The Great 2010 DPRK Nuclear Test Debate: Summarizing the Evidence of a Low-Yield Nuclear Test Carried out in North Korea in May 2010

Did they test or didn't they? The debate over whether North Korea conducted a low-yield nuclear test in May 2010 continues. On April 28, 2015 Jeffrey Lewis as well as Joel Wit and Sun Young Ahn posted on 38 North their negative (Lewis) and inconclusive (Wit and Ahn with some additional indications via satellite imagery) views on whether or not the North Koreans actually carried out a low-yield nuclear test in May 2010. As I started this debate with an article in Science & Global Security in late February 2012,[1] with its major message spread by Nature on February 3, 2012, I would like to correct a number of misunderstandings and maybe clarify a little bit more a story that might appear complicated for people not specialized in radionuclide analyses. My analysis was reinforced when new data appeared and when I also considered trapping of the xenon precursors antimony and tellurium in the analysis.[2]

In this article, I will try to describe the evidence below starting from the strongest argument and continue through arguments that some people might consider doubtful, unless they manage to see the whole picture.

Radionuclide Detections

First and foremost, the evidence that a nuclear explosion was carried out underground in East Asia came from detections of barium-140 and lanthanum-140 at a CTBT filter station at Okinawa (Japan).[3] Later, traces of cerium-141 were also noted in re-measurements at certified laboratories. Lanthanum-140 was detected at the CTBT station at Ussuriysk (Russia) as well, which virtually excluded that the source was to be found locally at Okinawa. All three radionuclides are offspring of very short-lived (1.7 and 13.6 second half-life) fission products of xenon and only a nuclear explosion can transport significant amounts of xenon out of a fissile matrix of plutonium or uranium within seconds.[4] Furthermore, none of the radionuclides (most specifically iodines) that always dominate a fission event with free (atmospheric nuclear tests) or fairly free (mishaps/accidents at reactors, criticality experiments or laboratories) access to the atmosphere were detected. That strongly implies there was an effective filter that only let noble gases escape the event. The obvious filter would be the rocks surrounding the test cavity of an underground test.[5] The ratio between the 140 and 141 activities could be used to calculate that it took a little less than ten seconds from the time of the explosion until the gases were injected into the atmosphere, an analysis that further supports the potential scenario.

The story could end here. There is already enough evidence for an underground nuclear explosion, which is why I did not check too many other “possible” sources. I was criticized for that, but later a paper was published by Christopher Wright, which very carefully did this exercise,[6] showing that there were no alternatives.

The second evidence was that longer-lived xenon isotopes (xenon-133 with a half-life of 5.243 days and xenon-135 with a half-life of 9.14 hours) were detected at a national noble gas station at Geojin in the northeastern corner of South Korea and at a CTBT station at Takaski (Japan). Both employ SAUNA (Swedish Automated and Unattended Noble gas Analyser) analysers. Doing the ratio analysis for the Geojin observation showed that the resulting scenario required a clear detection of metastable xenon-133 (xenon-133m), but as there was none, it was difficult to marry the mass 140 and 133/135 detections into a common event. But as the former was globally unique and the latter locally unique for the stations, it was reasonable to further consider that they actually derived from the same source. An official news flash from North Korea around midnight on May 11/12, 2010, reporting about a successful nuclear fusion experiment on April 15, suggested a hypothesis to solve the dilemma. Perhaps there had actually been two tests in the same chamber, one on April 15, the birthday of Kim Il Sung, the founder of DPRK, and one in early May. I speculated that maybe the fusion part did not work in April, but it did one month later, and that for ceremonial reasons, the success in May was stated as having occurred during the national holiday in April. As small nuclear explosions do not utilize fissile fuel very effectively, doing several low-yield tests in a single chamber would make it a virtual mine of nuclear fuel.

The two-test scenario was my biggest mistake from a credibility point of view, as many thought it was too fabricated and not so much in line with the Occam’s razor principle. I can understand that reaction, but I still think it was a valid hypothesis given what I knew at that time, and that precedent exists for nuclear tests to be conducted in the cavity of a previous test. It is not correct to say, as Lewis does, that I have concluded my first paper was in error.

