Dealing with Hard-to-Identify Seismic Events Globally and Those near Nuclear Test Sites

Transcription

1 Dealing with Hard-to-Identify Seismic Events Globally and Those near Nuclear Test Sites Lynn R. Sykes 1 and Meredith Nettles 2 Lamont-Doherty Earth Observatory and Department of Earth and Environmental Sciences, Columbia University, Palisades NY U.S.A. Under the CTBT the identification of nuclear explosions and either problem or anomalous seismic events is reserved for national CTBT authorities. Since their findings typically are classified, it is difficult to ascertain how well seismic events of various sizes can be identified in addition to being detected and located. We update through 2008 Sykes 2002 study Four Decades of Progress in Seismic Identification Help Verify the CTBT. Only about 70 events detected over the past 50 years were singled out in openly available scientific and governmental publications or the media as problem events whose identification potentially compromises the verifiability of the CTBT. They are a tiny fraction of the many earthquakes that have been reported. Special studies of problem events using a variety of techniques identify nearly all of them as either nuclear explosions, earthquakes, chemical explosions or mine-associated events. The sizes of problem events that require special attention have decreased dramatically with time as seismic instrumentation, analysis techniques and data transmission have improved. Since many occurred either near nuclear test sites or in a few countries that may be seeking nuclear capabilities, they have been, and continue to be, of great importance to policy makers. We also examine 38 seismic events located by the IDC (International Data Center) and its predecessor from 2000 through 2008 within 100 km of 6 nuclear test sites. Most occurred near the test sites of China, Nevada and Pakistan. No events were reported by the IDC near the Russian site at Novaya Zemlya or that of India. Only two seismic events were reported within 100 km of the North Korea nuclear explosions of 2006 and We examine and identify events of magnitude > 4.0 using long-period body and surface waves, focal mechanisms (CMTs), mb-ms, depth phases and the presence of foreshock/aftershock sequences. CMT mechanisms, which can clearly identify an event as an earthquake, now extend down to moderate-size events (for NTS to mb 3.4). Smaller events require an analysis of spectral ratios of regional seismic waves, which is the subject of a companion paper by Kim, Richards and Sykes (2009). s: 1 sykes@ldeo.columbia.edu; 2 nettles@ldeo.columbia.edu

2 INTRODUCTION AND MAJOR CONCLUSIONS Global Identification. Under the CTBT the identification of nuclear explosions and problem or anomalous seismic events is reserved for national authorities. Since their findings typically are classified, it is difficult to ascertain how well seismic events of various sizes can be identified in addition to their being detected and located. Nevertheless, only about 70 seismic events detected globally over the past 50 years were singled out either in unclassified scientific or governmental publications or the media as problem events whose identification potentially compromises the verifiability of the Treaty. That number is a tiny fraction of reported earthquakes. Special studies of problem events globally discriminate nearly all of them as either nuclear explosions, earthquakes, chemical explosions or mine-related events. See Figures 1 to 3. The sizes of problem events and their yields, if they had been nuclear explosions, have decreased dramatically with time. Identification of Events near Nuclear Test Sites. Many publicized problem seismic events occurred either near test sites or in countries that may be seeking nuclear capabilities. They have been, and continue to be, of great importance to policy makers. Here we emphasize identification, i.e. discrimination. We examine 38 seismic events of magnitude mb > 3.3 from 2000 through 2008 within 100 km of 6 nuclear test sites as located by the IDC. Most occurred near the test sites of China (Lop Nor), U.S.A. (NTS) and Pakistan. More than half of the total occurred near Lop Nor. No events were reported by the IDC near the test sites of Russia (Novaya Zemlya) or India. Two small seismic events were reported within 100 km of the North Korean test site and were identified as earthquakes.

