The Political Sensitivity of Earthquake Locations

Gregory E. van der Vink, IRIS Consortium and
Terry Wallace, The University of Arizona

Despite the often complex nature of the task, most seismologists regard locating earthquakes as a relatively routine exercise; not a task that might begin a chain of events leading to a crisis in international diplomacy or the outbreak of nuclear war. In the context of monitoring underground nuclear weapons testing, however, a mistake in the location of a seismic event can have severe political consequences.

In 1996 and 1991, earthquakes occurred in particularly sensitive locations during particularly sensitive times: one at a Russian test site during the international negotiations of a Comprehensive Test Ban Treaty (CTBT), and the other near the Indian nuclear test site during a period of heightened military readiness between India and Pakistan. If the location errors had not been recognized and quickly corrected, both of these seismic events could have undermined sensitive diplomatic negotiations.

Although there are strong arguments for keeping the overall capability of a monitoring system secret, these two events clearly demonstrate the importance of open and readily accessible data. It is through the open and independent assessment of the data that the concerns were resolved. In addition, debate exists over the relative negative impact of missing a violation versus falsely accusing a nation. For nuclear monitoring, either mistake can be dangerous.

Possible Derailment of the CTBT Negotiations

In the Spring of 1996, a potential derailment to the nuclear test ban treaty talks occurred when reports surfaced that Russia might have violated international agreements by secretly conducting a nuclear test. On March 7, The Washington Times revealed that "U.S. intelligence agencies suspect Russia secretly set off an underground nuclear test."1 An anonymous source in the article was quoted as saying "It was a low-yield test in mid-January" and the location was referenced as being at the Russian arctic test site of Novaya Zemlya. The next day, further support for the allegation was provided in a follow-up article under the title "Perry cites evidence of Russian nuke test" in which the Secretary of Defense was quoted as saying "there is some evidence on the subject, but there's also some ambiguity in the evidence".2

The two articles (Figure 1) inferred that the intelligence community had based their suspicions on observed activities at the Arctic test site, and that the activities they saw were similar to what is seen during a nuclear weapons test. The ambiguity is most likely because such activities are not unique to underground nuclear tests. For example, subcritical nuclear experiments are conducted underground at test sites in a manner similar to nuclear weapons tests. The experiments involve tens to hundreds of pounds of high-explosive charge and result in the dispersal of nuclear materials such as plutonium-239. Because the explosions do not produce neutrons, gamma rays, or a nuclear yield, however, they are not considered nuclear test explosions, and therefore are not restricted by international agreements. The United States, in fact, had plans to conduct a series of six subcritical experiments (two in 1996 and four in 1997) underground at the Nevada Test Site.3

Figure 1. Newspaper articles in the Washington Times revealing reports of a suspected nuclear weapons test at the Russian test site inNovaya Zemlya during January 1996.

While testing-related activity may not necessarily imply a nuclear test, a corresponding seismic event would be conclusive. According to the articles in The Washington Times, the lack of such a seismic signal lead some intelligence analysts to conclude that there was no test. Others, however, apparently argued that the test was either too small to generate a seismic signal, or else the test was conducted in such a manner that the seismic signal was muffled or masked.

The Search for a Seismic Smoking Gun

Despite its remote Arctic location, several seismic observatories have been installed around Novaya Zemlya to monitor the nuclear weapons testing program of the Soviet Union. As a result, the detection threshold for the area is significantly lower than the global average of around magnitude 4.0 to 4.5.4 In fact, the seismic arrays in Norway (called ARCESS) and in Spitsbergen (called SPITZ) are capable of clearly recording seismic events on Novaya Zemlya over 1000 times smaller - as small as magnitude 2.5 and perhaps even smaller.5 These observatories, however, did not list a seismic event corresponding to the activity on Novaya least initially.

