As discussed previously, the question of "Down to what level should we monitor?" becomes a cost-benefit evaluation of how many resources we are willing to allocate to the task and how much uncertainty we are willing to accept. No independent assessment of monitoring requirements has been made of the detection and identification threshold required for treaty verification. In addition, no comprehensive evaluation has been made of the role of seismology in the larger picture of monitoring that includes photo reconnaissance, radiation sampling, and other intelligence gathering systems. As a result, assessments of required monitoring levels remain subjective. As we shall now discuss, these subjective assessments are, in turn, highly sensitive to the assumptions we make about the nuclear testing practices we wish to monitor.
A simple fission weapon will probably have an explosive yield of at least 10-30 kilotons.1 As seen in the following table, most first-time tests have been in this range:2
|Yield (in kilotons)|
|United States (7/16/45)||21|
For the initial test of a first generation nuclear weapon with a yield of 10 kilotons or greater, decoupling is not a credible scenario and therefore the monitoring of such activity is simple. Explosions of 10 kilotons or greater produce seismic signals of roughly magnitude 4.8. There are approximately 1,500 such events per year. These events can be readily detected and identified with present capabilities.
If we extend the monitoring requirement beyond the first-time tester to nations with advanced nuclear weapons, or assume that a more sophisticated nuclear weapon could be created or obtained, then the monitoring requirements become more extensive. For example, a one kiloton nuclear explosion creates a seismic signal with a magnitude of approximately 4.0. There are about 10,000 seismic events each year of magnitude four or greater. Even at this magnitude, however, there is little disagreement that all such events could be detected and identified readily with the networks of seismic stations that currently exist or are being installed.
If we extend the monitoring task further to a fraction of a kiloton, or assume that a country could decouple the explosion in a large underground cavity, the monitoring task becomes increasingly difficult. For example, if a country were able to decouple successfully a one kiloton explosion in a large underground cavity, the seismic signal generated by the explosion might be equivalent to 1/70 of a kiloton, or approximately 15 tons. A 15 ton explosion has a seismic magnitude of about 2.5. Although a detection threshold of magnitude 2.5 could be achieved, there are several hundreds of thousands of seismic events each year at or above that magnitude level. Even if discrimination was 99.9% accurate, there could still be many events that are not positively identified. Furthermore, at low magnitudes such as 2.5, one must not only distinguish possible small nuclear tests from earthquakes, but also from chemical explosions used for legitimate industrial purposes. In the United States, for example, there are hundreds of chemical explosions of approximately that size each month.
It should, however, be kept in mind that uncertainty in the monitoring system does not necessarily create opportunities to violate the treaty. As we shall see in the next section, the question of monitoring capability is very sensitive to how the question is asked. In particular, one must decide what are the required levels of confidence, and how credible various evasion scenarios are for different nations. As a result, assessments of verification limits are determined not only by technical calculations, but also by judgments about what constitute reasonable assumptions about deterrence, risks, and benefits.
Return to:Table of Contents
Return to:Chapter IV
Continue to:Chapter V, How Should We Evaluate Monitoring Requirements and Capability?