3 Figs. 1, 2 and 3. Anomalous and Problem Seismic Events of past 50 years. We update through 2008 Sykes 2002 study Four Decades of Progress in Seismic Identification Help Verify the CTBT. Events, their dates and magnitudes (mb) are taken from him and more recent literature. Note that neither unidentified nor alleged nuclear explosions (Fig. 1) have occurred since 1990 of mb > 2.5. The magnitudes of earthquakes from the literature that required special study decreased substantially with time. One exception is the Lop Nor event of March 13, 2003, for which we used data of high signal-to-noise ratio from stations in Eurasia to obtain a reliable CMT focal mechanism and to identify it positively as an earthquake. E Kaz= eastern Kazakhstan, NZ= Novaya Zemlya.

4 6 Figure 1 5 E Kaz India Pakistan 10 kt Lop Nor Pakistan N Korea N Korea 1 Magnitude, mb 4 Lop Nor Ukraine N Korea NZ Iraq NZ India NZ.01 2 Nuclear Explosions Unidentified Not Identified Kola NZ NZ.001 kt Alleged Nuclear Explosions 1 "Nuclear Explosions " Date

5 6 Figure 2 CCD 72 5 E Kaz Urumqi 10 kt Pakistan Lop Nor Magnitude, mb 4 Kara N Korea Qinghai S. India NTS Lop Nor NZ Kara Sea NZ NZ.01 NZ 2 NZ Kara Kola.001 kt Earthquakes Identified After Special Study Not Identified Date

6 6 Figure 3 Germany 5 Wyoming 10 kt Urals Germany Wyoming 1 Magnitude, mb 4 3 Utah NY E Kaz Kursk E Kaz Colorado Utah Spitzbergen 0.1 Kola NTS.01 2 Chemical Explosions Identified.001 kt Mine Collapses Identified NTS Collapse Identified Date

7 We discriminate seismic events as earthquakes based on well-determined depths, focal mechanisms (CMTs), values of mb-ms, clear first motions, and the presence of foreshock/aftershock sequences. The ability to calculate a well-determined CMT (Centroid Moment Tensor) solution, which utilizes broadband seismic body and surface waves, can reliably identify an event as an earthquake. The results of CMT solutions are often superior to those of the older mb-ms technique, which uses a narrow and sometimes noisy range of periods of Rayleigh (but not Love) waves. CMT solutions can be obtained for moderate-size events as we illustrate for two Lop Nor earthquakes of 2003 of mb 4.0 and 4.3. CMT solutions for even smaller events are now available for earthquakes near the Nevada Test Site (NTS). Very small events generally require an analysis of spectral ratios of high-frequency regional seismic waves, which is the subject of a companion paper by Kim et al. (2009). Since earthquakes will occur in the future, their identification will continue to constitute a main task of insuring compliance with the CTBT. Summary of Events Identified Table 1 lists in red 34 seismic events identified as either earthquakes or the North Korean nuclear explosion of They are classified by their seismic magnitude mb for the 6 test sites studied. All events were identified except for 4 in Pakistan (in black), the largest of which was mb 4.1. Identification at mb > 3.3 for 5 test sites corresponds to a yield threshold of a small fraction of a kiloton if those events had been well-coupled explosions. The discrimination limit for Pakistan corresponds to about one kiloton well coupled. Identification is even better, about mb 2.5, for the Russian Arctic test site at Novaya Zemlya.

8 Table 1: Number Seismic Events, , of IDC mb within 100 km Test Site mb> Totals Lop Nor Pakistan NTS N. Korea Russian Rep India Totals Red =Identified: Earthquakes +N. Korean Explosion Black =Not Identified

