Proof readingsamplejobjungle

743 S(~IE"I'I Fie All E R[(~Al

OFFPRINTS

The Verification of a Comprehensive Nuclear Test
Ban
Sf 1 II:NTI 1-"1(: :1'1 I': H I f.. ~
~X-::TC6ER
.•.••• , ... ~ .• I I:" .•••

l·~j:~· ~1 .~. E::::.z b" •• :::.: •• ill -i1i:". ~ IIH''''':'' II •.•. lI.. i.:. II::. PI::::.t1" .• ':IJ -:1 il".I:1~ rl UII::I •• 1.5.! •. N:J :J--""I :JI !I" ':: ""'I!-=ril'. "11::1:- 1.:1::1 '~:J' :J:Jl.-l.:tI:J :.1:-' ::III'" 1I''':L.:'I::Inll::']!. r;1"r.-:) IJr=1r.I"Ir. r.f 1:18.:~."~I"JI':. j:~~~ l"1' ill I'I!?' "(I' 'II -=,: ::I j.:h: .•.. c,::r::JP"'I"= ~:':o:r.r!"!
I

ir.:.. rr.r 1n::J:o' it •.•... !".1r.rr.r. Ir :"J ".::1nr.'):'I1 .~:o':':t.::-', f1',1ra.'irr:ltt.YI ••••. r.t ••• i:r •••• ·:W: •... u4-J rl~ :JIJI..I·~ "':1 :.111'.'.:111:: U!.J.: ••.• I'IUUI w'II.L!I: I.IL!IIIII!-Z:I:I'I I:': I'll! F,JI:II!-.t.:::p. 11.1-1 Ir:II:~l:rr:ul-: .::,:1 1r:.lI;Er-~r"lil ""S(" ~ "lJ'T1r-1r:: J.~I-F!I(,:":'I~ ::Ir:(! ~hl;l .:II::II'II;!'!IVEI
I~~~ l,:'~ :.!!!!-I::llr:r;:: 1t.:: ~(! .1ri'! -hr. :",:'1 i': !':Ii.r...:.- ••.. "iI ':'1.1· ,"4 I H,o~ ••• 14· •.•• 1 III .1-0;1" 1"11;1 11:-11111;' ~..( "'''I,.,,",,=, J'=":"I.~ III:::rl:'il: ,111""l::f ""I;"II':=':' IU.iIlI. ~:: I::r lif r ~·lll;"ri:i:Ir. In.:.

The Verification of a Comprehensive Nuclear
Test Ban
Networks of seismic instruments could monitor a total test ban with high reliability. Even
small clandestine explosions could be identified even if extreme measures were taken to evade
detection
Two treaties put into effect over the past 20 years have set limits on the testing of nuclear weapons. The Limited Test Ban Treaty of 1963, which has been signed by
more than 120 nations, prohibits nuclear explosions in the atmosphere, the oceans and space, allowing them only underground. The Threshold Test Ban Treaty of 1976,
a bilateral agreement between the U.S. and the U.S.S.R., prohibits underground tests of nuclear weapom with a yield greater than 150 kilotons. In the present climate of
widespread pressure for more effective control of nuclear arms the idea of a comprehensive ban on all nuclear testing is receiving renewed attention. Such an agreement
would be an important measure. It might inhibit the development of new weapons by the major nuclear powers, and it might also help to prevent the spread of nuc1ear-
weapons technology to other countries.
A halt to all testing was the original goal of the negotiations that led to the 1963 Limited Test Ban. New talks with the aim of aChieving a total ban were begun in 1977
by the U.S., the U.S.S.R. and Britain, but the talks were suspended in 1980. In both cases the main impediment to a comprehensive treaty was the contention by the U.S.
and Britain that compliance with the treaty could not be verified becauSl; sufficiently small underground nuclear explosions could not be reliably detected and identified.
In July the Reagan Administration announced that the test-ban negotiations with the U.S.S.R. and Britain will not be resumed. Once again the primary reason given was
a lack of confidence in methods of verifying compliance.
by Lynn R. Sykes and Jack F. Evernden
In 1963 the reliability of measures for the verification of a treaty banning expLosions larger than about one kiloton may have been arguabLe, but it no longer is. We
address this question as seismologists who have been concerned for many years with the detection of underground expLosions by seismic methods and with means of
distinguishing underground explosions from earthquakes. We are certain that the state of knowledge of seismoLogy and the techniques for monitoring seismic waves are
sufficient to ensure that a feasible seismic network could soon detect a clandestine underground testing program involving explosions as sman as one kiloton. In short,
the technical capabilities needed to police a comprehensive test ban d'own to explosions of very small size unquestionabLy exist; the issues to be resolved are political.
An underground explosion sets up .fi elastic vibrations that propagate as seismic waves through the earth and along its surface. The waves travel great distances, and
seismic monitoring instruments in common use are sensitive enough to record even those generated by very small explosions. Once the waVeS have been detected the
main task is to distinguish the seismic signals of explosions from those of earthquakes. This can be done with a network of several widely separated seismometers.
Two types of elastic vibrations can propagate through the solid body of the earth, that is. through the crust and the mantle. The first waves to arrive at a seismometer are
compres.~ional waves, which are similar to sound waves in
air or water; the seismologicaL name for them is P (for primary) waves. The slower body vibrations are shear waves, whit:h are similar to the waves on a vibrating
string; they are catted S (for shear or secondary) waves. An underground explosion is a source of nearly pure P waves because it applies a uniform pressure to the walts
of the cavity it creates. An earthquake, on the other hand, is generated when two blocks of the earth's crust rapidly slide past each other along the plane of a fault.
Because of this shearing motion an earthquake radiates predominantly S waves.
A result of the spherical symmetry of the explosion source is that all the seismic waves it generates have a nearLy radial symmetry around the focus of the explosion. II).
contrast, the highly directional character of an earthquake source gives rise to seismic waves with strongly asymmetric patterns, The asymmetry in the amplitude of the
waves received at seismometers throughout the world provides the means whereby seismologists can determine the fauLting mechanism of a given earthquake.
In addition to the P and S body waves there are also two types of seismic waves that propagate only over the surface of the earth. They are called Rayleigh waves and
Love waves, and they result from complex reflections of part of the body-wave energy in the upper layers of the earth's crust. A simple explosion can generate Rayleigh
waves but not Love waves, whereas an earthquake generates waves of both types.
Seismologists characterize the size of a seismic event by means of magnitudes. A given event can be assigned several
3

U.S.

