES IN THE SOVIET UNION

Cavity construction for low-yield decouplingis possible in salt domes, bedded salt, and dryhard rock. These geologic categories excludefew areas of the Soviet Union. However, it isgenerally agreed that salt domes provide themost suitable host rock for large, stable cavi-ties. Salt domes are the most suitable becauseof the homogeneity of rock salt in domes, therelative simplicity compared to hard rock ofconstructing stable cavities explosively or bysolution mining, and the fact that numerouslarge cavities already exist in salt domes in theSoviet Union. Cavities confined to a single,homogeneous salt layer in bedded salt, on theother hand, are limited in size by the layerthickness. Assuming that the radius of a cavityin bedded salt should not exceed one-half thelayer thickness, decoupling opportunities areprobably limited to 1 or 2 kt in bedded salt.

Vast regions of the Soviet Union are under-lain by salt deposits. The general distributionof these deposits is indicated by the map infigure 6-3. However, we probably do not knowthe full extent of Soviet salt deposits. Further-more, although figure 6-3 indicates those areaswhere salt domes are prevalent, without ac-cess to detailed subsurface geologic data it isnot possible to rule out the presence of domesin any area of bedded salt in the Soviet Union.

Because the construction of large salt domecavities may be difficult to conceal, it is use-ful to estimate the decoupling opportunity pro-vided by already existing cavities created bySoviet underground explosions (presumedtamped). Table 6-1 summarizes this informa-tion. At each yield level, the table shows thenumber of existing holes large enough for fulldecoupling. These numbers refer to cavitiespresumed to have remained open following the

largest known Soviet salt dome explosions.The yields in table 6-1 were estimated by divid-ing the seismically estimated yields of thelargest Soviet salt dome explosions by theSalmon/Sterling yield ratio (5.3/0.38 = 14). Theuse of this ratio is justifiable because the cavityvolume created by a tamped explosion (in thiscase 5.3 kt, Salmon) is proportional to yieldand the largest fully decoupled explosion (inthis case 0.38 kt, Sterling) is proportional tocavity volume.

Table 6-1 indicates that Soviet decouplingopportunities at 1 kt and above, using exist-ing explosion-generated cavities, are limitedto three regions: the North Caspian region, theEast Siberian Basin, and a single site in Cen-tral Asia. On the basis of table 6-1, there are

Table 6-1.— Numbers of Decoupled Explosions of YieldGreater Than 1 kt That May Be Possible in Cavities

Created by Contained U.S.S.R. UndergroundExplosions, 1961-86a

Yield (kt)

Areas of known salt deposits 1 2 3 4 5b

North Caspian region:Azgir . . . . . . . . . . . . . . . . . . . . . . . . . . 2 4 2 0Astrakhan c . . . . . . . . . . . . . . . . . . . . . 1 — — — OOrenburg. . . . . . . . . . . . . . . . . . . . . ..2 — — — OKarachaganak. . . . . . . . . . . . . . . . . . . — 5 — — OLake Aralsor. . . . . . . . . . . . . . . . . . . . — 1 — — OIshimbay . . . . . . . . . . . . . . . . . . . . . . . — — — 1 0

East Siberian Basin:NW of Lake Baikal . . . . . . ........3 2 — — OWithin a few x 100 km of basin . . . 1 — — 1 0

Bukhara, Central Asia (explosion usedto extinguish fire in oil well) . . . . . . — — — 1 0

Full decoupling in salt: minimum radius (meters) = 25 ● (explosive yield (kt))l/3

Full decoupling in hard rock” minimum radius (meters) = 20. (explosive yield(kt))1/3%btained from yield of known explosion at site divided by yield ratio forSalmon/Sterling = 5.3/0.38 = 14.

b(or greater)cMany cavities capable of being used for full decoupling at yields Of about O 5

kt; some could be connected.

104

. . . . . . . .. . . . . ..,”. ,.

.

.,

105

no opportunities in existing explosion-gen- mation procedure used in constructing tableerated cavities above 4 kt. It should be noted 6-1 is one that most seismologists would sup-that the potential decoupled yields estimated port, within an uncertainty of about 50 per-in table 6-1 depend critically on seismic yield cent. This uncertainty translates into 50 per-estimates made for the corresponding cavity- cent uncertainty in the decoupled yields ingenerating explosions. The seismic yield esti- table 6-1.

PARTIAL DECOUPLING

The size of an explosion that could be decou-pled is limited by the maximum size of an air-filled cavity that could reasonably be createdand remain stable. Concern has been expressed,however, that even if a nuclear device is toolarge to be fully decoupled, it could perhapsbe partially decoupled, thus reducing its seis-mic signal to some extent. It has been furthersuggested that by partially decoupling largeexplosions, a country might be able to clan-destinely test above the 150 kt threshold levelof the Threshold Test Ban Treaty. As figure6-4 illustrates, however, partial decoupling isnot straightforward.

Figure 6-4 is scaled for the case of a 1 kt ex-plosion and shows how the size of the seismicsignal is affected by partial decoupling. As thecavity size first increases, the seismic signalactually gets larger, reaching a maximum fora cavity about 2 meters in radius. The largeseismic signal is produced at first in a smallcavity because less energy goes into meltingrock and more energy is transmitted into seis-mic waves. For the case of a 1 kt explosion,the radius has to exceed about 4 meters be-fore any reduction of the seismic signal occurs,and must exceed about 6 meters to obtain re-duction by more than a factor of 2. After that,further reduction occurs rapidly. If the rela-tionship for a 1 kt explosion is extrapolatedto larger explosions, the radius (in meters) ofthe cavity to begin partial decoupling = 6 *[size of explosion (kt)]1/3. For a 10 kt explosion,the size of the cavity required to begin partialdecoupling would be 6 * 101/3 = 13 meter ra-dius; for 150 kt, the radius would have to ex-ceed 32 meters.

Conducting a partially decoupled explosionalso has many risks that the potential evaderwould have to consider, including the following:

A. Partially decoupling a nuclear explosionis uncertain. As seen in figure 6-4, the re-duction in seismic signal occurs along asteep curve. It would be difficult to pre-dict from such a steep curve how muchactual decoupling there will be for a givenexplosion. If partial decoupling is notachieved, an explosion inside a cavity

Figure 6-4.— Partial Decoupling

—

I

.

Effect

I I I I I I I I I ● 1 I I0.1 1.0 10 100

Initial cavity radius (M) for a 1 kt explosion

The effect of partial decoupling scaled for the case of a1 kt explosion,

SOURCE: Modified from R.W. Terhune, C.M. Snell, and H.C. Rodean, ‘ r EnhancedCoupling and Decoupling of Underground Nuclear Explosions, ” LLNLReport UCRL-52806, Sept. 4, 1979,

106

B

might actually produce seismic signals plosion would have to be detonated in alarger than a well-coupled explosion. cavity near the maximum possible depthPartial decoupling creates greater pres- to minimize the pressure on the cavitysure on the wall of a cavity thin- fulldecoupling. For example, a 20 kt explo-sion set off in a cavity suitable for fulldecoupling of 10 kt will result in a dou-bling of the cavity pressure compared tothat for the 10 kt shot. Partial decouplingdamages the cavity wall and this makesit more difficult to be confident that no-ble gases and other bomb-produced iso-topes will not leak out of the cavity, reachthe surface, and be detected. A 20 kt ex-

wall. Risks of deformation or collapse in-crease with both the yield and the depthof the cavity. A risk trade off would beinvolved: the desire to minimize the es-cape of bomb-produced gases leads theevader to construct a cavity as deep aspossible, whereas construction difficul-ties, the time needed for construction, andthe risk of cavity deformation or collapseall become increasing problems at greaterdepths.

MONITORING CAPABILITIES CONSIDERING EVASIONThe previous two chapters discussed the

various thresholds for the detection and iden-tification of seismic events within the SovietUnion. These thresholds, combined with thefeasibility of successfully conducting clandes-tine decoupled nuclear explosions, effectivelydetermine what levels of nuclear test restric-tions can be monitored with high confidence.

As discussed in chapters 4 and 5, well-runseismic monitoring networks can detect andidentify underground nuclear explosions withyields well-below 1 kt if no attempt is madeto evade the monitoring network. However, acountry wanting to test clandestinely abovethe allowed threshold would presumably at-tempt to reduce the size of the seismic signalcreated by the explosion. For example, a coun-try might attempt an evasion scenario wherethe explosion is secretly decoupled in a largeunderground cavity and the muffled signalsare then masked by or attributed to a largechemical explosion that is simultaneously det-onated under the guise of legitimate industrialactivity. The problem for the monitoring net-work is to demonstrate a capability to distin-guish such an evasion attempt from the back-ground of frequent earthquakes and legitimateindustrial explosions that occur at low yields.

The monitoring burden placed on the seis-mic network by various evasion scenarios canbe greatly lessened if seismology gets some

help. Countering the various evasion scenariosneeds to be approached through a combinationof seismic methods, treaty constraints, andother monitoring methods that reduce thedifficulties and uncertainties of applying seis-mic monitoring methods to every conceivabletest situation. Specifically, the structure of anytreaty or agreement should create a testingenvironment such that a combination of pro-hibitions, inspections, and seismic methodswill provide the desired high levels of verifica-tion capability. Examples of the type of treatyconstraints that have been proposed to im-prove the capability of various monitoring net-works include the following:

Limitations on Salvo-Fired Chemical Explo-sions: All large salvo-fired chemical explo-sions above a certain size (depending onthe threshold being monitored and thearea) would be limited and announced wellin advance with inspections/monitoring tobe conducted on-site by the monitoringside at their discretion.Limitations on Ripple-Fired Chemical Ex-plosions: All ripple-fired chemical explo-sions above a certain size (depending onthe threshold being monitored and thearea) would be announced in advance witha quota of on-site inspections available tothe monitoring party to be used at theirdiscretion.

107

. Limitations to One Inspected and Cali-brated Test Site: All tests would be con-ducted within the boundaries of one de-fined test area in hard rock or below thewater table. Further, several calibrationtests recorded by in-country monitoringstations would be allowed with yieldsspanning the threshold yield. Inspectionsof the test site would be allowed beforeenactment of the treaty to ensure that nolarge cavities suitable for decoupling werepresent.

● On-going Test Site Inspections: A yearlyquota of on-site inspections by the moni-toring party would be allowed at the des-ignated test site.

● Joint On-site Inspections of Sites of Possi-ble Violations: A yearly quota of on-siteinspections would be allowed at sites des-ignated by the monitoring party withprompt access by a U.S.-Soviet technicalteam.

● Country-wide Network Calibration Tests:Agreement to conduct a number of largechemical explosions, of both salvo andripple-fired types, would be allowed toevaluate signal propagation characteris-tics and detection-identification capabil-ities in particular critical areas.

If these types of testing constraints can benegotiated within a treaty, reduced thresholdscould effectively be monitored. Keeping inmind the various types of networks and nego-tiated treaty constraints that are possible, thefollowing sections give a sense of the treatythresholds that could be monitored.

Monitoring Capability Within TheU.S.S.R. Using No Internal Stations

The threshold for detecting and locating seis-mic events within the Soviet Union (90 per-cent probability of detection at four or morestations), using a seismic network with no in-ternal stations, is at least as low as mb 3.5 (ch.4). The associated threshold for the identifica-tion of 90 percent of all seismic events withinthe Soviet Union is at least as low as mb 4.0(ch. 5). An mb of 4.0 corresponds to a well-

coupled nuclear explosion with a yield of about1 kt. Consequently, clandestine nuclear explo-sions above 1 kt would need to be decoupledto evade the monitoring network. Several con-siderations limit the threshold at which suchclandestine nuclear tests might be attempted.

Holes large enough to decouple explosionsabove a few kilotons would have to be madein salt domes. Almost all of the known saltdome regions of the U.S.S.R. and regions thathave any known types and thicknesses of saltdeposits are situated in areas of low naturalseismicity and good seismic transmission. Thedetection of seismic events from such areaswould probably be better than average.