9 DISCUSSION OF INDIVIDUAL TEST SITES Lop Nor The region within 100 km of the Lop Nor test site accounted for more than half of the total number of seismic events reported by the IDC for the 6 test sites we examined. Nevertheless, all of the Lop Nor events larger than mb 3.3 were identified as earthquakes. The Chinese test site is located near the southeastern side of the Tien Shan (Matzko, 1994), a region of moderate earthquake activity and contemporary horizontal compression where some of the plate motion between India and Eurasia is absorbed. The four focal mechanism (CMT) solutions (Fig. 4) are characterized by large components of reverse faulting and horizontal compression, as is found farther northwest along the Tien Shan. In contrast, no seismic activity was located in the triangular southwestern one-third of the area of Fig. 4 within the old and rigid lithosphere of the Tarim block. While no earthquakes as large as mb 5.0 occurred near Lop Nor from 2000 to 2008, four took place there between 1987 and The IDC did not report any measurements of surface waves that could be used to identify the event of 2003/03/13 as an earthquake by the mb-ms technique. It was the largest event (IDC mb 4.3) within 100 km of Lop Nor from 2000 to Selby et al. (2005) identified the event as an earthquake but with some equivocation. We obtain a reliable CMT (Dziewonski et al., 1981; Ekström et al., 2005) with a mechanism and depth very similar to theirs that we conclude clearly identifies it as an earthquake. Fig. 5 shows long-period seismograms of both Rayleigh and Love waves at two of the stations we used. For discrimination, CMT solutions are superior to

10 Fig. 4. Epicenters of seismic events within 100 km of the Lop Nor test site in northwestern China as reported in Reviewed Event Bulletins (REB) of the IDC. Triangles denote two events of mb 3.6 that were located beyond that distance by the IDC but were subsequently reported by the ISC and Beijing (BJI) as within it. Red circles denote locations of selected large earthquakes from 1987 through Sites of tunnels and shafts used for nuclear tests from Waldhauser et al. (2004). Four balloons in color indicate focal mechanisms (CMTs) with depths determined by that technique. Black numerals beside events denote depths calculated from depth phases pp and sp either by us, the ISC, Selby et al. (2005) or Heyburn and Bowers (2008). Note that all well-determined depths of earthquakes are between 5 and 35 km, much deeper than past underground nuclear explosions throughout the world. Fig. 5. Examples of surface waves used to determine CMT mechanism for event of 2003/03/13 near Lop Nor, China. Data seismograms deconvolved to displacement and filtered between 40 and 100 s, are shown in blue; model seismograms in red, with the inversion window indicated by triangles. Depth for this solution was fixed at 5 km, consistent with the results of Selby et al. (2005), and similar to unconstrained CMT inversion result of 13 km. Retrieved focal mechanism is shown in upper right.

11 42.5 La-tude N REB BJI&ISC, 2002&2004 ISC Tunnels Sha;s Longitude E

12

13 measuring mb-ms. The latter method uses only Rayleigh waves of a narrow range of periods in determining Ms and does not provide an estimate of event depth. CMT solutions often utilize the period band 25 to 100 seconds where earth noise is low. Table 2 depicts several of the techniques by which all of the seismic events within 100 km of Lop Nor between 2000 and 2008 were identified as earthquakes. All were identified by Kim et al. (2009) from the ratio of high-frequency P to Lg seismic waves at regional distances. Many also were identified from depths computed from the seismic waves pp and sp as were all of those listed in Table 2 between 1987 and Heyburn and Bowers (2008) calculated depths from those waves for many of the events between 1987 and A depth significantly greater than about 3 km indicates that an event cannot be an underground nuclear explosion. Only two events between 2000 and 2008 were identified as earthquakes based on what we regard as reliable reports of mb-ms by the IDC and ISC (International Seismological Centre). This likely occurred since Ms as measured traditionally between periods of about 18 to 21 s is difficult to detect in the presence of noise for events smaller than about mb 4. Also, Ms at those periods is likely to be smaller for earthquakes with depths greater than about 20 km than it is for shallower earthquakes, as in Nevada. The first motions of P waves from earthquakes with reverse-faulting mechanisms, as at Lop Nor, typically are compressional (upward) at stations at large (teleseismic) distances. In that case a seismic event cannot be discriminated as either an explosion or an earthquake. With the advent of data from stations at short distances, however, clear dilatational (downward) first motion of P is observed sometimes. This occurred for 7 of the events in Table 2 at the nearby station Urumqi (WMQ). Such an observation indicates that the event cannot be an explosion.