60

50

40

30

20

10

0

10

20

30

40

U.S.S.R.

50
a 40
:
~

>
-
a
:
w 30

c
.
.
C
Il
z

0 20

(
/
j
0

-
'
Q
. 10
x

w

a
:
- 0
0

:
W

-
'
C
)
::
:: 10
l
Z

IJ
..
0

0 20
:
w

<
II
:
:
;
::
:: 30
l
z

FRANCE

10

0

10

20

U.K.

':1 r---f'N Cl:J1i1
CHINA

::I
INDIA

J 19
1945 1950 1955
60

J~
,

1965
,

1970
1980
1975
magnitudes, each one based on a different kind of seismic wave. A magnitude is the logarithm of the ampliwde of a particular type of wave normalized for distance and depth of
focus. Of the numerous magnitudes that can be defined for a single seismic event we shall discuss only two, which in seismological notation are designated Ms and mb' The fonner
is generally based on Rayleigh waves with a period of 20 seconds, the latter on one-second P waves. The magnitude of a seismic signal is ultimately related to the energy released
at the site of the event. For a nuclear explosion the customary measure of energy release is the yield in kilotons, where one kiloton is the energy released by detonating 1,000 tons
of TNT.
Every year there are numerous earthquakes whose magnitudes are in the range corresponding to the yields of underground explosions. Several methods can be applied to several
types of waves to distinguish the seismic waves of explosions from those of earthquakes. The location of a seismic event and its depth below the surface are important criteria;
indeed. the great majority of routinely detected events can be classified as earthquakes simply because they are either too deep or not at a plausible site for an explosion_ The
remaining events can be reliably classified by the amount of energy radiated in the several kinds of waves at various frequencies.
The location of an event in latitude and longitude is a powerful tool for classification. The position is determined by recording the arrival time of short-period P waves at sever.u
seismographic Slations in various parts of the world. The travel time of the P waves to each station is a function of distance and depth of focus. From the arrival times it is possible
to determine the location of the source with all absolute error of less than 10 to 25 kilometers if the seismic data are of high quality.
The identification of seismic events at sea is quile simple. It is assumed that the network monitoring a test-ban treaty would include a small number of simple hydroacoustic
stations around the shores of the oceans and on a few critical islands to measure pressure waves in seawater. The hydroacoustic signal of an underwater explosion is so different
NUCLEAR TESTS CONTINUE to be carrie<! out at a rate of about 50 per year, principally by tbe two leading nudear-weapons powers: the U.S. and the U.S.S.R. As tbis bar chari sbows, the main
effect of the Limite<! Test Ban Treaty of 1963 (broken yerti.:alline) was not 10 reduce the number of test exptosious but merely to drive most of them underground. Nuclear test e:o:plosions in the :lImos-
phere and underwater are represented by colored bars, tbose und erground by grdY bars.

from that of an earthquake and can be detected at such long range that the identification of a seismic event at sea as an explosion or an earthquake is simplc anrl positive.
Hence any event whose calculated position is at least 25 kilometers at sea (a margin allowing for errors) can be classified as an earthq uake un the basis of its location and
the character of its hydroacoustic signal.
The accuracy with which the position of a seismic cvent can be determined in an area offshore of an island arc has been tested with an array of ocean-bottom seismometers
off the Kamchatka Peninsula and the Kurile Islands in the U.S.S.R. The tests indicate that the accuracy of a seismic network under these circumstances is much better than
25 kilometers. Holding to that standard, however, one finds that well over half of the world's seismic events are definitely at sea and are therefore easily identified as
earthq uakes.
Another large group of detected events have their epicenters on land but in regions where no nuclear explosions are to be expected; these events too can be safely
classified as earthquakes. Indeed. almost aU the world's seismic activity is in regions tha t are of no concern for monitoring compliance with a comprehensive test ban.
Thus the simple act of locating seismic events classifies most of them as earthquakes.
Calculating the depth of focus provides a means oC identifying a large fraction of the remaining earthquakes. From 55 to 60 percent of the world's earthquakes are at depths
of more than 30 kilometers; at least 90 percent are more than 10 kilometers deep. Any seismic event as deep as 15 kilometers is certainly an earthquake. No one has yet
drilled into the earth's crust as far down as 10 kilometers, and the deepest nuclear explosions have been at a depth of about two kilometers.
Several seismological procedures ean be employed to determine an event's depth of focus. In most cases the depth is calculated at the same time as the location. When a
seismic event is detected at 20 stations or more, a routinely calculated depth of 30 kilometers or more ensures with a 95 percent degree of confidence that the event was at
[east 15 kilometers below the surface.
A powerful technique for estimating depth can be applied if at least one seismological station is within a few hundred kilometers of the detected event. (A monitoring
network for a comprehensive test ban would be quite likely to meet this condition in areas where nuclear testing might be expected.) A pair of P and S waves generated at
the same instant and recorded by a station near the event follow identical paths but propagate at different speeds. The difference in their times of arrival, or in
P WAVE
I .
I I I
I I
- I _
S WAVE