Even if an explosion were successfully decou-pled and the seismic signal muffled down be-low the 90 percent identification threshold, itmight still be identified. Decoupled explosionsproduce seismic signals that are veryexplosion-like. Because there is no breaking ofrock or tectonic release, the signals from decou-pled explosions do not look like earthquakes.This makes the identification of a detectedevent as a decoupled explosion likely. Eventhough the magnitude of the clandestine testis below the identification threshold, the testwould in many cases still be well-detected andlocated. Also, note that the identificationthreshold is for 90 percent identification, thatis, 90 percent of all events above this magni-tude will be positively identified. There is nosharp boundary between identification andnon-identification. Even if the seismic magni-tude from a specific event fell somewhat be-low the identification threshold, there is a goodchance that it would be identified. As discussedearlier, the identification threshold is largelyset by the problem of distinguishing largechemical explosions from small decoupled nu-clear explosions, so that treaty constraints forhandling large chemical explosions would bevery helpful.

The largest air-filled cavity that could rea-sonably be created and remain stable wouldfully decouple a nuclear explosion of no morethan about 10 kt. While large explosions of upto 20 or more kt could theoretically be partially

108

decoupled to produce a seismic signal belowthe cautious identification threshold, such eva-sion scenarios are considered implausible be-cause of the practical considerations of con-tainment and predicting the decoupling. Infact, evasion scenarios for explosions above 10kt are not considered credible by most experts.This means that the monitoring of the SovietUnion with only an external network can beaccomplished down to a threshold of about 10kt. However, for accurate monitoring of a 10kt treaty, all experts agree that it would bedesirable to have stations within the SovietUnion for accurate yield estimation, plus treatyrestrictions for handling the identification oflarge chemical explosions in areas where de-coupling could take place.

Monitoring Capabilities Within TheU.S.S.R. With Internal Stations

The detection threshold (90 percent probabil-ity of detecting a seismic event at four or morestations) is mb 2.() - 2.5 using a seismic net-work with internal stations (ch. 4). The associ-ated threshold for the identification of 90 per-cent of all seismic events is at least as low asmb 3.5 (ch. 5) and could be reduced dependingon what provisions are negotiated to handlechemical explosions. This identification thresh-old corresponds to a well-coupled nuclear ex-plosion with a yield below 1 kt.

Seismic stations within the Soviet Unionwould permit lower thresholds to be monitoredby reducing the opportunities for evasion.Decoupling is possible for explosions withyields below 10 kt. Consequently, the networkof internal stations would be designed primar-ily to reduce the opportunities for decoupling.The most challenging evasion scenario for sucha network would be the situation where a small(1-5 kt) nuclear explosion is decoupled and thereduced seismic signal masked by or attributedto a simultaneous detonation of a legitimateindustrial explosion. As noted above, severalconsiderations limit the threshold at whichsuch clandestine nuclear tests might be at-tempted.

Almost all of the known salt dome regionsof the U.S.S.R. and regions that have anyknown types and thicknesses of salt depositsare situated in areas of low natural seismicityand good seismic transmission. The exceptionsinclude salt deposits in the Caucasus, Tad-jikistan, and near the Chinese border. An in-ternal monitoring network should involve theplacement of more seismic stations at closerspacing in those few areas. In addition, thoseareas are all near the southern border of theU.S.S.R. where the detection and identifica-tion thresholds either are currently better orcan be made better than the average identifi-cation threshold by monitoring from nearbycountries (i.e., Turkey for Caucasus) and sta-tions inside the U.S.S.R.

Many seismologists feel that the discrimi-nation threshold of mb 3.5 is too cautious aprediction for the capability of an internal seis-mic network. This identification threshold ismostly set by the large numbers of chemicalexplosions that occur below this level. The limi-tations imposed by identifying chemical explo-sions can be approached in two ways: first,limiting them by treaty (limiting their size andrequiring on-site observers and monitoring)and second, by further developing techniquesto make use of the expected differences be-tween the signals created by distributed ripple-fired chemical explosions and the concentratedpoint explosions characterizing decoupled nu-clear tests. While chemical explosions in theU.S.S.R. of mb 3.0 are likely to be more com-mon than those of mb 3.5, the monitoring needonly be concerned with those chemical explo-sions of mb 3.0 and larger that are located inareas of known or possible salt domes. Thisexcludes very large areas of the Soviet Union.A monitoring network with stations internalto the Soviet Union should concentrate onareas of poor transmission and areas wheredecoupling opportunities would be possible.Through such a strategy and with constraintson chemical explosions, many predict that theidentification threshold will be closer to mb

3.0. This would significantly reduce the sizeof decoupled explosions that could be clandes-tinely attempted.

109

All of the considerations so far have notmade allowances for increased verification ca-pability afforded by high-frequency recording.The recent N.R.D.C. recordings to very highfrequencies at distances of 200-650 km fromthree chemical explosions with yields of 0.01-0.02 kt are very impressive in this regard.From these data, it appears that explosionswith yields comparable to a fully decoupled 2.6-3.8 kt explosion (corresponding to magnitudemb 3.0) in areas of good transmission will pro-duce large signals with frequencies of 10-20Hz. This is also a frequency band in which thedecoupling factor will be small.

The decoupling reduction that is assumedfor these evasion scenarios is a factor of 70.If the monitoring system has even a modestcapability to record frequencies as high as 10-15 Hz, the effectiveness of the decouplingwould be greatly reduced. Figure 6-2 indicatesa decoupling factor of 30 for frequencies of 1-2 Hz and 50 as averaged from 1 to 5 Hz. Athigh frequencies, the decoupling factor willprobably be reduced from 70 to below 10.Smaller decoupling factors will result in a lower(better) threshold for the detection and iden-tification of decoupled explosions of a givenyield.

Decoupling combined with masking remainsa challenging evasion scenario even with ahigh-quality internal network. Opportunitiesfor such evasion, however, would be limitedby the many practical considerations describedabove and throughout this report. Attempt-ing evasion by this complicated scenario wouldentail further risk when viewed in conjunctionwith all types of intelligence gathering, ratherthan purely as a problem for seismic discrimi-nation. Detected seismic events in areas of pos-sible decoupling would be suspicious and pre-sumably focus attention. On-site inspectionscould play an important role as opportunitiesfor cheating could be still further reduced bynegotiated agreements requiring prior announce-ment and possible on-site inspections of largechemical explosions in areas of potential de-coupling.

Small differences of opinion concerning mon-itoring capability will always remain becauseparts of the debate are comparable to discus-sions of “half-full” versus “half-empty’ glassesof water. Some will review the complex opera-tion of seismic monitoring and will concludethat a country could cheat if any step in theprocess is uncertain. The chain is only as goodas its weakest link. Others will review the com-plicated evasion scenarios that have beenpostulated and conclude that evasion is too dif-ficult and uncertain to be credible. Cautiousassumptions about seismic monitoring capa-bility and generous assumptions about thelikelihood of successfully conducting clandes-

tine decoupled nuclear explosions can be com-bined to produce the conclusion that even withan internal network an explosion of up to 10or 20 kt could be partially decoupled in thelargest hole (capable of fully decoupling a 10kt explosion) to create a seismic signal belowthe mb 3.5 identification threshold. On theother hand, generous assumptions about mon-itoring capabilities and favorable assumptionsabout uncertainty and the role of other intelli-gence gathering systems can be combined toproduce the conclusion that even explosionsof a fraction of a kiloton fully decoupled canbe effectively monitored with high confidence.Considering all of the arguments, however, afew general statements can still be made con-cerning monitoring capability with an inter-nal seismic network.

Most experts agree that a high-quality net-work of internal stations combined with strin-gent treaty constraints, could monitor a thresh-old of around 5 kt. Differences of opinion rangefrom 1 to 10 kt and are due to judgments aboutthe level of constraints that can be negotiatedinto the treaty and what levels of motivationand risk the Soviet Union would be willing totake to test clandestinely slightly above thethreshold. Experts further agree that below1-2 kt, monitoring would become much moredifficult because additional methods of evasionare possible. Explosions of 1 or 2 kt could bedecoupled not only in salt but also in other me-dia such as granite and alluvium. At present,there is not a consensus that an internal net-

110

work would be capable of positively identify- perience of low-yield monitoring within the So-ing with high confidence all such evasion at- viet Union together with a high level of nego-tempts. If such a capability is possible, it will tiated supplementary measures to limit certainrequire demonstration through practical ex- evasion opportunities.

Chapter 7

Estimating the Yields ofNuclear Explosions

CONTENTS

PageIntroduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ...........113Measuring the Size of Seismic Signals.. . ............................113Determining Explosive Yield From Seismic Magnitude . ...............115Sources of Uncertainty. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ...........,.116Bias Correction for Soviet Nuclear Explosions. . . . . . . . . . . . . . . . . . . . . ...117Reducing Uncertainties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ..............120

Random Uncertainty. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. ... ... ... ..120Systematic Uncertainty . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .........,122

Yield Estimation Capabilities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ..122Soviet Test Practices and Test Ban Compliance . . . . . . . . . . . . . . . . . . . . ..124

BoxBox Page7-A. Calculations of the Six Largest East Kazakhstan Explosions.

By Sykes et al., Based on Unclassified Data . ....................124

FiguresFigure No. Page7-l. Computation of P-Wave Magnitude . . . . . . . . . . . . . . . . . . . . . . .......1147-2. Explosions in Granite . . . . . . . . . . . . . . . . . . . . . . . . .................1157-3. Yield Data From the Nevada Test Site . . . . . . . . . . . . . . . . . . . . .. ....1167-4. Yield Estimate Distribution . . . . . . . . . . . . . . . . . . . . . ...............1167-5. Schematic Illustration of Attenuation-Related Magnitude Bias .. ....1187-6. Comparisons of Explosions in Eurasia With Explosions in North

America. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ..1197-7. Uncertainty in Yield Estimates . . . . . . . . . . . . . . . . . . . . . . . . . . .......1217-8. mb Versus Time for Large Soviet Explosions . . . . . . . . . . . . . . . . . . . . .1257-9. U.S.S.R. Nuclear Tests 1966-86 According to Sykes and Davis .. ....126

Chapter 7

Estimating the Yields of Nuclear Explosions

For treaties that limit the testing of nuclear weapons below a specificthreshold, the yield of the explosion must also be measured.

INTRODUCTION

Once a seismic event has been detected andidentified as a nuclear explosion, the next stepis to estimate the yield of the explosion. Thisis particularly important for monitoring trea-ties that limit the testing of nuclear weaponsbelow a certain threshold. The process of esti-mating the yields of Soviet explosions involvesthree steps: 1) calculate the magnitude of theseismic signal; then, 2) make corrections to ad-just for the different geology at each test site;and finally, 3) convert the magnitude into ayield estimate.

The final yield measure describes the explo-sive energy of a nuclear explosion in terms ofkilotons, where 1 kiloton (kt) was originally de-fined as the explosive power equivalent to1,000 tons of TNT. This definition was foundto be imprecise,’ however, and so it was agreedin the United States during the Manhattanproject that the term “kiloton” would refer tothe release of 1012 calories of explosive energy.

1 kiloton = 1,000,000,000,000 = 1012 calories

While this convention is also followed in theSoviet Union, it does not necessarily mean that

*The definition is not precise for two reasons. First, there issome variation in the experimental and theoretical values ofthe explosive energy released by TNT (although the majorityof values lie in the range from 900 to 1,100 calories per gram).Second, the term “kiloton” could refer to a short kiloton (2x106

pounds), a metric kiloton (2.205x106 pounds), or a long kiloton(2.24x10 6 pounds).

the United States and the Soviet Union calcu-late explosive yields in the same way. Only thatpart of the total energy released in a nuclearexplosion that is immediately available (the so-called prompt energy release) is counted in theyield, and there does not appear to be any gen-erally accepted precise definition of what energyrelease time scale is considered “prompt.” Alsoa complication arises in determining the yieldof underground nuclear explosions due toenergy released by interaction of the neutronsfrom the explosive device with the surround-ing ground. Consequently, there might beslight differences in how the United States andthe Soviet Union measure yields.

Yields can be estimated not only throughseismic means, but also by other methods. Theother methods are based on analysis of the nu-clear byproducts of the explosion (radiochem-ical methods) or measurements of the speedof the shockwave generated by the explosionin the surrounding rock (hydrodynamic meth-ods). Neither radiochemical nor hydrodynamicmethods are currently used by the UnitedStates to measure routinely the yields of So-viet explosions because they require access tothe test site during the test, and in the caseof radiochemical methods, may require infor-mation about the weapon design that could re-veal sensitive information concerning the char-acteristics of the weapon.