After my first paper, I started to think about fractionation phenomena in a cavity that cools down after an explosion. I could not find any published technical paper on that, but looking at the relevant melting and boiling points, I concluded that the precursor elements antimony and tellurium could well be trapped in the condensing and solidifying rock materials. There were actually indications of that in more general US literature.[7] Due to different dynamics in the 135 and 133 decay chains,[8] the xenon-135/xenon-133 ratio in the residual gas depends on the effective trapping time. In my Journal of Radioanalytical and Nuclear Chemistry article, Figure 4 shows how the isotopic ratio is translated to explosion time and we can see that for the recently estimated seismic explosion time (00:08:45 May 12, 2010 UTC),[9] effective trapping would have happened about half an hour post-explosion (for both uranium and plutonium fission).

Before any positive seismic observation was found, the explosion time was estimated by analyzing the barium-140/lanthanum-140 ratio. That was not a very precise effort as the result was very dependent on the coincidence correction factor for the 487.0 keV line in the decay of lanthanum-140. The different Maximum Likelihood estimates are illustrated in Figure 3 of my Journal of Radioanalytical and Nuclear Chemistry article, where the one utilising most data, with the ratios treated correctly (not Gaussian) and allowing the impact of parameter uncertainties via Monte Carlo techniques, yields an explosion time between 8 am on May 10, 2010 and 2 pm on May12, with a most likely estimate at 5 pm on May 11 (all dates and hours UTC and confidence intervals based on half widths at half maximum).

Atmospheric Transport Modelling (ATM)

Advanced atmospheric transport modelling, both backward and forward in time, is inevitable to narrow down the suspected area for an emission. I initially used the tool, WebGrape, developed at the CTBTO and later Wright and Wotawa contributed their own analyses.[10] The results are sensitive to the assumed explosion time, which can be seen in Figure 5 in my Journal of Radioanalytical and Nuclear Chemistry article, where moving the estimated explosion time from afternoon May 11 UTC to midnight May 11/12 actually improves the correlations for the known DPRK nuclear test site. After the seismic analysis was published, Wotawa did some forward modelling that very well fit the arrivals of the clouds at all four stations involved.[11] Common for all modelling is, however, that one must assume the xenon-133 and xenon-135 emissions to happen about 30 hours after the explosion. This is consistent with the idea that this was an operational filtered ventilation to prepare for re-entering the tunnel/cavity.

Seismic Detection and Yield Estimates

In 2012, Schaff, Kim and Richards from Columbia University published a search of seismic signals in the time windows I had defined in my first paper.[12] It was based on data from the Mudanjiang Observatory 370 kilometers north of the North Korean test site and it put an upper magnitude limit on an undetected event between 1.4 and 1.7 (translated to 1-2 ton). Earlier this year, two other well respected seismologists, Miao Zhang and Lianxing Wen, published results from seven regional stations along the Chinese-DPRK border up to four times closer to the test site than Mudanjiang.[13]  With improved sensitivity, Zhang and Wen conclude there was a nuclear explosion at 00:08:45 on May 12 UTC with a magnitude of 1.44 ± 0.13, consistent with the Schaff et al. upper limit. The Chinese, however, translated this into an (apparent) yield of 2.9 ­± 0.8 ton.

With an advanced analysis close to the detection limit, it is natural that the Zhang and Wen conclusions are discussed in the seismic community, and there is a very recent paper by Sean Ford and William Walter at Lawrence Livermore National Laboratory[14] that confirms the Zhang and Wen detection, but argues that either the estimated yield is lower than 1 ton or the Zhang and Wen localization is off by several kilometers. Unfortunately they present their results having already converted from seismic magnitude to tons, making a comparison with Zhang and Wen difficult given the propensity for different authors to use different empirical transformations. Ford and William also question the conclusion that the explosion was nuclear and find a similar seismic detection from the test site lacking any radionuclide signature on June 6, 2010 to support that. I have also heard that a few papers on the May 2010 event will be presented at the CTBT “Science and Technology 2015” conference in Vienna in late June.

Something not yet openly discussed by seismologists in this case is “decoupling.” If a test is decoupled, its apparent yield can be many tens of times lower than the real yield. The fact that the 2010 event injected a very unique beam of short-lived xenon isotopes into the atmosphere is quite strong evidence that the test was decoupled (by being executed in a mined cavity). For a low-yield test of e.g. 100 ton, such a chamber does not have to be enormous. Literature reports a 30,000 m3 cavity to decouple 1 kiloton and as it scales with the yield, a 3,000 m3 chamber would be suitable to decouple a 100 ton test.[15] This is not so difficult to dig out without being especially noted in an active mining area. As a decoupled explosion melts and vaporizes much less of the surrounding rocks than a well coupled one, much more energy goes to instantly build very high pressure in the cavity gas that, in turn, pushes the hot debris through cracks and fractures in the rock. As the distance is of the order of hundreds of meters, the debris will cool down before it reaches the outside and all except the noble gases will stick to the walls of the cracks.