LOVE WAVE

RAYLEIGH WAVE

FOUR TYPES OF SEISMIC WAVE are illustrated. The two wave:; at the top propagate througb tbe solid body of the eartb; the two at the bottom propagate only nc"r the surface. The compressional
body waves called P (for primary) waves travel fastest Ilnd are the first On,'S to an;ve lit a seismometer; they arc the predominant type of body wave produced by all underground explosion. The
slower body waves called S (for shear or secondary) wav"" vibrate in a plane transverse to their direction of propagation; they are tbe predominant type of body wave produced hy an earthquake. The
surface waves called Rayleigh wan'S and love waves result from complex refleetiolls uf the P and S body waves in the upper layers of the eartb's eru..t.
other words the difference in their phases, therefore serves to determine the time of origin of the event. With experience the seismograms of a station near the event can be
successfully analyzed to detect at least one pair of such Pand S phases. Given the time of origin determined in this way and the arrival times of the P waves at only a few
distant receivers, an accurate estimate of the depth of focus can be made.
There may remain critical seismic regions whcre nearby stations do not cxist. Data from large events can then be employed to refine the calculated depth and location of
smaller events. The essence of the techniquc is to correct the observed times of small events by noting the differences between the observed and the calculated times f~r a
large event in the same area. The procedure is in routine use by several networks.
The combined effectiveness of location and depth in distinguishing earth-
quakes from explosions is impressivc. More than 90 percent of all earthquakes either are under oceans or are at [east 30 kilometers deep (or both). Most of the remaining
earthquakes are of little interest because they arc in countries that are unlikely to be testing nuclear weapons or in cOtmtries where clandestine testing would be
impossible. For the U.S., of course, the U.S.S.R. is the country of prime interest. About 75 percent of the earthquakes in and near the U.S.S.R. are in the eastern part of thc
country near the Kamchatka Pcninsula and the Kurile Islands. Almost all of the shocks in thcse arcas either have a focal depth greater than 50 kilometers or are well
ofTshore. It turns out that seismic events whose calculated position is on land in the U.S.S.R. or less than 25 kilometers at sea and whose calculated depth is less than 50
kilometers constitute only about .5 percent of the world's earthquakes. This amounts to about lOa earthquakes
5

per year with an mb magnitude greater than 3.8 for which other seismic discriminant~ mu~t be employed.
None of the measures we have discussed so far relies on the detailed characteristics of the waves radiated by earthqlJakes and explosions. Several powerful discriminants are based
on those characteristics, in particular on the relative amounts of energy in waves of different types and periods. For example, a shallow earthquake generates 20-second Rayleigh
waves with amplitudes at least several times greater than those of an explosion that releases the same amount of energy. In the notational practice of seismology the comparison of
the two magnitudes is referred to as the Ms: mb ratio, that is, the ratio of long-period to short-period waves.
Asecond spectral discriminant is . based on the observation that longperiod p and S waves are rarely or never seen in association with explosions but one type or the other is
routinely detected today by simple seismometers for most earthquakes that have a onesecond P-wave magnitude of at least 4.5. More sophisticated seismic stations and more
sophisticated analysis of the signals could lower the magnitude at which such waves can be detected.
A third distinction is that surface waves of the Love type are generated far more strongly by shallow earthquakes than they are by underground explosions, including even
abnormal explosions. Still another characteristic feature of the seismic signal from explosions is that the first motion of the earth stimulated by P waves is always upward because
the explosion itself is directed
outward; the first P-wave motion in an earthquake can be either upward or downward.
An important factor contributing to the separation of earthquakes from explosions on an Ms: mb diagram is that P waves from the two kinds of events have different radiation
patterns. Explosions radiate short-period P waves equally in all directions, whereas earthquakes have very asymmetric patterns. Hence most earthquake sources show a decrease of
from .4 to one magnitude unit from the peak values when the P-wave amplitudes are averaged over pertinent radiation angles. A simple explosion doe~ not initially radiate any
shear waves; earthquakes typically generate large shear waves. As a result Rayleigh waves generated by many types of earthq uakes have a larger amplitude than the corresponding
waves generated by underground explosions of the same mb .
There is a characteristic time for the formation of the source of a seismic event; the time is equal to the maximum source dimension divided by the velocity of source formation.
The source dimension for earthquakes is the length of the break where most of the short-period energy is released; it is from three to 20 times greater, depending on the state of
stress in the rocks, than the radius of the cavity and shatter 'lOne of a comparable explosion. The velocity of source formation for earthquakes is from somewhat less to much less
than the velocity of shear waves in the rocks surrounding the fault, whereas the relevant velocity for explosions is the velocity of shock waves in the rock, which is essentially the
velocity of compressional waves. As a result

