Physics and Society 24, 1-3 (April 1995)
________________________________________

Monitoring Nuclear Test Ban Treaties
David Hafemeister
Physics Department
California Polytechnic State University
San Luis Obispo, CA  93407 
 
	With the end of the Cold War, the emphasis on monitoring nuclear 
weapons tests has shifted from the 1976 Threshold Test Ban Treaty (TTBT), 
which confined the U.S. and U.S.S.R to a limit of 150 kilotons (kt), to 
the Comprehensive Test Ban (CTB) Treaty, which would ban all nuclear 
tests by the signatories.  This year, 1995, will be a watershed year to 
control the proliferation of nuclear weapons as the Nuclear 
Non-Proliferation Treaty (NPT) must be considered for renewal.  The fate 
of the NPT is dependent on the successful negotiation on the CTB because 
the NPT and CTB are certainly politically linked.  This paper will cover 
one aspect of these broader issues by discussing the monitoring of the 
TTBT and CTB.  It is important to study the past (TTBT) in order to learn 
for the future (CTB).
 
	TTBT Monitoring:  TTBT monitoring1 determined the yields with 
respect the 150-kt threshold primarily by measurements of teleseismic 
waves that traveled through the interior of the Earth.  Because the 
tectonic plate below the Nevada site has been extended and partially 
melted, seismic waves from the Nevada Test Site (NTS) explosions are 
diminished more than the waves from explosions at the Shagan River Test 
Site (SRTS) which has had no recent geological activity.  Because of this 
difference in magnitude (the "bias"), the U.S. explosions appear 
relatively smaller in yield compared to explosions of the same yield in 
Russia.  By ignoring this difference, the U.S. charged that the Soviets 
were "likely" violators of the TTBT.  As one way to sort out this 
dispute, the U.S. measured the yield of a Soviet explosion at the SRTS 
with electronic CORRTEX equipment in 1988.  The magnitude of this 
explosion was 6.03 (mb), with a yield of 106-118 kt from seismic data.  
Since a magnitude 6.03 explosion in Nevada corresponds to an explosion of 
450 kt, clearly there is a great difference in the geology of the two 
sites.  Using the CORRTEX data and other seismic data, such as the Lg 
surface wave, the U.S. dropped the noncompliance charge in 1990. 
 
	The classification of the seismic data prevented a thorough 
discussion of the data.  Parts of the U.S. policy community purposefully 
ignored the geological "bias" determined by its seismologists in order to 
maintain the "likely violation" charge against the Soviets.  This charge 
greatly retarded negotiations with the Soviets on both the TTBT, the CTB 
and other arms control treaties.  For example, in 1988 Acting Assistant 
Secretary of Defense Frank Gaffney stated2:  "The thinking goes like 
this: the more time wasted on discussion and experimentation of 
monitoring techniques irrelevant to the verification of an environment in 
which there are no legal tests, the easier it will be to stave off 
demands for the more constraining comprehensive test ban."
  
	With the end of the Cold War, yield data has been obtained from 
the former Soviets.  It is relevant to examine this data because the TTBT 
noncompliance issue can be used politically by those who do not favor the 
CTB.  We have plotted in Fig. 1, the recent estimates by Ringdal, 
Marshall and Alewine3 of the 38 largest post-1976 tests above 90 kt.  For 
comparison sake, we have plotted the projected distribution for tests at 
150 kt using an uncertainty factor F of 1.64 and 1.4. When one compares 
the data with the calculated probability for testing at 150-kt, one can 
only conclude that the Soviets were in compliance with  the TTBT.  It is 
possible that a very few tests could have been slightly above 150 kt, but 
this situation also prevailed for U.S. testing at NTS.  The value of the 
mb bias between the NTS and SRTS was crucial to the interagency process.  
Using the recently published data, we obtain an average value 0.38 for 
the three SRTS regions, which further strengthens the case for 
compliance.  This estimate does not take into account the smaller 
differences caused by the type of rock next to the explosion.  As the 
former technical lead for the State Department on nuclear testing issues 
in 1987, I was constrained to use a lower value of the bias value, but 
even with this value I still concluded (as did CIA and DOE, but not the 
U.S. government) that the Soviet Union was in compliance to the TTBT.  In 
testimony before the Senate Foreign Relations Committee on October 6, 
1988, OTA, LLNL and I stated that the Soviets were in compliance, but, of 
course, we were not allowed to use the specificity of the data that 
appears above. 
 
	What were the TTBT lessons for the CTB?  (1) The Soviets were in 
compliance to the TTBT.  (2) A multinational compliance process using 
unclassified data is needed to restrain politization.  
 