MEASURING THE SIZE OF SEISMIC SIGNALS

At present, U.S. estimates of Soviet yields waves, Rayleigh waves, and Lg waves.2 Theare generally made using seismic waves re- various magnitudes are averages based oncorded at teleseismic distances (distancesgreater than 2,000 km). Seismic magnitudes 2A description of the various types of seismic waves is pre-can be determined from the ampl i tudes o f P sented in Chapter 3, The Role of Seismology.

113

114

recordings at several stations. The magnitudesare then converted to explosive yields usingformulas developed through past experience.The formulas used are based on testing experi-ence at the Nevada test site and at the testsite operated in the Sahara by France in the1960s. The three magnitude measures mostoften used in yield estimation are: the P-wavemagnitude mb the surface wave magnitudeMs, and the Lg-wave magnitude mb(Lg).

The mb is computed from measurements ofP-wave recordings by the use of the formula

m b = log (A/T) + B.

As illustrated in figure 7-1, A is the largestP-wave amplitude in nanometers (0.000000001meters) measured peak-to-peak from a seismicshort-period recording during the first few se-conds of the P wave and correcting it for theinstrument magnification, T is the duration ofone cycle of the wave in seconds near the pointon the record where the amplitude was meas-ured (for P waves the period is typically 0.5to 1.5 seconds), and B is a distance-dependentcorrection term that compensates for the changeof P-wave amplitudes with distance.

The surface wave magnitude is determinedby measuring the Rayleigh-wave amplitudenear the point where the dominant period ofthe wave is nearest to 20 seconds on long-period vertical component records. The for-mula used is

Figure 7-1.–Computation of P-Wave Magnitude

A

s e c o n d so 1 2 3 4 5

Measurement made on P waves to obtain the magnitude ofa seismic event. The peak-to-peak amplitude (A) in the firstfew seconds of the P wave is corrected for the instrumentmagnification at the dominant period T.

Ms = log (A/T) + D,

where A and T are the amplitude and theperiod measured off long-period vertical com-ponent seismic recordings, again in nanometersand seconds, and D is a distance-dependent cor-rection term for Rayleigh waves.

The magnitude measure derived from meas-urements of Lg waves is computed from theformula

mb (Lg) = 5.0 + log [A(10km)/110],where A( 10 km) is the maximum sustained am-plitude of Lg on short-period vertical recordsin nanometers extrapolated backwards to a dis-tance 10 km from the source by dividing bythe geometrical spreading factor of d-5/6, whered is the source-to-receiver distance, and by theestimated attenuation along the path. The em-pirical mb(Lg) v. log Yield (Y) relationship alsoincludes a small second-order term, givingmb(Lg) = 3.943 + 1.124 log Y -0.0829 (log Y)2

for explosions in water saturated rocks suchas those at the Nevada Test Site.

The mb magnitude is routinely used at tele-seismic distances for yield estimation becauseP waves are detectable at large distances, evenfor small seismic events. This measure canalmost always be obtained for any seismicevent that is detected. The measurement ofMs requires a larger event, because Rayleighwaves are small for nuclear explosions. For ex-plosions below 50 kt, Ms may be missed al-together at teleseismic distances. The Lg am-plitude is similarly weak for small explosions.Consequently, it maybe important for seismicstations to be close to the explosion if surfacewaves and Lg waves are to be used for yield esti-mation of explosions less than 50 kt. This is oneof the reasons why seismic stations within theterritories of the treaty participants are desira-ble. The distance correction factors can be quitevariable regionally, and hence, some of themagnitude-yield relationships will need to beadjusted for different regions.

In addition to the conventional surface wavemagnitude Ms, a new measure of sourcestrength for surface waves is coming into wide

715

use. Called seismic moment (MO), it is an esti-mate of the strength of a compressional (ex-plosion-like) force at the explosion site. Seis-mic moment gives a direct description of theforce system, acting in the Earth, that wouldmake seismic waves of the size and shape ac-tually recorded. The advantage of using seis-mic moment is that the computation can cor-rect for the estimated effects of contaminationof the seismic signals due to earthquake-likemotion triggered by the explosion. This is use-ful because nuclear explosions often releasestress that has been built up in the area of the

explosion by geological processes. The releaseof built-up stress by the explosion creates asurface wave pattern similar to that observedfor earthquakes, which is seen superimposedon the signals of the explosion. Characteris-tics of an earthquake, such as Love waves andreversed polarities in the Rayleigh waves, areoften observed from a nuclear explosion, in-dicating release of pm-existing stress. If notremoved, this release of natural stress by theexplosion, called tectonic release, can distortyield estimates obtained from conventionalMs.

DETERMINING EXPLOSIVE YIELD FROM SEISMICMAGNITUDE

Once the seismic magnitude measurementshave been made, the next step is to relate themagnitude measurements to the yield of theexplosion. Because we know the actual yieldsonly of U.S. tests and some French nuclear ex-plosions (the Soviets have announced yieldsonly for a few of their tests), our knowledgeis based on data other than Soviet data. Theactual data used to derive this relationship areshown in figure 7-2. The relationship betweenthe yield of a nuclear explosion and the meas-

Figure 7-2.—Explosions in Granite

I

4.0

I

Yield (kt)

Data for explosions in granite from which the magnitude v.yield equation is derived.SOURCE: Modified from Air Force Technical Application Center.

ured seismic magnitude can be described usingan equation of the general form

M = A + B log Y + Bias Correction

where M is a magnitude measure (or moment)from surface waves, body waves, or Lg waves,A and B are constants that depend on whichmagnitude measure is used, and Y is the yieldin kilotons. The specific constants used by theUnited States for these calculations are clas-sified. The “Bias Correction” term is an ad-justment made to correct for the differencesin how efficiently seismic waves travel fromthe various test sites. This correction is par-ticularly important for mb, because short-period body waves are strongly affected by thephysical state (especially temperature) of themedium through which they travel.

The empirical magnitude-yield relationshipsfor mb that are used to estimate yields at in-accessible test sites in the U.S.S.R. and else-where have been revised several times duringthe last two decades. These revisions were im-provements in yield formulas and computa-tional procedures to correct for such problemsas difficulties in merging magnitude sets fromdifferent station configurations and instru-ments, clipping (limiting the maximum record-able amplitudes) of large signals by the record-ing systems, and not correctly accounting for

116

differences in the geology at different test sites.The early magnitude-yield formulas were basedon the simplifying assumption that all nuclearexplosions in granite at any site follow a simple

linear relationship between mb and log(yield).After the factors listed above were properlyconsidered, however, it became obvious thatbias corrections for each test site were needed.

SOURCES OF UNCERTAINTY

Under the ideal conditions of a perfectly uni-form and symmetrical Earth, it would be pos-sible to estimate yields of nuclear explosionsat any site from measurements at a singleseismic station. In practice, however, seismicmagnitudes and magnitude-yield plots showscatter. Using data from the InternationalSeismological Centre3 as an example, individ-ual mb measurements typically have a stand-ard deviation of 0.3 to 0.4 magnitude units be-fore station corrections are applied. Whenstation corrections are applied, the standarddeviation is reduced to 0.1 to 0.15 units. Fig-ure 7-3 illustrates typical scatter in a magni-tude-yield plot.

‘The International Seismological Centre is an organizationbased in England that gathers data for the research commu-nity from thousands of seismic stations operated all over theglobe.

Figure 7-3.— Yield Data From the Nevada Test Site

7 . 0

6.0

5.0

4.0

Yield (kt)

Mb versus yield for explosions at the Nevada Test Site. Thescatter is characteristic of yield measurements when onlyP-wave magnitudes are used.SOURCE: Modified from Air Force Technical Applications Center.

One reason for this variation is the small-scale geologic contrasts in the Earth that causefocusing and scattering of seismic waves. Fo-cusing effects near the recording seismometerscan create differences in estimated magnitudeseven when the stations are closely spaced.Focusing effects near the explosion can causebroad regional variations of seismic amplitudesso that seismic observatories over whole con-tinents may observe higher or lower averageamplitudes than the global average. Fortu-nately, the uncertainty introduced by focusingeffects can be reduced by averaging measure-ments from numerous stations if the stationsare well distributed around the test sites inboth distance and direction.

In addition to the scatter due to focusing,geological structures under individual stationsmay amplify seismic waves. Such effects may

Figure 7-4. —Yield Estimate Distribution

95% oftotal area

—

o 75 150 225 300 375

Yield estimate. in kilotons

50 % of


Probability of yield estimates for a 150 kt explosion meas-ured with a factor-of-2 uncertainty.

117

be corrected for by applying statistically de-rived station corrections that compensate forany such local effects.

After averaging many measurements andapplying appropriate corrections, estimates ofthe yield of a nuclear explosion are expectedto be distributed about the “true” value in themanner indicated in figure 7-4. The horizontalaxis in this figure is the yield estimate whilethe vertical axis is the probability that the esti-mate is correct. The area under the curve be-tween 2 yield values represents the probabil-ity that the actual yield is in this interval (thepercentage chance is 100 times the probabil-ity and the total area under the curve is 1, giv-ing a 100 percent chance that some value ofmagnitude will be measured). This figure showsthat it is most likely that the central yield value(150 kt in this case) will be close to the actualvalue and that outcomes become increasinglyless likely the larger the difference between theestimated value and the central value (see chap-ter 2 for a more detailed discussion of uncer-tainty and what it represents). The yield dis-tribution is asymmetric due to the normaldistribution of mb and the logarithmic rela-tionship between the yield of the explosion andthe measured seismic magnitude. Figure 7-3

is a typical empirical magnitude-yield curveobtained from actual data at the Nevada testsite that shows the measurements do not fol-low a single line but scatter around it becauseof measurement errors and variations in rockssurrounding the explosions.

Some of the uncertainty described above isdue to variations in how well explosions arecoupled to the surrounding rock. Also, explo-sion depth can influence the amplitudes of theseismic waves emitted, as can variations in thephysical properties of the Earth. For inacces-sible test sites, these effects result in increaseduncertainty in estimating yields. However, ifdata were exchanged and calibration shots per-formed, corrections could be made that wouldgreatly reduce the uncertainty. Nevertheless,there will always be some uncertainty in esti-mates of the yields of Soviet explosions, as inestimates of any physical quantity. This is notunique to seismology. Some uncertainty willexist no matter what type of measurement sys-tem is used. Such uncertainty should not nec-essarily be considered to represent opportuni-ties for cheating. Chapter 2 discusses themeaning of the various uncertainties and theirimplications for cheating.

BIAS CORRECTION FOR SOVIET NUCLEAR EXPLOSIONS

In estimating the yields of Soviet explosions,a major concern is how well the magnitude-yield formula for U.S. tests can be applied toSoviet test sites. Geophysical research hasshown that seismic P waves traveling throughthe Earth’s mantle under the main U.S. testsite in Nevada (and many other areas of theworld as well) are severely attenuated whencompared to most other continental areas,especially those with no history of recent platetectonic movements. If not corrected for, thisattenuation will cause a sizable systematic er-ror in estimates of the yield of Soviet ex-plosions.

The apparent reason for this attenuation isthe high temperature in the upper mantle un-

der Nevada and many other tectonically ac-tive regions. Regions of high temperatureschange the elastic and absorptive propertiesof the rocks, causing a large loss in the ampli-tudes of seismic waves traveling through them.Similar phenomena are thought to occur un-der the French test sites in Algeria and thePacific, though not under either the Soviet testsites in Kazakh and Novaya Zemlya or the U.S.test sites in Mississippi and Amchitka. If theP-wave magnitudes observed from U.S. testsin Nevada are used as a basis for estimatingyields, most Soviet explosions which have beenexploded in areas where the upper mantle iscool and there is little attenuation of P waveswill appear considerably larger than they ac-tually are.

———

118

The evidence for such attenuation effectscomes from many studies, including:

●

●

●

●

comparisons of P-wave amplitudes ob-served at the Nevada test site with obser-vations made at other sites in areas un-derlain by colder mantle;studies of short period S-wave amplitudes,which are very sensitive measures of man-tle temperature variations;studies of the frequency content of bothP and S waves, i.e., the relative loss of highfrequency energy in waves travelingthrough the upper mantle in both direc-tions under Nevada; andstudies of P- and S-wave velocities, which

In addition, there is a large amount of inde-pendent geophysical evidence supporting thenotion of anomalously high temperatures un-der most of the western United States. Thisevidence includes:

●

●

●

measurements of anomalously high heatflow,measurements of electrical conductivity,andthe low velocity P waves (Pn) and the ab-sence of S waves (Sn) that propagate justunder the Earth’s crust.