PATHS OF SEISMIC WAVES are t •.•• ced all a cross section of the earth. Body w~ves from an earthquake or an explosion travel tbrough the crust and mantle along the curved paths labcl...d P, S, pP and
pS. A pP wave is a compressional wave Ibat is produc...d by the reflection of a P wave from Ihe surface of Ibe earth just above an eartbquake Or an explosion; a pS wave is a sbelll' wave th~t results from the
conversion of part of tbe compressional energy of an upward P WaH into transverse energy :IS tbe P wave is reflected from tbe surface. Surface waves sucb iIli Rayleigh wav •..• and Love waves diminish
rapidly in amplitude with incrensing depth. The hypocenter is Ibe focal point of an e'u1:hquake or an undergronnd explosion from whicb tbe waves radiate. The epiccnler is tbe point nn the earth's surface
directly over the bypocenter.
6
of these differences in the size of the source and the velocity of source formation the characteristic times for earthquakes and explosions differ by a factor of from six to 40. It is
therefore not surprising that differences are observed between the short-period P-wave spectra of earthquakes and explosions.
Observations of several U.S. explosions have demonstrated the existence of a phenomenon called overshoot. It is related to .shock waves in strong rock, but it can be thought of as
the equivalent of cavity pressure rising to high values followed by a decrease in pressure by a factor of four or five; the lower pressure is thcn maintained for many tens of seconds.
Overshoot, when it occurs, provides additional P-wave spectral discrimination and augments discrimination by means of the Ms: fib ratio for larger events.
It was once thought that an explosion could not give rise to any Love waves at all. A phenomenon that was of great significance in thwarting President Kennedy's effort to achieve
a comprehensive test-ban treaty in 1963 was the observation that many underground nuclear explosions at the U.S. testing site in Nevada, particularly those in hard rock, generated
unmistakable Love waves. The failure of the qualitative criterion "No Love waves from explosions" (at a time when such quantitative criteria as the comparison of the magnitudes
of long-period and short-period waves were not adequately established) left seismologists unable to guarantee their ability to distinguish the seismic waves of underground
explosions from those of earthq uakes.
The presence of Love waves in the 1'1 evada tests has since been explained. What was not considered in the earlier analyses was the influence of the natural stressed state of the
earth on the waves generated by an explosion. The creation of a cavity and ils surrounding shatter cone by an underground explosion leads to the release of some of the natural
stress, which in turn generates seismic waves equivalent to those of a small earthquake, including Love waves. The observed waves are a superposition of the waves from the
explosion and from the release of the stress.
The release of natural stress also alters the amplilude of Rayleigh waves. The perturbation has never - been large enough, however, to put in doubt the nature of an event identified
by the ratio of long-period to shorr-period wav.es. Only rarely does the perturbation significantly affect the amplitude of P waves; it is not known ever to have changed the
direction of their first motion. Moreover, if the magnitude Ms is determined from Love waves ralher than Rayleigh waves, the ratio method (Ms: mb) provides an excellent
discriminant.

RADIATION PATTERNS of the P waves resulting frolll an ulldergronnd explosion (le]t) alld all earthquake (rig/It) arc compared. Tbe first motion of the P Waves [rolD an explosion is uniformly
outward and hence is gener.J.J]y obsr:rved as an upward tljsplai.:cmcnt at all.seismie stati •• n •. The first motion of the P waves frolll all earthquake is outward in some directions ::md inward in others; the
pattern o( the waves at tbe surface dep~nds Oll the orientation of the plane of tile
[n short, if seismologists had done their homework thoroughly by ]963, the nations of the world might well have achieved a comprehensive test-ban treaty then. Today the release
of natura] stresses in the earth is significant only as a perturbing factor that must be taken into account when the yield of an explo· sion is estimated from Rayleigh waves.
Reports that earthquakes occasiona]ly have Ms: mb values like those of e)(plosions have been cited as a factor that might impede the effective monitoring of a comprehensive test
ban. In analyzing a large se t of earthq uakes in all parts of the wor~d and of underground explosions in the U.S. and the U.S.S.R. we found only one example of this kind of
ambiguity. The focus of the event was far from the area in which the seismometer network gave its best results.
In 1972, at a meeting of the UN Committee on Disarmament, the U.S. submitted a list of 25 "anomalous" events that were said to be indicative of a proble.m in discrimination. In
1976 the 25 events were reanalyzed by one of us (Sykes) and two other seismologists, Robert Tathum and Donald Forsythe. It was established that about half of the events hadMs:
mb values that put them clearly in the earthquake population. Most of the original magnitudes had been determined from only one or two stations, and much existing information
had not even been consulted. When
LOVE WAVE
RAYLEIGH WAVE

.arlh'luake fault. In the compar:l.tivcly simple caSe of" vertical strike. slip fault, shown here, Ihe four-lobed radiatinn pattern ohserved at the surface for both P waves and Rayleigh waves is a simple
projection of the tbree..dimcnsional P-w:lve configuration. elnanating from the hypocenter of tbe e:ntbquake. The radiation pattern of tbe Love waves emitted by the .ame 'ouree is rotated by 45 degrees with
respert to the surface pattern of tbe l' waves and the Rayleigh waves.
the records of other available stations were examined, the events ceased to be "anomalous."
For the remaining problem events Ms: mb measurements based on 20-second Rayleigh waves gave values in the range characteristic of explosions. Sev. eral of these events were
at depths of from 25 to SO kilometers, where the possibility of nuclear testing can be excluded in any case, but the magnitude ratio nonetheless demanded explanation. It is known
from seismological theory that certain types of earthq uakes at these depths excite Long-period Rayleigh waves poorly. The theory predicts, however, that Love waves and
vibrations caJled higher-mode Rayleigh waves are in many instances vigorously generated in these circumstances. An analysis of recordings of the Love waves and the higher-
mode Rayleigh waves identified several more of the problem events as earthq uakes.
Only a single sequence of events at one place in Tibet remained as a problem. In that region underground nuclear testing is unlikely, but the nature of the events could not be
determined with certainty from the magnitude ratios. We think the reason is that with the seismographic networks of the 1960's, when the events were recorded, Love waves could
not be detected for small-magnitude events because they were obscured by background earth noise. New instal-
lations and new modes of data processing have greatly reduced the prob.Lem. If the same series of events or a similar series were to take place today, we think they would be
identified unambiguously. Long-period seismographs in bore. holes and routine digital processing of seismograms lead to a suppression of backgro und noise and increase the de-
tectability of many types of waves, including Love waves.
As it happened, the nature of the Ti·· betan problem sequence was resolved in spite of the inadequacies of the long-period data of the time. At several stations the first motion of the
P waves was downward, which is not possible for an e)(plosion. Hence the events must have been small earthquakes.
It seems reasonable to say that for the networks we shall describe below there should no longer be any problem events at mb 4 or more. We know of no Eurasian earthquake with a
one-second P_ wave magnitude of 4 or more in the past 20 years whose waves are classified as those of an explosion. (Of course, numerous smaller Eurasian earthquakes during
that period went unidentified because of inadequate data.) Furthermore, to our knowledge not one out of several hundred underground nuclear explosions set off in the same period
radiated seismic waves that could be mistaken for those of an earthquake. Our experience indicates an e)(tremely low proba-