14 Table 2: IDENTIFICATION OF LOP NOR SEISMIC EVENTS AS EARTHQUAKES Date mb mb CMT depth from Earthquake? **Hi freq P/Lg Down 1st IDC ISC depth pp&sp mb-ms regional waves motion P /13/ Yes/5 km Yes/6 km Yes/0.3* to 0.7 Yes Yes, Pg 12/19/ Yes 11/18/ Yes Yes 9/9/ Yes/12 km Yes 1/22/ Yes/22 km Yes/29 km Yes/0.3 Yes Yes 3/15/ Yes Yes 3/23/ Yes 8/25/ Yes Yes, Pg *10/1/ Yes *10/1/ Yes/35-60 Yes 8/5/ Yes 8/16/ Yes Yes 2/4/ Yes/29 km Yes 8/22/ Yes 2/13/ Yes/32 km Yes 5/17/ Yes 9/21/ Yes 6/15/ Yes 10/11/ Yes Yes 5/11/ Yes 4/9/ Yes 7/8/ Yes/33 km Yes /22/ Yes/19 km Yes/18 km Yes/0.5 1/30/ Yes/23 km Yes/24 km Yes/0.2 to /27/ Yes/8 km Yes/0.4 11/3/ Yes/20 km Yes/0.6 11/15/ Yes/16 km Yes/0.6 3/20/ Yes/23 km Yes/0.7 10/18/ Yes/16 km Yes/0.6 1/27/ Yes/20 km Yes/0.5 5/1/ Yes/17 km **Kim et al. (2009) *Doublet *Mw=4.4 not Ms 57 sec apart

15 Note that two Lop Nor events were identified by 5 different methods. Matzko (1994) concludes from their similar geological and geophysical properties that the Lop Nor site and the former test site in eastern Kazakhstan are characterized by efficient propagation of seismic P to large distances. Explosions in boreholes were conducted below the shallow water table and hence were well coupled. The mb-log (yield) relations developed for eastern Kazakhstan indicate that mb 3.4, the smallest seismic events indentified near Lop Nor, corresponds to a well-coupled explosion with a yield of about 0.03 kiloton (30 tons). For explosions above the water table that yield could be larger, about 0.1 kt. Such an event or one detonated in alluvium (uncompacted materials like sand and gravel) may well leak Xenon isotopes to the surface. Pakistan Fig. 6 shows 9 seismic events of mb > 3.3 from 2000 to 2008 that occurred within 100 km of the two Pakistani test sites used in Only Lop Nor had more seismic events, all of which were identified as earthquakes, during the same period and distance (Table 1). Pakistan was the only one of the six regions for which all of the events we studied could not be identified as earthquakes. This is attributed to India, Pakistan and some other surrounding countries not contributing seismic data to the IDC from stations on their territories. The Pakistani test sites are situated within the upper plate of the Makran subduction zone (Jacob and Quittmeyer, 1979). Its tectonic attributes, including the presence of volcanic centers, likely result in greater attenuation of P waves to distant stations. Earthquakes in the southern part of Fig. 6 probably occurred at depth within the shallow-dipping Arabian plate and are characterized by nearly horizontal

16 Fig. 6. Epicenters of seismic events from 2000 to 2008 within 100 km of two Pakistani test sites used in CMT solutions and depths of larger earthquakes from 1980 to 1999 are shown in red and those for two events in 2004 and 2005 in blue. Depths from pp and sp denoted in black. Note that all well-determined depths are between 11 and 61 km. Volcanic centers of Makran subduction zone from Jacob and Quittmeyer (1979).