A few days after The Washington Times articles appeared, however, the seismological equivalent of a smoking gun was thought to have been found. A re-examination of the seismic records revealed that a magnitude 2.5 event occurred at 1717Z on January 13, 1996. As illustrated in Figure 2, it is easy to understand why no event was recognized initially. The signal is barely discernable above the noise level, even when the data are high pass filtered above 8Hz. A magnitude 2.5 seismic event is relatively small. On a global scale, over 200 magnitude 2.5 earthquakes occur each day. However, while magnitude 2.5 may be small for an earthquake, it represents a significant explosion. The relationship between seismic magnitude and explosive yield is variable, particularly at low yields. Factors that relate to how well the force of the explosion is coupled to the surrounding rock, and the extent to which the seismic waves are attenuated as they pass through the Earth, create variability in the relationship.

Figure 2. Seismic recordings of the January 13 event from the ARCESS array in Norway and the SPITZ array in Spitzbergen. The blue data is unfiltered, the red data is filtered above 8Hz. P and S wave arrivals are identifies on the data.

While an exact explosive yield can not be precisely determined, the magnitude of 2.5 does allow some bounds to be placed on the range of possible yields. For a stable tectonic setting such as Novaya Zemlya, the relationship between explosive yield and magnitude is approximately: mb = 4.45 + 0.75 log Y.6

A magnitude 2.5 seismic signal would therefore correspond to a well-coupled explosion of a few tons of TNT. The yield determination becomes further complicated, however, if one assumes that the explosion could have been detonated in a large underground cavity in an attempt to "muffle" or reduce the seismic signature. Based on experiments conducted in salt deposits in both the former Soviet Union and the United States, such decoupling could in principle reduce the seismic signal by a factor of 70.7 Making the worst-case assumption that the event was caused by an explosion successfully decoupled at the maximum reduction of a factor of 70, the explosion may have been as large as 700 tons.

If the event was caused by a nuclear explosion, the yield would have been between a few tons and 700 tons. An explosion in this range would exceed the limits for subcritical experiments, and would be construed as a meaningful nuclear test. The test would indicate the end of Russia's declared moratorium on nuclear testing. Furthermore, because Russia did not announce the test in advance to the United States, the test would also be a violation of the notification agreements associated with the 1974 Threshold Test Ban Treaty. Such violations, in turn, would seriously undermine international efforts to attain agreement on the Comprehensive Test Ban Treaty.

The Seismic Event Re-analyzed

The standard procedure for locating events with seismic arrays uses the difference in arrival times of the various seismic phases at each of the array elements. These differences can be used to determine the direction (azimuth) of the arriving wavefront and its apparent velocity. The velocity of the wavefront, in turn, is dependent on the distance between the earthquake and array locations. For the ARCESS and SPITZ arrays, we obtained results shown in Table 1.

The difference in azimuth determinations between the Pn and Sn phases at both stations is due to the relative low signal to noise ratio of the recorded signal and the fact that the different phases traveled along multiple-paths to the array. The Sn phases are used for distance but not for azimuth determination. Using the apparent azimuths from the Pn phases and the picks for the Pn and Sn phases, we determine the location of the event at 51.27N 75.40W, seventy kilometers west of the preliminary location of 75.3N, 55.3E. The error ellipse for the epicenter determination covers 560 square kilometers, and represents the 90% confidence level. Although factors such as station and path corrections create uncertainty in the location, we can determine with high confidence that the event did not occur on the island of Novaya Zemlya. The location therefore precludes the event from being an underground nuclear explosion. If the seismic recordings were produced by an underwater explosion, the explosion would have produced a "bubble pulse" reverberating between the surface and the ocean floor. Such reverberations would be visible in the seismic signature of the event; and none are observed in the waveforms.

At regional distances, the seismic recordings of earthquakes and explosions have some subtle differences. The differences in the seismic recordings arise from the fundamental differences in the source of the signal. Explosions exert a primarily compressive motion over a relatively small volume of rock. Earthquakes, on the other hand, have a shearing motion along extended fault surfaces. One of the best methods for discriminating earthquakes from explosions is therefore the ratio of the high frequency S (shear waves) to the high frequency P (compressional waves). In general, the S/P ratio is much smaller for explosions than earthquakes. The regional S/P ratio discriminant must be calibrated for individual regions and paths because both compressional and shear waves are affected by the structure and physical properties of the regional geology.8 The pathway between the ARCESS array and Novaya Zemlya is well calibrated, and ratios of greater than one are known to be characteristics of earthquakes.9 As can be seen in Figure 2, the ARCESS recording shows that the amplitude of the S waves is greater than that of the P waves for the frequency band of 1 to 8 Hz, thus indicating a ratio of greater than one and providing further indication that the event was an earthquake.