bility that an event will remain unidentified when all the available tt::hniques of discrimination arc brought to bear.
No morritoring techrrology can offer an absolute assurance that even the smallest illicit explosion would be dete:ted. We presume that an ability to detect and identify eveuts
whose seismic magnitude is equivalent to an explosive yield of about one kiloton would be adequate. It is often assumed that for the U.S. to subscribe to a comprehensive test ban
it would require 90 percent confidence of detecting any violation by another party to the treaty. Developing a new nuclear weapon, however, generally req uires a series of tests,
and the pro bability that at least one explosion will be detected rises sharply as the number of the tests is increased. Moreover, a 90 percent level of confidence for the detection of
even a single explosion probably is nor needed. For a country seeking to evade the treaty the expected probability of detection would certainly have to be less than 30 percent, ahd
perhaps much less, even if only one illicit test were planned. The test-ban agreements that have been considered over the years all include an "escape clause" through which a
country could renounce its treaty obligations. Unless the probability of detection were very low, a country whose national interest seemed to demand a rt;~umption of testing
would presumably invoke thc escape clause rather than risk being caught cheating.
Given these standards of reliability for a monitoring system, it is possible to specify the size and the sensitivity of the
ARCTIC EARTHQUAKE SEPTEMBER 8, 1972 mb;5.9
seismic network that would be needed to verify compliance with a comprehensive test ban. Two kinds of network can be considered for maintaining seismic surveillance of the
U.S.S.R. One network consists of 15 station·s outside the borders of the U.S.S.R. In the second network the 15 external stations liTe supplemented by 15 internal ones.
The ultimate limit on the detection of seismic signals is imposed by microseisms, or random vibrations of the earth's surface. Most microseisms are induced by the earth's
atmosphere and oceans. [n order to detect a one-kiloton explosion in much of the U.S.S.R. a monitoring network would have to be able to recognize above the background noise
any event with a short-period Pwave magnitude of 3.8 or more. In order to distinguish an explosion from an earthquake by comparing the long-period magnitude with the short-
period one, . the network would also have to be able to detect surface waves with an Ms magnitude of 2.5 or more. The network-of 15 external stations could achieve these goals.
Indeed, since almost all the seismic areas of the U.S.S.R. are along its borders, the external network would be sensitive to events of even smaller magnitudc there. The mere
detection of a seismic event in most areas of the interior would constitute identification of the event as an explosion.
The lower limit of one kiloton on the yield of an explosion that could be detected by an external network is based on the assumption that the coupling between the explosion and
the seismic radiation is efficient and that the explosion

NUCLEAR EXPLOSION. U.S,S.A. AUGUST 28, 1972 mb;6.3

tup
------, ,... ..
l DOWN P

1 MINUTE

~ r-
was not set oU during or soon after a large earthquake: If one must consider the possibility that a country would try to evade a test-ban treaty by decoupling, or m ullling, an
explosion and thereby reducing the amplitude of the emitted seismic signals, an improved network would be required. In principle such illullling could be done by detonating the
explosion in a large cavity or by using energy-absorbing materia I in a smaller cavity. The former stratagem might reduce the seismic signal of an explosion by 1.9 magnitude units
as measured by one-second P waves (that is, by mb). The latter stratagem might bring a reduction of one unit.
The use of an oversize eavity is clearly the more worrisome possibility, but .it could be attempted only in cerrain geologic formatiom: a salt dome or a thick sequence of bedded
salt deposits. Few areas of the U_S.S.R_ have deposits of salt in which the construction of a cavity large enough for decoupling a severalkiloton explosion would be possible. The
maximum size of a cavity that could reasonably be constructcd and maintained sets a limit of two kilotons on explosions that might be illumed in this way and escape detection by
the IS-station external network.
Another way to reduce the amplitude of radiated seismic waves is by detonating an explosion in a low-coupling medium such as dry alluvium. The maximum thickness of dry
alluvium in the U.S.S.R. sets a limit of 10 kilotons on explosions that might be concealed by this means, again assuming that only the 15 external stations were installed.

SURFACE WAVES

SEISMOGRAMS OF LONG.PERIOD WAVES [rom an earthquake in the Arctic near tbe U.S.S.R. (l(Jp) aud an und.rgrouud nuclear explosion in tbe U.S.S.R. (bol/(Jm) were recorded at a seismic
,tation in Elath, Israel, roughly equidistant from the two ~vellts. The short-period body W3Ves generated by the two shocks were observed to have almost the same magnitude. The magnilude of the long-
period Raylcigh waVes recorded in these tr~ccs, in contrast, is c1elll'ly
8
much greater for the earthquake Ihan il is for the e:<plusion. The ra, liD of long-period surfac.e waves 10 .hon-period body waves hilS been shown to be a reliable criterion fur distinguishing the seismic
waves of ellrthquakcs from those or e:<l'lo.ions. In addition the P wIlve of the explosioo has more high-frequellcy .nergy Ih:lI1 the I' wave of the earthquake. The S waVe of lbe earthquake ;. lar~e,
whercllS thut of the ':<plo.sion is small and 1101 ea.;ly identifi.d ill the .eismogr:1m.

