the U.S. Geological Survey´s PDE is incorrect; or the Pakistanis mis-identified the event; or the Pakistanis called attention to the event for other reasons.4The U.S. response to the Pakistani inquiry was based on the analysis of global seismic data such as that routinely collected by the U.S. Geological Survey and published in its seismic catalog entitled "Preliminary Determination of Epicenters" (commonly referred to as the "PDE").1 The PDE listing for April 28, 1991 included a seismic event that occurred not at 9:00 a.m., but rather at 8:33 a.m., and the magnitude was not 5.6, but rather 4.6. In addition, the PDE location for the event (latitude 30.726N, longitude 71.806E) indicated that the event was not near the Indian test site, but rather in the center of Pakistan and at a depth of 26 kilometers. 2 The differences in magnitude and time may be due to translation errors as the information was passed along to the U.S.: the 9:00 could simply be a rounding off of 8:33, and the 4.6 versus 5.6 could be a typographical error.3 However, the difference in location could not have been introduced during the relay of information, for it is the basis of the inquiry. One can therefore assume three possibilities:
To determine if the information within the U.S. Geological Survey´s PDE was correct, the Pakistanis could have accessed data from open stations outside their country. The following figure shows both the PDE location and the location of the Indian test site. Although 51 stations were used in the USGS´ location of the event, we show below how as few as four high quality stations can be used to identify the event as an earthquake. Triangles mark these open seismic stations installed under the IRIS and U.S. Geological Survey´s Joint Seismic Program with the Commonwealth of Independent States. These stations are part of the Global Seismic Network. Four of the stations are in Russia: Obninsk (OBN), Arti (ARU), Talaya (TLY), and Kislovodsk (KIV). One is in Kyrgyzstan, Ala-Archa (AAK). The concentric circles are at approximately 1000 kilometer intervals from the PDE location.
The Kislovodsk (KIV) station was turned off for 20 minutes on the morning of April 28 while the station operator changed the recording tape.5 The 20 minute gap in data recording unfortunately corresponds to the arrival time of the seismic signals from the event. The other four stations, however, recorded the event and the data can be accessed openly. The top figure on the next page shows the vertical component of the unfiltered broad-band data (BHZ) from the AAK, ARU, OBN, and TLY stations. The two lower figures show the same data filtered through high (0.5-2.0Hz) and low (10-100 secs.) bandpass filters. By looking at both the closest station (AAK) and the furthest station (OBN), one can see how different stations complement each other in resolving the nature of the event.
AAK is the closest station and shows clear signals in both high and low frequencies. The body waves (both P & S) are clearly identifiable in the high-pass filtered record. By measuring the time difference between the arrivals of the P and S waves, it can be determined that the distance is consistent with the PDE location. In addition, measurement of the P-wave magnitude at AAK confirms the 4.6 body-wave magnitude listed in the PDE. At the close distance of AAK, however, the signal looks explosion-like because it appears to have large body waves and relatively small surface waves. While the AAK data can be used for location and body-wave magnitude, the short distance between the station and the event does not allow an accurate determination of the surface-wave magnitude Ms. The ratio of Ms (surface-wave magnitude) to mb (body-wave magnitude) is one of the most common methods for discriminating the signals created by earthquakes from those created by explosions.
For OBN, the unfiltered data are difficult to analyze. The OBN station is located near a city and the broadband recordings show a high-level of seismic noise. If the data are filtered, however, the body waves can be seen in the highpass record with sufficient resolution to make an mb estimate. Similarly, the surface waves (R) can be seen after the low-pass filtering with sufficient resolution to make an Ms estimate.6 Using filtered OBN data, one can calculate the Ms/mb ratio. The relatively large surface waves (and hence large Ms/mb ratio) are consistent with seismic waves generated by earthquakes. Scientists unaccustomed to looking at broad-band data might have dismissed the OBN data as noisy and therefore not useful.7 Through simple digital filtering, however, the OBN data provide strong evidence that the seismic event was in fact an earthquake.
 |
In this example, a party with basic analytical skills and access to data from open seismic stations could validate independently the information within the PDE and identify the seismic event as an earthquake. The following three elements, however, are necessary:
As can be seen from this example, it is not sufficient to have merely a catalog of events or openly accessible seismic stations. By its very nature, seismology is an international science. Seismology requires cooperation because data from many different geographic locations must be gathered to answer scientific questions. In deterring proliferation, this requirement can be turned into a strength. Through the installation of an open global seismic network, seismological researchers within different countries become linked on an international basis. Analytical skills are shared as data are exchanged. As coverage expands and analytical skills are developed more extensively on an international basis, ambiguous events such as the April 28 event will be resolved quickly. |
Return to:Table of Contents
Return to:Chapter III
Continue to:Chapter V, Assessing Monitoring Requirements