17 La-tude N Pakistan FIG. 6 REB ISC Nuclear Explosions Volcanic Centers Longitude E

18 extension. Two of the CMT mechanisms farther north are characterized, however, by horizontal compression; they likely occurred within the upper (Eurasian) plate. The Eurasian-Indian plate boundary is located farther east than Fig. 6. Six events were identified as earthquakes by their CMT solutions. The seismic phases pp and sp indicate that the depths of 5 events were between 11 and 54 km, discriminating them as earthquakes. Several seismic events were identified from Ms-mb values (Table 3). We have not yet examined the short-period ratio P/Lg or searched for stations recording dilatational (downward) first motion, which may identify additional events. We note that data from seismic stations in Afghanistan, Oman and the United Arab Emirates have become available recently. Nevada Test Site Fig. 7 shows 4 seismic events of mb > 3.3 from 2000 to 2008 that occurred within 100 km of various sub-areas that were used for nuclear testing at NTS. Those and 4 other seismic events between 1992 and 1999 were identified as earthquakes based on their CMT focal mechanisms. Several were identified based on their small mb-ms values as compared to those of underground nuclear explosions (Table 3). Walter et al. (2004) identified several seismic events within several hundred kilometers of NTS from the ratio P/Lg at 6 to 8 Hz. Unlike Lop Nor and Pakistan, earthquakes near NTS are quite shallow and hard to identify from their depth phases pp and sp. CMT solutions by Herrmann (2009) also identify as earthquakes 4 events of about mb 3.0 between 2000 and Reliable depths computed by the University of Nevada at Reno likely identify 2 other events of mb 3.0 and 3.2 as earthquakes.

19 Fig. 7. Epicenters of seismic events from 2000 to 2008 within 100 km of subregions of Nevada Test Site (NTS). CMT solutions of larger earthquakes from 1980 to 1999 are shown in red; CMT depths in black. CMT solutions determined by BRK (Berkeley), GCMT (Global Centroid Moment Tensor project, e.g. Ekström et al., 2005) and SLM (Herrmann, 2009).

20 Nevada Test Site Region Fig CMT SLM CMT SLM 6 KM CMT SLM 15 KM GCMT 8 KM CMT BRK 15 KM GCMT CMT LLNL CMT SLM 6 KM CMT SLM 11 KM La,tude N 36.5 REB mb >=3.4 Nuclear Test Areas Las Vegas Larger Eqs REB mb < Longitude W

21 TABLE 3: IDENTIFICATION OF SEISMIC EVENTS AS EARTHQUAKES, mb > 3.3 Date mb mb CMT depth from Earthquake? Hi freq P/Lg IDC ISC depth pp&sp mb-ms regional waves PAKISTAN Yes/45 km Yes/48 km Yes/46 km Yes 0.5 to Yes/61 km Yes Yes/54 km in Afghanistan Yes/43 km Yes/ Yes/39 km Yes/33 km Yes/23 km Yes/ Yes/18 km Yes/11 km Yes/-0.3 to 0.1 NEVADA TEST SITE Yes/ 8 km Yes/ Yes/ 6 km Yes, Walter et al Yes/11 km Yes/0.2 to Yes/ 6 km Yes/ Yes/ 15 km Yes/ Yes/ 11 km Yes/ Yes Part sequence of more 4 events Yes/ 15 km Yes/ -0.4

22 All seismic events near NTS of magnitude mb > 3.3 from 2000 to 2008 are identified as earthquakes as are many of mb 3 to 3.3. NUCLEAR TESTING OF MILITARY SIGNIFICANCE We have shown that seismic events can be identified with high confidence at Lop Nor for mb > 3.3. The identification level for NTS is at least as good, perhaps mb 3.0. That for the Russian test site at Novaya Zemlya is about 2.5 (Sykes, 2000). These correspond to yields in the 10 s to ~100 of tons of nuclear yield (0.01 to 0.1 kilotons). Fig. 8 shows Kidder s (1985) histogram of the frequency of U.S. nuclear tests at NTS as a function of yield. Its most prominent peak occurs between 7 and 20 kt, indicative of the high military significance of testing at those yields. The yields of past Russian tests have a pronounced peak near 20 kt. About 5% of U.S. tests, mostly effects tests, were at yields below 1 kt, indicative of their low priority. Identification for Lop Nor, Novaya Zemlya and NTS is now comparable to even the smallest test in Fig. 8. In fact, estimating yields below 1 kt becomes increasingly uncertain since the number of well-calibrated nuclear events is very small. Remarks. Few papers, especially recent ones, are available in the open literature on the discrimination of seismic events as either earthquakes or nuclear explosions. We encourage seismologists to work on and publish analyses of problem events, especially those near nuclear test sites. If qualified professionals do not do so, we fear that leaks to the press and blogs will fill the gap, sometimes with highly erroneous identifications and misinformation. Much to their credit, the British group that works on verification has long published detailed analyses of many problem seismic events. Seismic data from hundreds of stations around the world are now openly available in real or near real time. Timely access to seismic data from the IDC would further this.