These “symptoms” of high attenuation havebeen observed in many other areas of the worldand are recognized as such by most geophysi-

are also influenced by temperature. cists. The sketch in figure 7-5 illustrates how

Figure 7-5.—Schematic Illustration of Attenuation. Related Magnitude Bias

Signal from Signal fromSoviet test U.S. test

Soviet Union’stest site

Us.test site

Seismic body waves crossing parts of the upper mantle with high temperatures become anonymously reduced in amplitude.Seismic signals from the Soviet Union’s test site appear much larger than signals from an identical explosion conducted atthe U.S. test site.

SOURCE: Modified from Defense Advanced Research Projects Agency.

119

this attenuation is created in the Earth andaffects estimates of the size of the wave source.Seismic body waves crossing the hatched highattenuation zones in the upper mantle are re-duced in amplitude and high frequency com-ponents of wave motion relative to waves thatbypass such zones.

Magnitudes derived from Rayleigh wavesand Lg waves are less influenced by tempera-ture variations in the mantle because theytravel along the surface and largely bypass thehigh attenuation zones in the upper mantle.A plot of P-wave magnitudes (rob) against sur-face wave magnitudes (Ms) should, therefore,show the attenuation of P waves relative tosurface waves. By plotting the Ms - mb ratioof explosions for different test areas, the at-tenuation indifferent regions can be compared.Figure 7-6 shows the results of an early studythat compared the Ms - mb for explosions inEurasia with the Ms - mb for explosions inNorth America. It can be seen that the twogroups are offset, with explosions of the sameMs value having lower mb values in NorthAmerica than in Eurasia. Results like this ledto early speculation about the existence of highattenuation zones in the upper mantle.

In general, if the P-wave magnitudes areplotted against the Rayleigh or Lg magni-tudes for the Nevada and Soviet test sites, the2 sets of data are offset by about 0.3 to 0.4magnitude units (or an amplitude factor ofabout 2 for P waves).4 The most likely expla-nation for this offset is the reduction of P-wavemagnitudes due to attenuation at the Nevadatest site. Such data constitute additional sup-port for the idea of an attenuation bias for Pwaves. Offsets can also be brought about byother factors such as contamination of the MS

measurement by tectonic release. However,such contamination can be detected by thestrong Love waves the release generates, and

“See for example P. Il. Marshall and P. W. Basham, “Dis-crimination Between Earthquakes and Underground ExplosionsEmploying an Improved M. Scale, ” GeophysicaZJoumal of thelibyzdAstronomicaJ Soa”ety, vol. 28, pp. 431-458,1972, and OttoW. Nuttli, “L~ Magnitudes of Selected East Kazakhstan Un-derground Explosions, ” BulZetin of the Seismological Societyof Ame~”ca, vol. 76, No, 5, pp. 1241-1252, October 1986.

Figure 7-6.—Comparisons of Explosions in EurasiaWith Explosions in North America

7.0

6.0

5.0

I I 1 I3.0 4.0 5.0 6.0

M s

Explosions

● Eurasia o North America

Explosions with the same MS value have lower mb values inNorth American than in Eurasia. This led to early specula-tion about the existence of high attenuation zones in theupper mantle.

SOURCE: Modified from P.D. Marshall and P.W. Basham, Geophysical Journalof the Royal Astronomical Society, 1972, vol. 28, pp. 431-458.

reduced by using seismic moment instead ofsurface wave magnitude (Ms) for yield esti-mation.

Various government-supported scientificpanels of seismologists, after considering thetotality of the geophysical evidence, haverepeatedly recommended during the last dec-ade that U.S. yield estimates of Soviet explo-sions be reduced by subtracting a larger “BiasCorrection” term from the magnitudes to ac-count for the attenuation effect on mb. As aresult, the bias correction has been increasedon several occasions over the last decade asnew scientific evidence indicated that suchchanges are appropriate.

The size of the bias correction was deter-mined simply by averaging the correction in-ferred from a number of independent or semi-independent estimates of the attenuation ef-fect made by different researchers. Most evi-dence for an attenuation “bias” has been in-direct thus far, although the evidence from

76-584 0 - 88 : QL 3 - 5

120

global seismic studies and seismological experi-ence gives strong support to the idea. Moredirect measurements of this bias may soon be-come available. The United States and SovietUnion have recently agreed on experiments tocalibrate seismic yield estimation methodsthrough measurements at each others testsites. Explosions will be measured with seis-mic methods and the yields confirmed inde-pendently by hydrodynamic methods. In addi-tion, several seismic stations have been set uprecently in the Soviet Union near the Kazakhtest site by a group of U.S. scientists supportedthrough the Natural Resources Defense Coun-

cil. Data from these stations will help improveestimates of the bias correction and assess theefficiency of seismic wave propagation at highfrequencies to regional distances.

The bias correction is currently used as a sim-ple, yield-independent adjustment to the in-tercept, A, in the rob-log Y curve. The valuecurrently used by the U.S. Government is in-tended to be the most appropriate value foryields near the 150 kt threshold of the 1974Threshold Test Ban Treaty. A different biasmay be appropriate for yields that are eithermuch larger or much smaller.

REDUCING UNCERTAINTIESThe estimated yield of an underground of nu-

clear explosion, like any quantity derived frommeasurements, has some error associated withit. The error comes from a variety of sources.Some of the error is considered to be randomin that it varies unpredictably from one meas-urement to another. Other errors are not ran-dom but are systematic. Systematic errors arethose that always act in the same way, for ex-ample, the bias between test sites. If system-atic errors are understood, corrections can bemade to reduce or eliminate the error.

The distinction between random and system-atic errors, however, has no clear boundary.If everything about the Earth and seismicwaves were known, almost all errors in seis-mology would be systematic. In general, ran-dom errors usually turn out to be systematicerrors once the reason for the error is under-stood. However, if the systematic errors arenot understood, or if there are lots of system-atic errors all operating indifferent ways, thenthe systematic errors are often approximatedas random error. In such cases, random uncer-tainties are inflated to encompass the unex-plained systematic uncertainties.5

‘As discussed in Chapter 2, random errors do not provide op-portunities for cheating. However, if systematic errors are foundto be 1) of sufficient size, 2) usable for an advantage by one side,and 3) unrecognized as being systematic by the other side, thensuch errors can be exploited under some situations. A treatyshould, therefore, contain provisions to reduce the uncertaintyof yield estimates and counter evasion opportunities.

Random Uncertainty

Different methods of yield estimation havedifferent accuracies and uncertainties. At theNevada test site, the most precise method usesP-wave magnitudes (rob). Less precise meth-ods use Lg waves (mb(Lg)), surface waves (MS),

Ms)and seismic moment ( At the Nevada testsite, yields estimated from mb alone have arandom uncertainty factor of 1.45 at the 95percent confidence level, whereas those frommb(Lg) have an uncertainty factor of 1.74, andthose from Mo have an uncertainty factor of2.13.

Recently, the Defense Advanced ResearchProjects Agency (DARPA) has been able toreduce the random uncertainty in seismic yieldestimates at the Nevada test site by combin-ing measurements made by the three differ-ent methods. Scientists have shown, usingdata from U.S. explosions at the Nevada testsite, that the random errors of the three typesof magnitude measures for a given event canbe considered statistically independent. Con-sequently, an improvement in the accuracy ofyield estimates can be achieved by combiningseveral methods to produce a “unified” mag-nitude measure. By forming a weighted aver-age of the three magnitudes, a “unified seis-mic magnitude” with an uncertainty factor of1.33 (figure 7-7) has been derived. Most seis-mologists believe that if this method were now

121

Figure 7-7.—Uncertainty

6.2

5.4

1.74

in Yield Estimates

/100 200 300400

/2.13

Yield (kt)

95%Factor

1.33

1 I100 200300400

Yield (kt)

Uncertainty in yield estimates can be greatly reduced through the use of a unified seismic yield estimate. On the left are threeplots of mb, mb(Lg), and MO versus yield at the Nevada Test Site. On the right is a similar plot of the unified seismic yield esti-mate versus the actual yields. The 95 percent uncertainty factors are shown to the right of each plot.SOURCE: Modified from Defense Advanced Research Projects Agency.

applied to estimating yields at the Soviet testsite in Eastern Kazakh, the uncertainty wouldbe reduced to a factor of 1.6- 1.5 for explosionsaround 150 kilotons.6 What limits the uncer-tainty from being reduced to the level of the

‘Consider, as an example, the situation where there are 3 sta-tistically independent methods of calculating the yield of anexplosion, all of which (for the sake of this example) have a fac-tor of 2 uncertainty in a log normal distribution:

# of Methods Resulting Uncertainty1 2.02 1.63 1.5

By combining methods, the resulting uncertainty can be re-duced. This methodology, however, can only reduce the randomuncertainty. Systematic uncertainty such as differences betweenthe test sites will remain and limit the extent to which the un-certainty can be reduced unless calibration is performed.

Nevada test site (a factor of 1.3) is the system-atic uncertainty or bias correction. As we willsee later, however, this systematic uncertaintycan be reduced through calibration shots.

The expected precision given above are onlyfor explosions where all waves are used for theestimation. For smaller explosions, the re-gional Rayleigh and L waves are not always

strong enough to travel the long distances re-quired to reach seismic stations outside the So-viet Union. Consequently, for monitoring low-yield explosions, stations within the SovietUnion may be necessary to obtain the improvedaccuracy of the “unified seismic yield” esti-mation method. The relationship between mag-nitude and yield for the stations within the So-viet Union will also have to be established.

—

122

As noted above, the formulas derived fromthe Nevada test site data that describe the rela-tionships between yield and mb(Lg) and mo-ment MO are not directly applicable to the So-viet test site in Kazakh without some as yetunknown adjustments. The values of these ad-justments can be determined if stations areplaced within the Soviet Union and the Soviettest sites are calibrated. Test site calibrationssuitable for lower yields could only take placeafter an internal network is installed. If themb(Lg) and MO v. yield curves are suitablycalibrated, the absolute yields of explosionsat Kazakh should be measurable with the “uni-fied seismic method” just as accurately as atthe Nevada Test Site.

The above analysis applies only to explosionsat known test sites observed by a large set ofwell-distributed seismic stations for which theappropriate station corrections and bias cor-rection have been determined. The accuracywith which the yields of “off-site” nuclear testscould be estimated would be less than thatstated above. Therefore, to maintain high ac-curacy in yield estimation, nuclear testingshould be prohibited outside specified, cali-brated test sites.

Most yield estimation research has concen-trated on yields around 150 kt, so the accuracythat could be achieved by seismic methods atlower yields is not yet well known. In any fu-ture low threshold test ban treaty, it might beexpected that the initial uncertainties in yieldestimation for explosions below 10 kt wouldbe large. These uncertainties would then be re-duced as more data were gathered, as ourknowledge of wave propagation properties forvarious paths in the monitored regions was re-fined, and as calibration information was ob-tained.

Systematic Uncertainty

The yield estimation precision describedabove for teleseismic data are limited becauseof systematic uncertainties. As discussedabove, the systematic uncertainty can be re-duced by calibrating the test site. Calibratinga test site involves exploding devices whoseyields are either known or accurately deter-mined by independent means, and then meas-uring the magnitudes at a large number ofmonitoring stations. By doing so, the yieldsof other events can be determined by compar-ing the amplitudes of the seismic waves at com-mon seismic recording stations with thoseoriginating from the events with known yields.

This approach reduces the systematic uncer-tainties caused by having to estimate the vary-ing properties of the rocks surrounding theexplosion and any focusing effects near the ex-plosion sites. As long as these factors remainapproximately unchanged within a geologi-cally uniform area, the calibration improvesthe estimation of yields.

The sizes and numbers of geophysically dis-tinct subdivisions in any test site depend onthe geological structures of the area. A spe-cific calibration maybe valid only for a limitedarea around the shot if, at larger distances, therock properties and focusing effects change.The distances over which the relevant condi-tions change vary, depending on the local ge-ology. Testing areas that are large or containvarying geology would obviously need morecalibration shots than areas that are geologi-cally uniform. If calibration were performedat the Soviet test site, the expected seismicyield estimation capability would be compara-ble to the existing seismic capability at the Ne-vada Test Site.