Another possible drawback of an exclusively external network should be mentioned. Confusion could arise when signals from two or more earthquakes reached a station simultaneously. The
effect would be most troublesome when the long-period waves from a small event in the U.S.S.R. arrived at the same time as similar waves from a much larger earthquake elsewhere in the
world . Under these circumstances it might be difficult to establish with certainty by comparing Ms with mb the identity of the event in the U.S.S.R. With a network of 15 external statioos
there would be a few events per year in which the smaller earthq uake was in the territory of the U.S.S.R. or within 25 kilometers of its borders and at a depth of less than 50 kilometers.
Amonitoring network made up of 15 seismographic stations outside the U.S.S.R. and 15 inside it would largely eliminate the problem of coincident earthq uake signals and would greatly
reduce the maximum yield of an explosion that might escape detection, even if decoupling were attempted. The internal monitoring stations would be simple unattended o[1es, with U1e
capability of measuring vertical ground motion and two orthogonal components of horizontal motion, so that the dista[1cc and direction' of a nearby event could be estimated from the data of
a single shortrange station. With such a network in place, and assuming that muffling was attempted in llie presence of normal earth noise, the largest explosion that would have a 30 percent
chance of escaping detection in any setting except a salt dome would be .5 kiloton.
For salt domes the main area of concern in the U.S.S.R. is the region north of the Caspian Sea. Our hypothetical network has three stations there. Even a small explosion in a large salt-dome
cavity would emit certain P and Swaves with an amplitude large enough to be detected by nearby stations. Furthermore, detection by even one of the stations would immediately identify the
event as an explosion because the area has no natural seismic activity. As a result evasio[1 would not be likely to be attempted al a yield greater than one kilotem even in the salt-dome area.
A possible strategy for evasion mat has been mentioned from time to time is the one of hiding the seismic signal of a nuclear explosion in the signal of a large earthquake, which might be near
the sile of the explosion or far from it. For the U.S.S.R. the only credible possibility is a distant earthquake because the only possible testing sites where earthquakes are frequent enough to
make the effort worth while are on the Kamchatka Peninsula and in llie Kurile Islands. Clandestine testing there iR nnt likely becHlse seismic activity in the area can be moni-

•
7 •
EARTHQUAKES

•••
• • • •
•
-;;; 6
•

6 •
• •
•• • •
(j)
•
•
w
• •• •

~ •
· .
.-
• •
•••
•
w
• • •• ••
• •
u • •• ••
0

•
Lt • ••• •
•
•• •
• •• • •
0 0

•
cr:
•
•• •
::>
(j)
• • • ••• 0

•••••••••
•
0
••• •••••
z

0
• •••••
u •••• 0

•••••
w
•
•
(j)
••••• 0

6
• 0
+
•
'" 4 • +
0

lJ...
• 0
0 0
0
•
0 • 0

W ++ ++
0 Eeffi
::> +
I- +
Z
+
(!l
«: EXPLOSIONS
:'Z 3
0
0

0

2

4 4.5 5 5.5 6 6.5
MAGNITUDE OF ONE·SECOND BODY WAVES (m.)

CLEAR DISTINCTION betwoen earthquakes and explnsions is evideut ill tbis plot of U,e magllitude of long-pe~iod surface waves (M ) against that of short-pe~iod body waves (01,). The 383
s
earthquakes ~epresented by the black dots were compiled f~om a set of all the earthquakes ~ecord~'" worldwide in a six-mnuth period tllat bad an Inb vwue of 4.5 or more and a focal depth of less than
311 kilometers. (There are fewer dots than earthquakes becau>e the magnitudes of Some of them coincided.) The colorcd sqnares designate underground explosions in the U.S. and the colored cros"es
uuderground explosions in tbe U.S.S.R. Only one earthquake falls within the explosion population, as defined by the straight line .separating the two groups of events. This single evenl, whieh bad the
sman""t magnitude of any of the earthquakes in the survey, took place in the sDuthwest Pacific Ocean, a region wbere tbe sensitivity of the network of seismic statiDus is poorer than it is in mnst of tbe
Northern Hemisphere. The 01 values were adjusted to take into aCCOullt regional variations in the amplitud~ of short-period "'aves.
0

DEPTH ." 50 KM.
LOCATION:> 25 KM. AT SEA DEPTH •• 30 KM.
Ms:m.

DEPTH'" 50 KM.
Ms:m.
LOCATION:> 25 KM. AT SEA
DEPTH ." 30 KM.
MS:mb LOCATION:> 25 KM. AT SEA
METHODS OF DISTINGUiSHING earthquakes from explosions were tested by applying the methods to aUthe eartbquakes with a magnitude of 4.5 or mo~e recorded during a 162-day period in 1972.
The .ample consisted of 948 events. Many of them coutd be classified as earthquakes (ruther than explosion,,) by their locatiol Dr their depth. The remaining events could be etassified by comparing tbe
maglitude of long-period surface waves with tbe mal!:nitude •• f sbort-perind body waves (the rntio M,,: RIb)' Tbe sequence ill which lbe lesls were applied afreeted the efficiency of the p~ocedure, but an
events were identified regardt""" of the sequence.
9