	CTB Monitoring:  The CTB requires a different monitoring approach 
which requires the detection and identification of a nuclear explosion 
rather than the quantification of a nuclear yields.  Roughly speaking, a 
1 kiloton explosion, tightly coupled (tamped) in hard rock, will have a 
seismic magnitude of about mb = 4.  It is generally accepted that a 
coupled explosion of about 0.1 kiloton could be detected and identified 
with the two types of networks now being proposed.  In 1988, the 
Congressional Office Technology Assessment1 concluded:  "Based on 
cautious assumptions for a network of 30 internal arrays [to USSR] or 
about 50 three-component internal stations, it appears likely that a 
detection threshold of 2.5 mb (90 percent probability of detection at 
four or more stations) could be reached."  Since it is more difficult to 
identify than detect, one should generally add about 0.5 mb units to the 
detection level when discussing identification.  
 
	Model calculations by Claassen, Unger and Leith5 show that the 
present network of 79 stations (including the four arrays) on the 
Eurasian continent can typically detect down to the 3.6 mb level (90% 
probability that four P-wave arrivals will be observed at 4 different 
stations).  They also project that a network increased to 128 stations 
would be able to typically detect a threshold of 3.2 mb, as shown in Fig. 
2.  If the criteria of 4 P-wave arrivals is relaxed to 2 P-wave and 2 
S-wave arrivals, the typical detection level is lowered to 2.8 mb.  For 
the regional networks to be most successful it will be necessary to study 
the regional seismic characteristics.  The combination of the near 
real-time alpha network of arrays and the beta network of broad-band 
triple-axes seismographs should be able to monitor coupled nuclear tests 
to less than one  kiloton.
 
	Decoupled Explosions:  If a nuclear weapon is placed in a cavity 
of sufficient size, such that the blast pressure on the cavity wall is 
below the elastic limit of the surrounding media, the seismic signal 
strength1 can be reduced by a factor of 7-70.  The cavity size necessary 
to obtain these decoupling factors has a radius of 20-25 meters per 
cube-root kiloton.  Thus, a 30 kt explosion would need a cavity radius of 
60-75 m to achieve full decoupling -- an extraordinary engineering 
challenge when one considers the requirement for secrecy.   Many experts 
have concluded that the decoupled signal would still be detectable and 
identifiable.  If a 1 kt weapon had an unexpected yield of 5 kt, which is 
quite possible for a clandestine new program, it would require a cavity 
radius of 35-45 meters (diameter of 70-90 meters), a factor of 1.7 larger 
than for the 1 kt cavity.  The tester's problems are further complicated 
by possible venting which could be easily detected; 30% of Soviet tests 
vented and the U.S. had severe venting problems with its earliest tests.  
In particular, it appears that smaller tests can be harder to contain 
than larger ones.  The last four U.S. explosions that vented were from 
explosions with yields less than 20 kilotons.  It is hypothesized that 
smaller explosions would not sufficiently glassify the cavity, and also 
would not rebound sufficiently to close fractures with a stress cage.  
Thus, the smaller explosions, which one might think were easier to hide, 
are more likely to vent and could be detected by the release of 
radioactivity.  For these same reasons, it is further hypothesized that 
partially decoupled tests would also be difficult to completely contain.  
There is very little data6 on decoupled tests in cavities, and only one 
has been carried out with a yield greater than one kiloton.  Other 
intelligence means can also gather evidence on clandestine decoupled 
nuclear tests.  It is widely felt that a clandestine execution of a few 
kiloton or larger shot, that was decoupled to a degree that enabled the 
test to escape detection by seismic means and which did not have yield 
excursions, would require the resources of a very technologically 
sophisticated nation.     
 
1.  Office of Technology Assessment, Seismic Verification of Nuclear 
Testing Treaties  (OTA-ISC-361, 1988) and The Containment of Underground 
Nuclear Explosions  (OTA-ISC-414, 1989).
 
2.  G. van der Vink and C. Paine, Science and Global Security 3, 261-288 
(1992).
 
3.  F. Ringdal, P. Marshall, and R. Alewine, Geophys. J. Int. 109, 65-77 
(1992).
 
4..  P. Richards, IEEE Tech. and Society 9, no. 4, 40-52 (1990).
 
5.  J. Claassen, The Potential Monitoring Contribution of the Open 
World-wide Network, Sandia National Laboratory (SAND 93-1931).  J. 
Claassen, J. Unger and W. Leith, IRIS Newsletter 12, 1-7 (Aug. 1993).
 
6.  L. Glenn and P. Goldstein, J. Geophys. Res. 99, 11723-30 (1994)
 
Figure Captions
 
Fig.  1.  Yields of post-1976 Soviet nuclear tests3 over 90 kilotons at 
Shagan River Test Site.  The theoretical curves are based on tests at 150 
kt and uncertainty factors F=1.6 (wider curve) and F=1.4.
 
Fig. 2.  Detection threshold estimates5 in mb(Lg) values with 128 seismic 
stations and arrays.  A probability of 90 percent that at least four 
stations detected the event was used to determine the contours.