Mistaken Identity

A seismic signal emanating from the Russian activities on Novaya Zemlya would have been strong evidence of major violations of international agreements. Taken together, the location, magnitude, and character of the January 13 seismic records, however, strongly indicate that they were created not by an explosion on Novaya Zemlya, but rather by a small earthquake in the Barents Sea. Although the absence of a seismic signal can not be construed as proof of innocence, it does shift the burden of proof. Quite possibly, the Russians were conducting a subcritical experiment. Such activities are permissible under current test ban agreements, and therefore do not violate international agreements. The lack of a seismic signal makes such an alternative explanations credible. Paradoxically, the signing of a CTBT would allow for on-site inspections, thus creating a mechanism for resolving such ambiguous activities in the future. Thus the very agreement that was threatened by the ambiguous nature of the activity, will provide a mechanism to resolve such ambiguity in the future.

Although the relocation of the seismic event was able to reduce suspicion that the Russians may have conducted a test, it is frustrating to note that this reassuring information never made it to the attention of the press that publicly raised the concern in the first place. In fact, just the opposite seems to have occurred. Three months after the first report, an article in The Washington Times about Russia’s efforts to modernize their nuclear forces also cited the event. By this time, the seismic evidence had been completely, and erroneously, turned around. In a self-perpetuating fashion, the presence of a seismic signal, which had been cause for doubt in the earlier stories, had now become the basis for the allegation, and was touted as follows as evidence of Russian duplicity:

In January, as this paper revealed and Defense Secretary William Perry confirmed, the Pentagon detected seismic activity consistent with a low-yield nuclear blast at the underground arctic nuclear test site in Novaya Zemlya, even though Moscow had pledged in 1992 to stop such detonations.10

Not the first time

Unfortunately, the Novaya Zemlya event is not the only example of where a mis-located seismic event had the potential to create international conflict. On April 30, 1991 the Government of Pakistan notified the U.S. Ambassador that Pakistan had detected a seismic disturbance in India's Rajasthan Desert, the location of India’s 1974 nuclear test (Figure 4). Because the seismic event appeared to have no aftershocks, the Pakistanis were suspicious that the event was not an earthquake, but rather by inference, that India had conducted an underground nuclear test.

Figure 4. Telex from the United States Embassy in Islamabad Pakistan requesting information about a seismic event with no aftershocks that was thought to have occured near the Indian test site on April 28, 1991.

Since the end of British colonial rule in 1947, territorial battles between India and Pakistan have fueled an arms race that according to testimony by the Director of the Central Intelligence Agency: poses perhaps the most probable prospect for future use of weapons of mass destruction, including nuclear weapons. 11

Both India and Pakistan are thought to possess nuclear weapons, or at least to be capable of assembling nuclear weapons rapidly. The Pakistanis asked the U.S. Ambassador for an assessment of the April 28, 1991 seismic event. The United States was able to respond to Pakistan, and assure them that the seismic event that they had recorded was an earthquake and not an Indian underground nuclear weapons test. The U.S. response was based primarily on the location of the event. The event was listed in the U.S Geological Survey's global seismic catalog. It had been recorded by several dozen stations around the world. The location for the event was not near the Indian test site, but rather in the center of Pakistan and at a depth of 26 kilometers (Figure 5).12 The Pakistanis had mislocated at the Indian test site an earthquake that had occurred within their own country.

Figure 5. Actual location of the April 28 seismic event as determined from several dozen seimsic stations and listed within the U.S. Geological Survey's Preliminary Determination of Epicenters global catalog.


We would like to thank Doug Baumgardt for helpful discussions and preliminary analysis of the array data, and Mark Tinker for technical assistance.

Return to: IRIS Newsletter Information
Return to: Title Page and Table of Contents
Return to: Previous Article

Continue to: First Feature