tared in detail from stations in Japan and the Aleutian Islands. Indeed, oceanbottom seismometers and hydroacoustic sensors could be placed just offshore.
The first dcfeme against cvasion by the masking of a test in a large earth-· quake is the questionable feasibility of the subterfuge. Unless the evader maintained several testing sites the number
oC opportunities per year for clandestine testing would be quite limited. In addition the evader would have to maintain his wea pons in constant readiness for firing. To attain the evasion
capability given below he would have to set off an explosion within IOO seconds of the time of arrival of the short-period waves of the earthquake. He would have to estimate the maximum
amplitude and the decay ratc oC the earthquake waves with high accuracy, and he would have to be certain of the amplitude of the P waves generated by the cxplosion to within ,I magnitude
unit. Even after taking these precautions the evader would have to accept a high probability that the event would be detected by at least one monitoring station and a small probability that three
stations would detect it. He would also have to install'and operate his testing site (including a large cavity) and his own seismological network in total secrecy over a period of years.
In contrast to these daunting require. ments for successful evasion, the only requirements for a monitoring nation are to operate a network of high-quality seismic stations and to process the
data with determination. Against a network of 15 external stations and 15 internal
ones the only effective cvasion schemes at yields of one kiloton or more would require both decoupling and hiding the explosion signal in an earthq liake.
The issues relating to the monitoring of a comprehensive test ban can be summarized as follows. The understanding of seismology and the le,~ting of seismometer networks arc sufficiently
complete to ensure that compliance with a u·eaty could be verified with a high level of confidence. The only explosions with a significant likclihood of escaping detection would be those of
very small yield; less than one kiloton provided the monitoring system includes stations in the U.S.S.R.
It is important to view the question of yield in the context of the nuclear weapons that have been tested up to now. The ones that ushered in the nuclear age in 1945 had a yield of from 15 to 20
kilotons. Yields increased rapidly to the point where the U.S.S.R. tested a 58,OOO-kiloton weapon in 1961. The largest underground explosion had a yield of almost 5,000 kilotons. Unclassi-
fied reports place the yield of the weapons carried by intercontinental missiles in the range from 40 to 9,000 kilotons. The yields of underground explosions that might go undetected or
unidentified under a comprehensive test ban are therefore much smaller than those of the first nuclear weapons. If the threshold of reliable detection and identification is one kiloton, that is
only one-I 50th of the limit specified by the Threshold Test Ban Treaty of 1976.

THRESHOLDS OF DETECTION for seismi~ evenls ill the Ea.<;tern Hemispher. are delineall'1l by lhe lwo sets of COlltoUrs drawn 011 llUs pair of maps for a proposed network of 15 seismic stations established
outside lhe U.S,S.R. Colon'<l dots giv.the locatioll of 12 of the 15 staHunsj three otlJers are off lhe maps. 'fhe nnmber On each cnnlonr indicates that au ..:v..:nt of that maj;tllitude Of" larger ha!li at leaSt a 90
10
From the viewpoint of verification a comprehensive test ban wo uld actually establish the equivalent of a very low threshold, since weapons of extremdy low yield could be tested underground
without the cenainty of being dctected and identified. A treaty that imposed a threshold near the limit of seismological monitoring capability might therefore be considered an alternative to a
comprehensive test ban. Such a treaty might be preferable to the present quite high threshold, but it would have the disadvantage that arguments could arise over the exact yield of tests made
near the threshold. Indeed, the judgment of whether or not a test has taken place will always be less equivocal than an exact determination of yield.
In recent years there have becn repons that the U.S.S.R. may have repeatedly violated the 1976 treaty by testing devices with a yield greater than the ISO-kiloton limit. Such reputed violations
were recently cited as evidence that the threshold treaty, which has not been ratified by the u.s. Senate, is not verifiable and should be renegotiated. On the basis of our analysis we conclude
that the reports are erroneous; they are based On a miscalibration of one of the curves that 'relates measured seismic magnitude to explosive yield. When the correct calibration is employed, it
is apparent that none of the Russian weapons tests exceed 150 kilotons, although several come close to it.
Observations at the Nevada Test Site (NTS), where American nuclear-weapons tests are held, indicate there are lin-

percent probability of being delected by four or more slations. 'rbe conlours on th. map atlh. left «prescot tbe detection thresholds Eor short-period body waves and lhose on the map al the right the detection lhtL'sholds for
long-period surface waves. Ollll,ese maps and the oncs On the apposite page the only seismic ltOlst: b.kt'n intu account is the microseismic noise ~enerated by lhe atmospbere and oceans.

IMPROVED DETECTION THRESHOLDS for seismic events ill the Eastern nemispnere :ll"e delineated by the two sets of contours drown on this pair of maps for a proposed n.ctwork of 30 seismic
stations: 15 outside the U.S.S.R. and 15 inside. For most of the U.S.S.R. the effect of adding tbe 15 interior stations would be to lower the de-
ear correlations between the logarithm of the explosive yield and the two magnitude values, Ms and mb, for ex:plosions with yields greater than 100 kilo· tons. When the measured
Ms and mb val ues of explosions at the Russian test site near Semipalatinsk are inserted into the NTS formulas, however, the resulting estimates of yield given by mb are more
than four times as great as those given by Ms. For ex:plosions in hard rock at many test sites estimates of yield based on the NTS Ms formula have invariably agreed with actual
yields, whereas estimates based on the NTS mb formula have sometimes been in drastic disagreement with the actual yield.
A strong correlation has been found between mb values measured at individual stations and P-wave travel times to these stations. The U.S.S.R. routinely publishes seismological
bulletins that include P-wave arrival times of earthquakes, and it is straightforward to interpret the times for stations in central Asia in terms of the expected pattern of mb values
near Semipalatinsk. From an analysis of the P-wave signals it is predicted that the mb value for an ex:plosion at Semipalatinsk is 40 percent greater than an equivalent explosion at
NTS. This is the same correction that must be applied to the curve relating mb to yield at NTS to make the mb estimates of the yield of Russian explosions consistent with the Ms
estimates. Thus two modes of analysis lead to the conclusion that there is an essentially universal relation between Ms and yield whereas the curve relat;ng mb to yield must be
calibrated for each test site.