YIELD ESTIMATION CAPABILITIES

In considering the capability of all methods is radiochemical methods. Radiochemical meth-of yield estimation, it must be kept in mind ods of yield estimation have an uncertainty ofthat it is never possible to determine a yield about 10 percent (a factor of 1.1). Also, experi-without some uncertainty. The standard against mental devices often detonate with yields thatwhich yield estimation methods are measured are slightly different from what was predicted.

123

This uncertainty in predicted yield was recog-nized during the negotiations of the Thresh-old Test Ban Treaty and provisions were estab-lished for unintended breaches (see ch. 2).

The yields of Soviet underground explosionscan be seismically estimated with a much bet-ter capability than the factor-of-2 uncertaintythat is commonly reported.7 New seismic meth-ods have greatly improved yield estimation ca-pabilities. Further improvements would occurif the test sites were calibrated and, for smalltests, if stations were present within the So-viet Union during the calibration. The capa-bilities depending on these variables can besummarized as follows:

●

●

Without Calibration: For large explosions(above 50 kilotons) seismic yield estima-tion could be improved with the additionaluse of the other methods including: sur-face waves, Lg waves, and seismic mo-ment. Through such a combined method,it is estimated that without calibration So-viet yields can be seismically measuredwith present resources to a factor of 1.6to 1.5 uncertainty.With Calibration: Further reductions inthe uncertainty of yield estimates can beaccomplished if the Soviet test site werecalibrated. At a defined, well-calibratedSoviet test site, it is estimated that yieldscould be seismically measured with thesame factor of 1.3 uncertainty that isfound for seismic estimates at the Nevada

‘See, for example, VerifyingIVuclear TestingLim.itations: Pos-sible U.S.-Soviet Cooperation (Washington, DC: U.S. Depart-ment of State, Bureau of Public Affairs, Special Report No. 152,Aug. 14, 1986)

●

Test Site. In fact, Soviet seismologistshave told U.S. seismologists that they areable to estimate yields seismically at theirown test site with only a factor of 1.2 un-certainty.Small Explosions: For small explosions(below 50-kt), the regional seismic waves

may not always be strong enough to travellong distances to seismic stations outsidethe Soviet Union. Consequently, seismicstations within the Soviet Union may benecessary (in addition to calibration) to ob-tain the 1.3 factor of uncertainty fromcombined seismic methods for explosionwith yields below 50 kt. At yields below10 kt small variations of the physical envi-ronment may produce greater uncertainty.Therefore, at yields below 10 kt, the un-certainty may be inherently greater.

A 1.3 factor of uncertainty (for yields above50 kt) is the claimed capability of the hydro-dynamic yield estimation method using CORR-TEX data8 that has been proposed as an alter-native means for improving yield estimation.Consequently, hydrodynamic yield estimationwill not provide a significantly superior yieldestimation capability over what could be ob-tained through well-calibrated seismic means(also a 1.3 factor of uncertainty). Hydrody-namic yield estimation is, however, one of themethods that could be used to provide inde-pendent estimates of the yields of calibrationshots to improve seismic methods. Once a testsite was calibrated using hydrodynamic meth-ods, there would be no need to continue theuse of those intrusive methods.

‘See appendix, Hydrodynamic Methods of Yield Estimation.

124

SOVIET TEST PRACTICES AND TEST BAN COMPLIANCESpecific concern over compliance with test 1. The mb of several Soviet tests at their

ban treaties has been heightened with findings Shagan River (E. Kazakhstan) test site areby the Reagan Administration that: significantly l arger than the mb for U.S.

Soviet nuclear testing activities for a num- tests with yields of 150 kt.ber of tests constitute a likely violation of the 2. The pattern of Soviet testing indicateslegal obligations under the Threshold Test that the yields of Soviet tests increasedBan Treaty.9 after the first 2 years of the treaty.

Such findings are presumably based on netassessments of all sources of data. In measur-ing yields near the 150 kt limit of the Thresh-old Test Ban Treaty, however, seismic evidenceis considered the most reliable basis for esti-mating the yields of Soviet underground nu-clear explosions.10 It is, therefore, the seismicevidence that has received particular attention.

Concern about whether the Soviet Union isactually restricting its testing to a maximumyield of 150 kt is motivated by two arguments:

“’The President’s Unclassified Report on Soviet Noncompli-ance with Arms Control Agreements, ” transmitted to the Con-gress Mar. 10, 1987.

Conclusion of the Defense Intelligence Agency Review Psnelas stated in a letter from Roger E. Batzel, Director of LawrenceLivermore National Laboratory to Senator CMbome Pen onFeb. 13, 1987.

The validity of the first argument is depen-dent on how the Soviet yields are calculated.Because of the uncertainty in measuring theyields of Soviet tests using only mb and be-cause of differences in opinion as to what thecorrect bias value for Soviet tests should be,there is disagreement as to whether the mb

values of the largest Soviet tests do, in fact,represent violations of the 150 kt limit of theThreshold Test Ban Treaty. For example, whencalculations such as those in table 7-1 are madeusing both mb and MS measurements and abias correction of 0.35, they indicate that thefew remaining yields estimated as above 150kt are well within the expected random scat-ter, and do not support claims of a violation.

The second argument is dependent on as-sumptions about probable Soviet behavior.Two years after the signing of the Threshold

Box 7-A.—Calculations of the Six Largest East Kazakhstan Explosions. By Sykes et al.,Based on Unclassified Data

6 Largest East Kazakhstan Explosions 1976-1986Yield from Yield averaged from

Date mb only mb & MS

23 June 79. . . . . . . . . . . . . . . . . . . . . . . . . . 152 14914 Sept 80 . . . . . . . . . . . . . . . . . . . . . . . . . . 150 150*27 Dec 81 . . . . . . . . . . . . . . . . . . . . . . . . . . 176 1614 July 82 . . . . . . . . . . . . . . . . . . , . . . . . . . . 158 158*14 July 84 . . . . . . . . . . . . . . . . . . . . . . . . . . 140 140*27 Oct 84. . . . . . . . . . . . . . . . . . . . . . . . . . . 140 140*

Average: 152.7 (± 13.4 kt) Average: 149.7 (± 8.8 kt)(*based on mb only; no Ms determined)All estimated yields are well within the uncertainty expected for observance of the 150 kt threshold limit.

SOURCE: Calculations of the six largest East Kazakhstan explosions made by Sykes et al., based on unclassified data. Body wave measure-ments from International Seismological Centre Bulletins and United States Geological Survey Reports. Station corrections deter-mined to be 0.02 to 0.04 from mean. Surface wave calculations made by Sykes et al. Calibration corrected for bias using a valueof 0.35 to make body wave data consistent with surface wave data.

125

Test Ban Treaty, the size of the largest Sovietexplosions at their eastern Kazakh test siteincreased markedly (see figure 7-8).11 This in-crease has been interpreted by some to inferthat the Soviets have been violating the 150kt threshold limit in the later tests. The argu-ment assumes that the Soviets were testingup to the limit for the first 2 years and, there-fore, by inference, have been testing above thelimit in violation of the treaty ever since 1978.Alternate interpretations for this apparentyield increase have been offered. It has beenpointed out that a similar pattern of testingoccurred at Kazakh for the 2 years prior to thetreaty. It has also been speculated that thisincrease in yields may reflect a Soviet decisionto move their high yield testing from the NovayaZemlya test site to the Kazah test site.12 As

~There w= &IO speculation that this increase was coincidentwith a change in the U.S. official method of yield estimation.For example, Jack Anderson, “Can’t Tell If Russia Cheats OnTest Ban,” The Waslu”ngton Post, Aug. 10, 1982, p. C15.

‘*See for example, “Nuclear Test Yields” (Letter to the Edi-tor), J. F. Evemden and L. R. Sykes, Science, vol. 223, Feb. 17,1984, p. 642.

anon-technical consideration, it can be arguedthat if the Soviets had tested above the limitof the Threshold Test Ban Treaty at the Kazakhtest site, they would never have offered to al-low the United States to calibrate their testsite using CORRTEX and Soviet test explo-sions. The calibration will reduce the uncer-tainty of yield estimates, a reduction that ap-plies to past as well as future explosions andhence can provide more accurate evidence con-cerning past compliance.

Because of the statistical nature of all yieldestimates, the question of compliance can beaddressed best not by looking at individualtests but rather by examining the entire pat-tern of Soviet testing. It is particularly usefulto compare the testing programs of the UnitedStates and the Soviet Union. It can be seenfrom figure 7-9 that if a bias value lower than0.35 is used, there appears to have been about10 (out of over 200) Soviet tests since the sign-ing of the Threshold Test Ban Treaty in 1974with yield central values above the 150 ktthreshold limit. When the same method of yield

Figure 7-8.— mb Versus Time for Large Soviet Explosions

E

5.0 ‘

4.4 “ ‘

*o ● .

I I I I 1 I1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 198’1 1982 1983

. Kazakhstan O Novaya Zemlya

The m b versus time for all large Novaya Zemlya and Kazakhstan explosions. It can be seen that a large increase of the maxi-mum yield for explosions at the Eastern Kzazkh test site occurred about 2 years after the Threshold Test Ban Treaty was signed.

SOURCE: T.C. Bache, S.R. Bratt, and L.B. Bache, “P Wave Attenuation mb Bias, and The Threshold Test Ban Treaty,” SAIC-86-1647, submitted to AFGL, March 1966, p. 5.

126

Figure 7-9.—U.S.S.R. Nuclear Tests 1966-86 According to Sykes and Davis5,000

2,000

1,000

500

100

50

201966 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985

● Kazakhstan O Novaya ZemlyaThese yields were calculated by combining P-wave and surface-wave magnitudes and using a bias correction of 0,35. Barsdenote the estimated standard deviations of the estimates. The few tests that do appear to have exceeded 150 kt are wellwithin the expected scatter. If a lower bias correction is applied and only P-wave determinations are used, then slightly higheryield estimates will result and additional central values will be greater than 150 kt.

SOURCE: Modified from L.R. Sykes and D.M. Davis, Scientific America, vol. 258, No. 1, January 1987, pp. 29-37.

estimation is applied to U.S. tests, approxi-mately the same number of U.S. tests also ap-pear to be above the 150 kt threshold limit.This, however, does not mean that one or theother or both countries have cheated; nor doesit defacto mean that seismology is an inade-quate method of yield estimation. It is inher-ent in any method of measurement that if bothcountries are testing up to the yield limit, theestimated yields of some tests will have cen-tral values above the yield limit. Because ofthe uncertainty of measurements using anymethod, it is expected that about half the So-viet tests at 150 kt would be measured asslightly above 150 kt and the other half wouldbe measured as slightly below 150 kt.

All of the estimates of Soviet tests are withinthe 90 percent confidence level that one wouldexpect if the yields were 150 kt or less. Exten-sive statistical studies have examined the dis-tribution of estimated yields of explosions atSoviet test sites. These studies have concludedthat the Soviets are observing a yield limit.The best estimate of that yield limit is thatit is consistent with compliance with the 150kt limit of the Threshold Test Ban Treaty .13

I$such gtatistic~ studies have been carried out extensivelyby Lawrence Livermore National Laboratory, The conclusionof these studies was reported in open testimony before the Sen-ate Armed Services Committee on Feb. 26, 1987 by Dr. MiloNordyke, Leader of the Treaty Verification Program.

Appendix

Appendix

Hydrodynamic Methods of Yield Estimation

Hydrodynamic methods could be used to complementseismic methods of yield estimation

Introduction

The yield of an underground nuclear explosionmay be estimated using so-called hydrodynamicmethods. These methods make use of the fact thatlarger explosions create shock waves that expandfaster than the shock waves created by smaller ex-plosions. Three steps are involved in making a yieldestimate. First, the properties of the geologic me-dia at the test site that may affect the expansionof the shock wave are determined. Second, the ex-pansion of the shockwave caused by the explosionof interest is measured during the hydrodynamicphase, when the ambient medium behaves like afluid. Finally, the yield of the explosion is estimatedby fitting a model of the motion of the shock frontto measurements of the motion.

Although the algorithms used by different indi-viduals or groups can (and usually do) differ in de-tail, most of the algorithms currently in use areof four basic types: insensitive interval scaling,similar explosion scaling, semi-analytical modeling,and numerical modeling. Before considering thesealgorithms and their application to test ban veri-fication, it will be helpful to have in mind how theshock wave produced by an underground nuclearexplosion evolves during the hydrodynamic phaseand how this evolution is affected by the proper-ties of the ambient medium.