23 1.2 DISTRIBUTION OF EXPLOSIVE YIELDS AT NTS: 1980 THROUGH 1984 Fig df(y)/d(logy) HEDF Yield (kt) TTBT

24 Fig. 8. Histogram of numbers of nuclear explosions at NTS as a function of yield in kilotons from (Kidder, 1985).

25 Acknowledgments. We thank W.-Y. Kim for information on short-period seismic waves and first motions for events near Lop Nor. R. Herrmann and W. Walter responded very quickly to our request for information on seismic events near NTS. We thank G. Ekström, P. Richards and W.-Y. Kim for reviewing the manuscript References Dziewonski, A. M., T. A. Chou, and J. H. Woodhouse (1981). Determination of earthquake source parameters from waveform data for studies of global and regional seismicity, J. Geophys. Res., 86, Ekström, G., A. M. Dziewonski, N. N. Maternovskaya, and M. Nettles (2005). Global seismicity of 2003: centroid-moment-tensor solutions for 1087 earthquakes, Phys. Earth Planet. Inter., 148, Herrmann, R. (2009). Focal mechanisms of North American earthquakes obtained from broadband data, Heyburn, R., and D. Bowers (2008). Earthquake depth estimation using the F trace and associated probability, Bull. Seismol. Soc. Amer., 98, Jacob, K.H., and R.L. Quittmeyer (1979). The Makran region of Pakistan and Iran: trench-arc system with active plate subduction, in Geodynamics of Pakistan, A. Farah and K.A. DeJong eds., Geological Survey Pakistan, Quetta, pp Kidder, R. (1985). Militarily significant nuclear explosive yields, FAS Public Interest Report, 38, #7. Kim, W.-Y., P.G. Richards, and L.R. Sykes (2009). Discrimination of earthquakes and explosions near nuclear test sites using regional high-frequency data, this vol. Matzko, J.R. (1994). Geology of the Chinese nuclear test site near Lop Nor, Xinjiang Uygur autonomous region, China, Engineering Geology, 36, Selby, N.D., D. Bowers, A. Douglas, R. Heyburn, and D. Porter (2005). Seismic discrimination in southern Xinjiang: the 13 March 2003 Lop Nor earthquake, Bull. Seismol. Soc. Amer., 95, Sykes, L.R. (2000). False and misleading claims about verification during the Senate debate on the comprehensive nuclear test ban treaty, FAS Public Interest Report, 53, #3, 12 pp.

26 Sykes, L.R. (2002). Four Decades of Progress in Seismic Identification Help Verify the CTBT, Eos Trans. Amer. Geophys. Union, 83, Waldhauser, F., D. Schaff, P. G. Richards, and W.-Y. Kim (2004). Lop Nor revisited: underground nuclear explosion locations, , from double-difference analysis of regional and teleseismic data, Bull. Seismol. Soc. Amer., 94, Walter, W.R., K.D. Smith, J.L. O Boyle, T.F. Hauk, F. Ryall, S.D. Ruppert, S.C. Myers, R. Abbot, and D.A. Dodge (2004). An assembled western United States dataset for regional seismic analysis, Lawrence Livermore National Laboratory, UCRL-TR , 22 pp.