tediOD threshold for short-p~riod body WllVcS from a magnitude of 3.8 10 aile of 3.4 (left); the c<>rnsponding effect for long-period surface wav,," would be to lower tbe detectioo threshold from a
magnitude of 2.6 to one of 2.3 (right). Tbe interior stations would also provide more :Decurate information on the focal deptb of a seismic event.
A comprehensive treaty would have an additional advantage over a lowthreshold treaty: all technological uncertainties would work against the potential evader. A country planning
a surreptitious n.uclear test could not know the exact seismic-detection capability of other nations or the exact magnitude of the seismic waves that would bc generated by his test.
A ban on nuclear explosions of all sizes would also have the important conceptual value that nuclear weapons, no matter what their size, would be recognized as inherently dif-
ferent from conventional weapons.
It is sobering to consider how the state of the world would differ if a full test ban had been achieved in 1963. The number of nuclear weapons has grown tremendously since then
and is now estimated at from 50,000 to 100,000. The loss of life and the social damage that would be inflicted in a ma. jor nuclear exchange are vastly greater than they were in
1963. Furthermore, both the U.S. and the U.S.S.R. are less secure now than ever before, not because of any failure to develop arms but because of the growing stockpiles of
weapons and the inability of any nation to defend itself against nuclear attack.
A comprehensive test-ban agreement should not be regarded as a substitute for disarmament. Meaningful reductions in the nuclear threat must include a continuing and serious
process of arms control; in this process, however, a comprehensive test-ban treaty could have an important part. The problems of negotiating such a treaty are overwhelmingly
political rather than technical and must be recognized as such.
Before the suspension of negotiations between the U.S., Britain and the U.S.S.R. in 1980 tentative agreement had been reached on anum ber of issues. All three nations agreed that
a test-ban treaty would include a prohibition of all tests of nuclear weapons in all environments, a moratorium on peaceful nuclear explosions until arrangements for undertaking
them could be worked out, provisions for on-site inspections, a mechanism for the international exchange of seismic data and the installation of tamperproof seismic stations by
each country in the territory of the others. The proposed treaty would have a term of three years. The agreements on the long-standing issues of on-sire inspection, peaceful
explosions and the placement of monitoring stations in each country represented important breakthroughs. It wouLd be a setback for the cause of international security if this hard-
won ground were now lost.
For many years the stated policy of the U.S. has emphasized the desirability of a complete test ban if verification could be ensured. The policy was not fundamentally altered by
the recent decision of the Reagan Administration to put off further negotiations on the test ban. On the contrary, it was reported that the Administration still supports the ultimate
goal of a comprehensive ban on nuclear testing but has doubts about the efficacy and reliability of seismic methods of verilication. As we have attempted to show here, there can
be no substance to such doubts.

The Authors
LYNN R. SYKES and JACK F .. EVERNDEN are geophysicists who in addition to their scientific work have spent
con~iderable time OIl means of distinguishing underground nuclear explosions from earthquakes. Sykes is Higgins
Professor of Geological Sciences at Columbia Uni.versity and head of the eanhq uake studies group at the university'::;
Lamont-Doherty Geological Observatory. His B.S. and M.S. were both awarded in 1960 by the Massachusetts lnstitute
of Technology. His Ph.D. in geology was granted in 1964 by Columbia and he has remained at the university since,
taking up his present jobs in 1971. Sykes's scientific work has centered on plate tectonics, including the plate-tectonic
str ucture of Asia, which may be significant in the verification of a testban treaty. Evcrnden is program manager of the
U.S. Geological Survey's National Cenwr for Earthquake Re:iearch. His B.S. in mining geology .( 1948) and his Ph.D. ht
geophysics (1951) are from the University of California at Berkeley_ He writes: "I, of course, agree with all conclusions
presented in this artide, but that agreement should not be construed as acceptance of those conclusions by either the
U.S. Geological Surveyor the United States Government."
Bibliography
TOWARD A COMPREHENSIVE NUCLEAR TEST BAN TREATY. Hearing before the Subcommittee on Arms Control, In-
ternational Law and Organization of the Committee on Foreign Relations, U.s. Senate, May 15, 1972. U.S. Gov-
ernmentPrinting Office, 1972.
THE GREAT TEST-BAN DEBA"rE. Heibert F. York in Scientific American, Vol. 227, No.5, pages 1:5-23; November,
1972.
NUCLEAR EXPLOSIONS AND EARTH-
QUAKES; THE PARTED VEIL. Bruc.e A. Bolt. W. H. Freeman and Company, 1976.
STUDY OF SEISMOLOGICAL EVASION, P AR-r I: GENERAL DISCUSSION Of V AR· [QUSEVASIONSCHEMES. p, RT2:EvALUATION
OF EVASION POSSIBILITIES UsING NORMAL MICROSEISMIC NOISE. PART 3: EVALUATION Of' EROSION PosSIBIUTIES USING
CODAS OF LARGE EARTHQO/KES. Jack F. Evernden in Bulletin of the Seismological Sodety oj America, Vol. 66,
No.!. pages 245280 (Part 1), pages 281-324 (Part 2), February, 1976; No.2, pages 549~592 (Part 3), April. 1976 ..
THE COMPREHENSIVE TEST BAN. Stock~ holm International Peace Research Institute in World Armaments and
Disarmament: SIPRI Yearbook 1978. Ta.ylor and Francis Ltd., 1978.
TRlPARTITE REPORT TO THE COMMITIEE ON DISARMAMENT (UNITED NATIONS). In World Armaments and Disarma-
menl: SIPRI Yearbook 1981. Taylor and Francis Ltd., 1981.

Proof readingsamplejobjungle

More Related Content

What's hot (20)

Similar to Proof readingsamplejobjungle (20)

Recently uploaded (20)

Proof readingsamplejobjungle