Shock Wave Evolution

The hydrodynamic evolution of the shock waveproduced by a large, spherically symmetric explo-sion underground may be usefully divided intothree different intervals. These are listed in tableA-1, along with the times after detonation at whichthey begin for 1 and 150 kiloton (kt) explosions ingranite. The characteristics of these intervalsfollow.

Self-Similar Strong-Shock Interval

At the very earliest times, the energy of the ex-plosion is carried outward by the expandingweapon debris and by radiation. Soon, however,

Table A-1.—Characteristic Times in the Evolution ofa Shock Wave Caused by an Underground

Nuclear Explosiona

Event in the evolution Time (µs) for a Time (µs) for aof the shock wave 1 kt explosion 150 kt explosion

Beginning of the Self-SimilarStrong Shock Intervalb – - 1 0

Beginning of the TransitionInterval c. . . . . . . . . . . . . - 2 0 - 8 0

Beginning of the PlasticWave Intervald. . . . . . . . . . . . . . . -2,000 -10,000

aFOr an idealized Spherically-syrnrnelrlc exploslon In granitebf+ere fhe shock wave is assumed 10 become self-similar when If reaches a radws of 1 m (see

text) No time is gwen for a 1 kt explosion because the shock wave caused by such an exploslonfyplcally weakens before it has time to become self -s!mllar

coeflned as the time when the denslfy just behind the shock front falls to 80 percent Of Its maXU’rWM

limiting valued~fined as the time when the speed of the shock wave falls to 120 perCent Of the plaStlC Wavespeed In reality, granite undergoes a phase transition sllghlly before the shock speed reachesthis value, a complication that has been neglected In this Illustration

SOURCE’F K, Lamb, “Hydrodynamic Methodsof Yield Estlmatlon ‘‘ Report Iorthe U S Congressoffice of Technology Assessment, Feb 15, 1988

a shock wave forms and begins to move outward.At this time the speed of the shock wave is muchgreater than the speed of sound in the undisturbedambient medium, the pressure behind the shockwave is predominantly thermal pressure, and theratio of the density behind the shock wave to thedensity in front is close to its limiting value. Thisis the strong shock interval.1

If the shock wave envelops a mass of materialmuch greater than the mass of the nuclear chargeand casing while it is still strong, and if energytransport by radiation can be neglected, the shockwave will become self-similar, expanding in a par-ticularly simple way that depends only weakly onthe properties of the medium.’ The time at which

‘See Ya. B. Zel’dovich and Yu. P. Riazer, Physics of Shock Wavesand High-Temperature Phenomena (New York, NY: Academic Press,1967 ~nglish Translation]), ch. XI. As the strength of a shock waveis increased, the ratio of the material density immediately behind it tothe material density immediately in front of it generally increases, untila value of the ratio is reached beyond which an increase in the strengthof the shock wave produces little or no further increase in the densityof the post-shock material. This density ratio is referred to as the limit-ing density ratio. In typical rocks, pressures behind the shock front ofabout 10-100 Mbar are needed to produce a density ratio close to thelimiting value.

‘Ibid., ch. I and XII.

129

130

the motion becomes self-similar depends in parton the design of the nuclear charge and diagnosticequipment and on the size of the emplacement hole.As the shockwave expands, it weakens and slows,the density behind the shock front drops, and thewave enters a transition interval in which the mo-tion is no longer self-similar. No time is given intable A-1 for the beginning of the self-similarstrong-shock interval for a 1 kt explosion becausethe shock wave produced by such an explosiontypically weakens before it has time to become self-similar.

Transition Interval

As the shock wave weakens and slows, it entersa broad transition interval in which the thermalpressure is not much greater than the cold pres-sure of the medium. The motion of the shockwavechanges only gradually and so the time at whichthe transition interval is said to begin is purely con-ventional. In this report the shockwave is consid-ered to have entered the transition interval whenthe density just behind the shock front has fallento 80 percent of its maximum limiting value. Thespeed of the shock front is only a few times greaterthan the relevant sound speed-the speed of theso-called plastic wave—in the medium over muchof the transition interval and hence the motion ofthe shock wave in this interval is more sensitiveto the properties of the medium than it is in thestrong shock interval.3

Plastic-Wave Interval

In the absence of phase transitions and othercomplications,4 the shock wave weakens and slowsstill further, entering an interval in which the pres-sure behind the shock front is predominantly thecold pressure of the compressed ambient mediumand the shock speed is close to the plastic wavespeed in the medium. Again, the motion of theshock wave changes only gradually and so the timeat which the plastic-wave interval is said to beginis purely conventional. In this report the shockwave is considered to have entered the plastic-waveinterval when its speed is less than 120 percent ofthe plastic wave speed in the medium. In practice,phase transitions and other effects complicate theevolution of the shock wave in this interval forrocks of interest.

‘Ibid., pp. 741-744,‘Ibid., ch. XII.

Theoretical models and experimental data showthat the evolution of the shock wave in all threeintervals depends on such properties of the rockas its chemical composition, bulk density, plasticwave speed, and degree of liquid saturation. Theseproperties vary considerably from one rock toanother. As a result, the shock wave generally de-velops differently in different rocks. For example,the characteristic radius at which the shock waveproduced by a 150 kt explosion changes from astrong, self-similar wave to a plastic wave5 variesfrom about 30 meters in wet tuff to over 60 metersin dry alluvium.

Measuring the Position of theShock Front

Several techniques have been used to measurethe position of the shock front as a function of time.During the 1960s and early 1970s, extensive meas-urements were made using the so-called SLIFERtechnique. 6 In the mid-1970s an improved tech-nique, called CORRTEX, was developed.7 This isthe technique the Reagan administration has pro-posed as a new technique to monitor the Thresh-old Test Ban Treaty (TTBT).8

In the CORRTEX technique, an electrical sens-ing cable is lowered into a vertical hole to a depthgreater than the depth at which the nuclear explo-sion will take place, typically hundreds of metersfor explosives with yields near 150 kt. The hole maybe the one in which the nuclear explosive is placed(the emplacement hole) or one or more other holes(so-called satellite holes) that have been drilled spe-cifically for this purpose. The latter geometry isshown in figure A-1. If satellite holes are used, theymust be drilled at the proper distance(s) from theemplacement hole, typically about ten meters foryields near 150 kt. Then, if the sensing cable isstrong enough that it is not crushed by other dis-

Y3ee F. K, Lamb, ACDIS WP-2-87-2 (University of Illinois Programin Arms Control, Disarmament, and International Security, Urbana,IL, 1987).

‘M. Heusinkveld and F. Holzer, Review of Scientific Instruments, 35,1105 (1964). SLIFER is an acronym for “Shorted Location Indicationby Frequency of Electrical Resonance.”

7C.F. Virchow, G.E. Conrad, D.M. Holt, et.al., Review of ScientificInstruments, 51,642 (1980) and Los Alarnos National Laboratory pub-lic information sheet on CORRTEX (April 1986). CORRTEX is an acro-nym for “Continuous Reflectometry for Radius v. Time Experiments. ”

W.S. Department of State, Bureau of Public Affairs, “U.S. Policy Regarding Limitations on Nuclear Testing, ’’Special Report No. 150, Au-gust 1986; and U.S. Department of State, Bureau of Public Affairs,“Verifying Nuclear Testing Limitations: Possible U.S.-Soviet Coopera-tion,” Special Report No. 152, Aug. 14, 1986.

131

CORRTEXrecorder

Sensingcable

Figure A-1 .—Use of the CORRTEX Technique

Surface

Sensingcable

point

Typical cable emplacementin satellite hole

CORRTEXrecorder

4

tic

Workingpoint

/

Surface

nostic

I//

\

\Shock front /progression

/

Moving shock wave fromnuclear detonation

crushes and shortens cable

SOURCE: U.S. Department of State, Bureau of Public Affairs, “U.S. Policy Regarding Limitations on Nuclear Testing,” Special Report No 150, August 1986

132

turbances but weak enough that it is crushed bythe pressure peak at the shock front, it will be elec-trically shorted close to the point where the shockfront intersects the cable (see figure A-l). As theshock front expands with time, the changing dis-tance from the surface to the shallowest point atwhich the shock wave intersects the sensing cableis measured at preset time intervals by electricalequipment attached to the cable and located aboveground. The CORRTEX technique is much less af-fected by disturbing early signals from the explo-sion than were earlier techniques.

The time at which the explosion begins is takento be the time at which the first signal, producedby the electromagnetic pulse (EMP) from the ex-plosion, arrives at the CORRTEX recorder. If theexplosion is spherically symmetric, the length ofthe unshorted cable decreases rapidly andsmoothly with time as the shock front expandsaway from the center of the explosion and the ra-dius of the shock front at a given time can be cal-culated using simple geometrical equations. If theexplosion is not spherically symmetric, due to theshape of the canister, the design of the nuclearcharge, or inhomogeneities in the ambient medium,the interpretation of CORRTEX data is more com-plicated and could be ambiguous or misleading un-der the conditions encountered in treaty verifica-tion. Problems of this kind can be prevented bycooperative agreements, as discussed below.

An error of 1 meter in the measured distance ofthe crushing point from the center of the explo-sion will cause an error of about 50 kt in the yieldestimate, for yields near 150 kt. Thus, an accuratesurvey of the satellite hole is required in order tomake an accurate yield estimate. Surveys are cur-rently made with special laser or gyroscopic equip-ment. In some yield estimation algorithms, thelateral displacement from the center of the explo-sion can be treated as one of the unknowns in esti-mating the yield.

Insensitive Interval ScalingOnce measurements of the radius of the shock

front as a function of time are in hand, an estimateof the yield of the explosion can be made by com-paring the measurements with a model of the mo-tion of the shock front away from the center of theexplosion. The simplest algorithm currently in useis insensitive interval scaling. This is the algorithmthat the Reagan administration has proposed touse in analyzing CORRTEX data as an additionalnew method of monitoring compliance with the 150kt limit of the TTBT.

Insensitive interval scaling is based on the as-sumption that the radius of the shock wave for anexplosion of given yield is independent of themedium during a certain interval in time and ra-dius called here the insensitive interval. Indeed,studies indicate that for the collection of rockswithin U.S. test experience (mostly silicates), rockproperties are correlated in such a way that thereis a time during the transition interval when theradius of the shock front produced by an explo-sion of given yield varies relatively little from onerock to another.9

In using insensitive interval scaling, the shockwave sensing cable must be placed close enoughto the center of the explosion that it samples theinsensitive interval. Yield estimates are then de-rived by fitting a simple empirical formula, calledthe Los Alamos Formula, to the shock radiusversus time data in this interval.10 The Los AlamosFormula is a power law that approximates the ac-tual radius versus time curve during the insensi-tive interval. This is illustrated by figure A-2,which compares the Formula with a model of theevolution of the shock wave produced in graniteby a spherically-symmetric point explosion witha yield of 62 kt.

Yield Estimation Algorithms

Hydrodynamic methods of yield estimation areevolving as research aimed at gaining a better un-derstanding of underground explosions and im-proving yield estimation methods continues. Atpresent, four basic types of algorithms are com-monly in use. In order to simplify their descrip-tion, the explosion will be assumed to be spheric-ally symmetric (the complications that can ariseif it is not will be addressed later).

The rocks for which the United States had good data or models arethe dry alluvium, partially saturated tuff, saturated tuff, granite, ba-salt, and rhyolite at the test sites used, almost all of which are at theNevada Test Site. At present, the reason for the correlation of rock prop-erties that gives rise ta the insensitive interval is not well understmdfrom a fundamental physical point of view. Moreover, it is known thatthe radius of the shock wave in this interval is very different for othervery different kinds of rocks. Thus, the existence of an insensitive in-terval must be established by test experience or modeling, and is onlyassured for certain geologic media.

1’The Los Alamos Formula for the shock radius in meters is R(t) =a.W%(t/W1/3)6, where W is the yield of the explosion in kilotons, t is theelapsed time since the beginning of the explosion in milliseconds, anda and /3 are constants. Different values of a and 6 have been used bydifferent individuals and groups and have changed with time. The valuesof a and /3 used here are 6.29 and 0.475 (see M. Heusinkveld, Journalof Geophysictd Research, 87, 1891, 1982).

133

Figure A-2.—Comparison of the Los Alamos FormulaWith a Semi= Analytical Model of the Shock Wave in

Granite Caused by a Spherically asymmetric PointExplosion With a Yield of 62 Kilotons

1000

100

10

1

— Grani te Model

— Los Alamos Formula

.0001 .001 .01 .1 1 10 100 1000

Time ( ins )

SOURCE: F.K. Lamb, “Hydrodynamic Methods of Yield Estimation,” Report forthe US. Congress, Office of Technology Assessment, Feb. 15, 1988.

In practice, the Los Alamos Formula is usuallyfirst fit to a broad interval of radius versus timedata that is thought to include the insensitive in-terval. The result is a sequence of yield estimates.Due to the departure of the Formula from the ac-tual radius versus time curve at both early and latetimes, the sequence of yield estimates typicallyforms a U-shaped curve. This is illustrated in fig-ure A-3, which shows the sequence of yield esti-mates obtained by applying the Formula to therelatively high-quality SLIFER data from thePiledriver explosion in granite. If the assumptionson which the algorithm is based are satisfied, theyield estimates near the bottom of the curve ap-proximate the actual yield of the explosion, In theusual form of the algorithm, only the radius versustime data that fall within a certain predeterminedinterval chosen on the basis of previous experience(the so-called algorithmic interval) are actually usedto make the final yield estimate. The length of thealgorithmic interval and the time at which it oc-curs are both proportional to W1/3, where W is theyield of the explosion (see table A-2). The algorith-

Figure A-3.—Application of the Los Alamos Formulato an Explosion in Granite

200

100

■ ✎

1 1

1 2 3 4 5Time (ins)

SOURCE: F.K. Lamb, “Hydrodynamic Methods of Yield Estimation,” Report forthe U.S. Congress, Off Ice of Technology Assessment, Feb. 15, 1988.

mic interval is indicted in figure A-3 by the twovertical bars at the bottom of the figure. In thisexample, the assumptions of the algorithm aresatisfied and the average of the yield estimatesthat lie within the algorithmic interval is very close

Table A-2.—Algorithmic Intervals for Various Yieldsa

Time Time interval (ins) Radius interval (m)

1 0.1-0.5 2-4.510: : : : : : : : : : : : : : : : 0.2-1.1 4.5-1o50. . . . . . . . . . . . . . . . 0.4-1.8 8-17100. . . . . . . . . . . . . . . 0.5-2.7 10-21150. . . . . . . . . . . . . . . 0.5-2.6 11-24aThe time intewals used by various individuals and groups varies. Throughout

this report the algorithmic interval Is taken to be from O.l W’~ milliseconds to0,5W1~ milliseconds after the beginning of the explosion, where W is in kilotons,

SOURCE: F.K. Lamb “Hydrodynamic Methods of Yield Estimation” Report forthe U.S. Congress, Office of Technology Assessment, Feb. 15, 1988.

134

to the announced yield of 62 kt. Studies indicatethat yield estimates made using this algorithmhave a precision of about a factor of 1.2 at the 95percent confidence level for spherically-symmetricexplosions with yields greater than 50 kt conductedat the Nevada Test Site (NTS). Estimates of theaccuracy of the algorithm made by comparing re-sults with the usually more accurate radiochemi-cal method are similar. According to official state-ments, the insensitive interval algorithm isexpected to be accurate to within a factor of 1.3at Soviet test sites for explosions with yieldsgreater than 50 kt in media within U.S. test ex-perience.” Some scientists believe that the uncer-tainty would be somewhat larger.

The insensitive interval algorithm does not workas well if the assumptions on which it is based arenot satisfied. This is illustrated in figure A-4, whichshows the yield estimates obtained by fitting theLos Alamos Formula to good-quality SLIFERdata from atypical low-yield explosion in alluvium.In this example the radius and time data have beenscaled using the actual yield so that the derivedyield should be 1 kt. However, the yield estimatesgiven by the Los Alamos Formula are systemati-cally low, ranging from 30 to 82 percent of the ac-tual yield, and do not forma U-shaped curve. Theaverage of the yield estimates that lie within thealgorithmic interval is about 60 percent of the ac-tual yield. The overall appearance of the yieldversus time curve shows that the assumptions ofthe algorithm are not satisfied.

A common misconception has been that the al-gorithmic interval lies within the strong shock re-gion and that the relative insensitivity of yield esti-mates to the properties of the medium stems fromthis.12 As explained earlier, radius versus time datain the algorithmic interval would indeed be rela-tively independent of the medium if this were so,and would follow a power-law curve similar to theLos Alamos Formula. However, the shock waveis not strong during the algorithmic interval, be-cause in this interval the shock speed is only a fewtimes the speed of sound and the post-shock pres-sure is much less than the pressure required toachieve the maximum limiting density. Indeed, theexponent of time usually used in the Los AlamosFormula is significantly greater than the value

“U.S. Department of State, op. cit., footnote 8.“For example, “The accuracy of the method is believed to be rela-

tively, but not wholly, independent of the geologic medium, providedthe satellite hole measurements are made in the ‘strong shock’ region. . . “ (Ibid.) This misconception may have arisen from the fact that the

interval formerly used to estimate the yields of nuclear explosions inthe atmosphere using hydrodynamic methods is within the strong shockregion.

Figure A-4.—Application of the Los Alamos Formulato a Low-Yield Explosion in Alluvium

O.8

0.7

0.6

0.5

.

.■

.●

.

......

........

m. .

.

.

0.1 0.3 0.5 0.7 09 1.1

Time (scaled ms)SOURCE: F.K. Lamb, “Hydrodynamic Methods of Yield Estimation,” Report for

the US. Congress, Office of Technology Assessment, Feb. 15, 1988.

appropriate for a strong shock wave. The LosAlamos Formula is, as noted earlier, an empiricalrelation, which was obtained by fitting a power-law expression to data from a collection of explo-sions in a variety of different rocks and approxi-mates actual radius versus time curves over a por-tion of the transition interval.

Similar Explosion Scaling

If a given explosion occurs in the same mediumas a previous explosion at a different site, and ifradius versus time data and the yield are availablefor the previous explosion, then the yield of thegiven explosion can be estimated by similar explo-sion scaling. The reason is that for explosions inthe same medium, the radius versus time curve de-pends only on the yield of the explosion, and this

135

dependence is known and is simple.13 Hence, anestimate of the yield of the given explosion can bemade by comparing the two sets of radius versustime data. This algorithm can make use of dataoutside the insensitive interval and works well ifthe ambient media at the two explosion sites aresufficiently similar. However, in practice it hassometimes proved difficult to ascertain whetherthe relevant properties of the media are similarenough to give the desired accuracy. Similar ex-plosion scaling has been proposed as a supplementto insensitive interval scaling for TTBT verifi-cation.

Semi-Analytical Modeling

Semi-analytical modeling is another approachthat is useful for studying the evolution of shockwaves in geologic media and for estimating yields.In this approach both the properties of the ambientmedium and the motion of the shock front aretreated in a simplified way that nevertheless in-cludes the most important effects. The result is arelatively simple, semi-analytical expression for theradius of the shock front as a function of time. Ifthe required properties of the ambient medium areknown and inserted in this expression, the yieldof an explosion can be estimated by fitting the ex-pression to measurements of the shock wave mo-tion with time.14 Semi-analytical algorithms can inprinciple make use of more of the data than canthe insensitive interval algorithm and can also beused to estimate the uncertainty in the yield causedby uncertainties in the properties of the ambientmedium.

Numerical Modeling

If a treatment that includes the details of theequation of state and other properties of the am-bient medium is required, or if the explosion isasymmetric, modeling of the motion of the shockfront using numerical hydrocodes may be neces-sary.15 In principle, such simulations can Provide

“See H. L. Brode, Annual Review of Nuclear Science, 18, 153-202(1968).

“For an early semi-analytical model, see M. Heusinkveld, ReportUCRL-52648 (Lawrence Livermore National Laboratory, Livermore, CA,1979). For improved semi-analytical models, see F. K. Lamb, ACDZSWP-87-2-1 (University of Illinois Program in Arms Control, Disarma-ment, and International Security, Urbana, IL, February 1987); W. C.Moss, Rep. UCRL-96430Rev. 1 (Lawrence Livermore National Labora-tory, Livermore, CA, July 1987); and R. A. Axford and D. D. Helm,Proc. NEDC (Los Alamos, October 1987).

15 For a discussion of numeric~ models currently in use ~ F. K. L~b,

“Monitoring Yields of Underground Nuclear Tests, ” Threshold TestBan Treaty and Peaceful Nuclear Explosions Treaty, Hearings Beforethe Comnu”ttee on Foreign Relations, Um”ted States Senate (Washing-ton, DC: U.S. Government Printing Office, 1987), pp. 359-370.

radius versus time curves that extend over muchof the shock wave evolution, making it possible tobase yield estimates not only on data from the tran-sition interval but also data from later phases ofthe shock wave evolution. In practice, the yieldestimates obtained using such a procedure arefairly sensitive to the equation of state of the am-bient medium, which is known with sufficient ac-curacy for only a few geologic media. If adequateequation of state data are lacking, numerical mod-eling may not be warranted.

In summary, the shockwave produced by an un-derground nuclear explosion propagates differ-ently in different media and different geologicalstructures. As a result, knowledge of the ambientmedium and local geological structures is requiredin order to make accurate yield estimates usinghydrodynamic methods. Several different yield-estimation algorithms have been developed. Thesealgorithms, like those based on seismic methods,involve some complexity and require sophistica-tion to understand and apply correctly. Some keyterms that have been introduced in this discussionare listed and explained in table A-3.

Application to Monitoring Treaties

Assuring Accuracy

Ambient Medium.-The physical properties andgeologic structure of the ambient medium enterdirectly into yield estimates based on hydro-dynamic methods. Incorrect assumptions aboutthe average properties of the ambient medium maybias the yield estimate, decreasing its accuracy,while small-scale variations will cause scatter inthe radius versus time data, decreasing the preci-sion of the yield estimate. Thus, it is important togather information about the types of rock presentat the test site and their properties, including theirchemical composition, bulk density, and degree ofliquid saturation, as well as the speed of sound inthe ambient medium and any specific features ofthe local geologic structure that could affect theyield estimate. Availability of the required datawould need to be assured by appropriate coopera-tive measures.

Some information about the geologic medium atthe test site could be obtained by examining thecontents of the hole drilled for the CORRTEX sens-ing cable. Verification could be improved by coop-erative arrangements that would also allow obser-vation of the construction of the emplacement hole,removal and examination of rock core or rock frag-

136

Table A-3.–Glossary of Hydrodynamic YieldEstimation Terms

Term/Explanation

Strong Shock Interval: The interval in radius and time duringwhich the speed of the shock wave is much greater thanthe speed of sound in the unshocked medium

Transition Interval: The interval in radius and time outsidethe strong shock interval in which the speed of the shockwave approaches the speed of sound in the unshockedmedium

Plastic Wave Interval: The interval in radius and time outsidethe transition region in which the speed of the weakeningshock wave is approximately the plastic wave speed

SLIFER Technique: A technique for measuring the positionof the shock wave expanding away from an undergroundexplosion by determining the resonant frequency of anelectrical circuit that includes a sensing cable placed ina hole in the ground near the site of the explosion

CORRTEX Technique: A technique for measuring the posi-tion of the shock wave expanding away from an under-ground explosion by determining the round-trip travel timeof electrical pulses sent down a sensing cable placed ina hole in the ground near the site of the explosion

Insensitive Interval Scaling: A yield estimation algorithm inwhich the Los Alamos Formula is fit to measurements ofthe position of the expanding shock wave as a functionof time during the algorithmic interval

Los Alamos Formula: The empirical formula used in the in-sensitive interval scaling algorithm to make yield estimatesby fitting to shock radius versus time data

Algorithmic Interval: The special time interval used in insen-sitive interval scaling during which the radius of the shockwave is relatively insensitive to the ambient medium; usu-ally assumed to be 0,1-0.5 scaled milliseconds after thebeginning of the explosion

Similar Explosion Scaling: A yield estimation algorithm inwhich data obtained from a previous explosion in the samemedium are scaled to fit measurements of the positionof the expanding shock produced by the explosion underconsideration

SOURCE: F.K. Lamb “Hydrodynamic Methods of Yield Estimation” Report forthe US. Congress, Office of Technology Assessment, Feb. 15, 1988,

ments from the wall of the emplacement hole, ex-amination of any logs or drill core from existingexploratory holes, removal and examination of rockcore or rock fragments from the walls of existingexploratory holes, and if necessary, constructionof new exploratory holes.

There is precedent for such cooperative arrange-ments in the Peaceful Nuclear Explosions Treaty(PNET), which explicitly established the hydrody-namic method as one of the monitoring methodsthat could be used for large salvos and specifiedverification measures like these.16

Test Geometry.–CORRTEX data must betaken very close to the center of the explosion in

1aArms control md Disarmament Agreements (U.S. Arms Controland Disarmament Agency, Washington, DC, 1982).

order to cover the insensitive interval (see tableA-2). As a result, yield estimates can be affectedby the arrangement of the nuclear charge and thecanister or canisters containing it and the diagnos-tic equipment.17 In particular, any properties of theexperimental set-up or the surrounding geologicmedia that cause the shock front to be distortedat the radii of interest could affect the accuracyof the yield estimate. The reason is that a CORR-TEX sensing cable measures only the depth of theshallowest point where a pressure wave firstcrushes it, at a single lateral displacement fromthe explosion. Thus, unambiguous interpretationof the data may become difficult or impossible ifthe explosion is not spherically symmetric.

For example, explosions of nuclear charges intunnels may be accompanied by complicated (andunanticipated) energy flows and complex shockwave patterns. If significant energy reaches thesensing cable ahead of the ground shock and shortsit before the ground shock arrives, the CORRTEXdata will describe that flow of energy and not themotion of the ground shock. Alternatively, the mo-tion of the ground shock itself could be sufficientlydistorted that interpretation of the shock positiondata becomes ambiguous or misleading. Asanother example, a large canister or double explo-sion could short the CORRTEX cable in such awaythat only part of the total yield is sensed over mostof the interval sampled by the CORRTEX cable,as shown in figure A-5. The physical size of

lrThe imwrt~ce of these disturbing effects is less for high-yield thanfor low-yield explosions.

Figure A-5.—Effect of Nuclear Test Design onShock Wave Radius Measurements Using CORRTEX

Equipment

CORRTEXRecorder

SOURCE: F.K. Lamb, “Hydrodynamic Methods of Yield Estimation,” Report forthe U.S. Congress, Office of Technology Assessment, Feb. 15, 1988.

137

canisters and diagnostic lines-of-sight tend to posemore of a problem for nuclear directed-energyweapons than for traditional nuclear weapons.18

In using hydrodynamic methods to estimate theyields of one’s own tests, the design and placementof the nuclear charge and related equipment areknown and can be taken into account. This is notnecessarily the case when monitoring the nucleartests of another party. Cooperative agreements tomake possible optimal placement of sensing cablesand to exclude nuclear test geometries that wouldsignificantly disturb the yield estimate wouldtherefore be required.19

Such agreements could, for example, limit thelength of the canister containing the nuclear chargeand the cross-sectional dimensions of the emplace-ment hole, and mandate filling of the nuclearcharge emplacement hole with certain types of ma-terials. Such agreements could also provide for ob-servation of the emplacement of the nuclear chargeand the stemming of the emplacement hole, con-firmation of the depth of emplacement, and limi-tations on the placement of cables or other equip-ment that might interfere with the CORRTEXmeasurement. For test geometries that includeancillary shafts, drafts, or other cavities, additionalmeasures, such as placement of several sensing ca-bles around the weapon emplacement point, maybe required to assure an accurate yield estimate.For tunnel shots, sensing cables could be placedin the tunnel walls or in a special hole drilled towardthe tunnel from above. Again, there is precedentfor such cooperative measures in the PNET.20

The restrictions on the size of canisters and diag-nostic lines-of-sight that would be required evenwith the sensing cable placed in a satellite holewould cause some interference with the U.S. nu-clear testing program at NTS. However, these re-strictions have been examined in detail by the U.S.nuclear weapon design laboratories and the De-partment of Energy, and have been found to bemanageable for the weapon tests that are plannedfor the next several years. In assessing whetherhydrodynamic methods should be used beyond thisperiod, the disadvantages of the test restrictionsmust be weighed against the potential contribu-

‘8 See Sylvester R. Foley, Jr., Assistant Secretary for Defense Pro-grams, Department of Energy, letter to Edward J. Markey, Congress-man from Massachusetts, Mar. 23, 1987.

1’S. S. Hecker, Threshold Test Ban Treaty and Peaceful Nuclear Ex-plosions Treaty, Hearings Before the Comnu”ttee on Foreign Relations,United States Senate (Washington, DC: U.S. Government Printing Of-fice, 1987), pp. 50-61 and 226-235; M. D. Nordyke, Ibid., 67-71 and 278-285; Foley, ibid

200p. cit., footnote 16.

tion to treaty monitoring made by these methods.In summary, the accuracy that could be achieved

using hydrodynamic methods to estimate theyields of underground nuclear explosions dependson the amount of information that can be gatheredabout the medium in which the explosion occurs,and the nature and extent of cooperative arrange-ments that can be negotiated to optimize the place-ment of sensing cables and to limit disturbingeffects.

Hydrodynamic methods for estimating theyields have not yet been studied as thoroughly oras widely as the seismic methods currently inuse, although they have been examined morethoroughly than some seismic methods that havebeen proposed for the future. Tests and simulationsto identify troublesome configurations have beencarried out, but only a few explosions have beenmonitored with the CORRTEX sensing cable in asatellite hole.21 Given the possibility thathydrodynamic yield estimation may have to beused to monitor treaty compliance in an adversar-ial atmosphere, the possibility of deliberate effortsto introduce error or ambiguity, and the tendencyfor worst-case interpretations to prevail, additionalresearch to reduce further the chances of confu-sion, ambiguity, spoofing, or data denial would bevery useful.

Minimizing Intrusion

Hydrodynamic yield estimation methods aremore intrusive than remote seismic methods forseveral reasons:

1. Personnel from the monitoring country wouldbe present at the test site of the testing coun-try for perhaps 10 weeks or so before as wellas during each test, and would therefore havean opportunity to observe test preparations.The presence of these personnel would posesome operational security problems.22

2. The exterior of the canister or canisters con-taining the nuclear charge and diagnostic

equipment must be examined to verify thatthe restrictions necessary for the yield esti-mate to be valid are satisfied. For tests of nu-clear directed energy weapons, this examina-tion could reveal sensitive design informationunless special procedures are followed.23

*’Op. cit., footnote 8. Approximately 100 tests have been carried outwith the CORRTEX cable in the emplacement hole, and SLIFER datafrom satellite holes are available for several tens of earlier explosions.

IZR. E, Batzel, Threshold Test Ilan . . ., op. cit., footnote 19, PP. 48+0and 210-225, and Foley, op. cit., footnote 18.

21 Batzel, ibid.

138

3. Sensing cables and electrical equipment willtend to pick up the electromagnetic pulse(EMP) generated by the explosion. A detailedanalysis of the EMP would reveal sensitive in-formation about the design and performanceof the nuclear device being tested.

Intrusiveness could be minimized by careful at-tention to monitoring procedures and equipment.For example, the electrical equipment required canbe designed to avoid measuring sensitive informa-tion about the nuclear devices being tested. CORR-TEX equipment has been designed in this way, andthe United States could insist that any Sovietequipment used at NTS be similarly designed. Thesecurity problems posed by opportunities to ob-serve test preparations are more severe for nucleardirected energy weapon tests, since they tend tohave more and larger complex diagnostic systemsand canister arrangements which, if fully revealedto the Soviets, might disclose sensitive informa-tion. The United States has determined that theSoviet personnel and activities that would be re-quired at NTS to monitor U.S. tests would beacceptable both from a security standpoint andfrom the standpoint of their effect on the U.S. testprogram. Detailed operational plans have been de-veloped to accommodate such visits without ad-verse impact on operations.24

Specific Applications

Threshold Test Ban Treaty.–As noted earlier,hydrodynamic yield estimation has been proposedby the Reagan administration as a new routinemeasure for monitoring the sizes of nuclear tests,in order to verify compliance with the 150 kt limitof the TTBT. To reduce the cost and intrusivenessof such verification, it could be restricted to testswith expected yields greater than some thresholdthat is an appreciable fraction of 150 kt. Hydro-dynamic measurement of the yields of one or morenuclear explosions at each country’s test site orsites has also been suggested as a method ofcalibrating seismic yield estimation methods.25

24 FOley, Op. cit., footnote 18-ZfiIt has ban sugge9ted that the nuclear calibration chwges to be det-

onated at the test sites could be provided by the monitoring country.If they were, a hydrodynamic estimate of the yield might not be needed,since, as explained inch. 7, the yields of certain types of nuclear chargesare accurately reproducible. Knowledge of the surrounding medium andgeologic structure would still be needed to provide assurance that thecoupling of the explosion to seismic waves was understood. However,some way would have to be found to provide assurance that no sensi-tive information about the design of the nuclear weapons of either coun-try would be revealed.

From the point of view of the United States, pos-sible advantages of being able to use hydrodynamicyield estimation methods at Soviet test sites in-clude the additional information on yields that thiswould provide, establishment of the principle ofon-site inspection at nuclear test sites, and the pos-sibility of collecting data on the ambient media andgeologic structures at Soviet test sites. Obviously,the larger the number of explosions and the greaterthe number of test sites monitored, the more in-formation that would be obtained. Possible dis-advantages for the United States include the po-tential difficulty of negotiating routine use ofhydrodynamic methods at Soviet test sites, whichcould impede progress in limiting nuclear testing,and the operational security problems at NTScaused by the presence of Soviet monitoring per-sonnel there.

Peaceful Nuclear Explosions Treaty.–As itstands, the PNET does not provide for use ofhydrodynamic yield estimation except for salvosin which the “planned aggregate yield” is greaterthan 150 kt.26 Thus, if the TTBT is modified to al-low hydrodynamic yield estimation for all weapontests with planned yields above a certain value, thepurpose of the modification could in principal becircumvented by carrying out weapon tests as“peaceful” nuclear explosions of “planned yield”less than or equal to 150 kt, unless the PNET isalso modified to close this loophole.

Low-Threshold Test Ban Treaty. -Undergroundnuclear explosions as small as 1 kt produce shockwaves that evolve in the same way as thoseproduced by explosions of larger yield. However,such explosions can and usually are set off at shal-low depths and can be set off in alluvium. As a re-sult, the motion of the ground can be markedlydifferent from that on which standard hydro-dynamic yield estimation methods are based, caus-ing a substantial error in the yield estimate (seefigure A-4). There can also be significant variationsin the motion of the ground shock from explosionto explosion under these conditions.

In addition, serious practical, operational, andengineering problems arise in trying to usehydrodynamic methods to estimate the yield ofsuch a small explosion. For one thing, the sensingcable must be placed very close to the nuclearcharge (see table A-3). Drilling a satellite holewithin 3 meters of the emplacement hole to thedepths of typical nuclear device emplacement, aswould be required in order to use hydrodynamic

‘eArms . . . , op. cit., footnote 16.

methods to estimate the size of an explosion witha yield near 1 kt, would be challenging, to say theleast. The need for such close placement would ne-cessitate further restrictions on the maximum sizeand orientation of the canister used to contain thenuclear charge and diagnostic instrumentation.Such restrictions might be deemed an unaccept-able interference with test programs. However, useof small canisters with numerous diagnostic lines-of-sight to the detonation point could disturb theCORRTEX measurements. Because the shockwave radii to be measured are much smaller at lowyields, survey errors become much more important.

Possible solutions to these problems have notyet been carefully and thoroughly studied. Thus,

139

at the present time hydrodynamic yield estimationmethods could not be used with confidence to mon-itor compliance with threshold test bans in whichthe threshold is less than several tens of kilotons.

Comprehensive Test Ban.–As their name im-plies, hydrodynamic yield estimation methodshave been developed to measure the sizes of un-derground nuclear explosions. They are a poten-tially valuable component of a cooperative programto monitor limits on yields, but are neither intendednor able to detect, identify, or measure the yieldsof unannounced or clandestine nuclear tests. Thus,they are not applicable to monitoring a comprehen-sive test ban.

o