The unique 2010 prompt noble gas emission could have been due to inexperience with decoupled low-yield testing. It is, of course, fully possible that DPRK has carried out more low-yield tests since then, but is more careful to prevent dynamic leaks. A corresponding learning curve was seen in the sequence of the three higher-yield tests, where in 2006, xenon was detected, in 2009, no xenon was observed, and in 2013, nearly two months delayed releases were detected.

Radionuclide observations from a foreign nuclear test can generally not be used to infer the explosive yield. With modern ATM calculations, however, the dilution of the nuclides between release and detection can be estimated and from that, the size of the release can be expressed in Bq or in tons of fission yield. Wotawa reported a 0.2 EBq prompt release of xenon-140[16] and an approximately 100 TBq release of xenon-133 some 30 hours later. This corresponds to a prompt leak of 1 ton (uranium fission) or 3 ton (plutonium fission) and a delayed release of about 20 ton (both U and Pu). Broadly these numbers indicate a lower limit on the yield of the explosion and as the leaks certainly are only some fractions of the total yield, the latter can well have been in the range of 100 tons.

At around 100 ton, it is reported in the literature that useful experiments can be done to develop boosters, which are necessary to minimize the physical size of the devices.[17] It is widely described in the press that the DPRK needs to make their charges fit the missiles they have at hand. And note that in February 2013, they made a test where they said they had succeeded in making that device physically small.

The Telegram

The North Korean news flash around midnight May 11/12 UTC about a successful fusion experiment (on April 15, 2010, but assumed to refer to the May event) has been the source of some confusion as both Lewis and Wit and Ahn claimed that the telegram came several hours before the Zhang and Wen dating of 9 minutes past midnight. I have my dating from a reporter at Reuters who saw the telegram arrive on his screen. With a very short time between experiment and telegram the latter must have been prepared in advance, which appears somewhat odd. But so is the telegram itself, being perhaps the most exotic amongst the many curious statements that periodically come out of DPRK. And if the April 15 alternative were true, it is also odd to announce something significant almost a month after it occurred, especially as its propaganda value would very well fit the celebrations on “The Day of the Sun.” Although it would be a staggering coincidence if the news flash and the May nuclear explosion were not related, the telegram should, with all its unknown details, be considered the weakest link in the chain of arguments summarized here, which implies that a low-yield underground nuclear test was actually carried out at the North Korean test site in early May 2010.

Lars-Erik De Geer (Sweden) is a nuclear physicist retired from the Swedish Defence Research Agency (FOI) and the Comprehensive Nuclear-Test-Ban Treaty Organization (CTBTO) in Vienna. He has had a long career in detecting and analyzing radionuclide signatures from nuclear weapon tests and radiological accidents.


[1] Lars-Erik De Geer, “Radionuclide Evidence for Low-Yield Nuclear Testing in North Korea in April/May 2010,” Science and Global Security 20, no. 1 (2012): 1-29.

[2] Lars-Erik De Geer, “Reinforced evidence of a low-yield nuclear test in North Korea on 11 May 2010,” Journal of Radioanalytical and Nuclear Chemistry 298, no. 3 (2013): 2075-2083.

[3] Lars-Erik De Geer, “Radionuclide Evidence for Low-Yield Nuclear Testing in North Korea in April/May 2010,” Science and Global Security 20, no. 1 (2012): 1-29.

[4] Richard Garwin, a world famous authority in the field, agreed with this reasoning when I spoke to him on the telephone before my first paper was published.

[5] This has not to my knowledge been seen before, even though I have more than 40 years of experience in global nuclear test detection by a Swedish sampling network that several times picked up leaks from the nuclear test sites. Twice we saw signatures from Soviet underground tests, where barium/lantahanum-140 were just enhanced compared to other fission products because their xenon precursor escaped fallback ejecta from shallowly buried engineering tests (PNEs). The detection of solely lanthanum-140 that Jeffrey Lewis mentions was also detected in Sweden and I was part of the effort to find the source, Bourges (France), where lanthanum-140 without any barium-140 produced by irradiation in a reactor was used for decontamination experiments.

[6] Christopher M. Wright, “Low-Yield Nuclear Testing by North Korea in May 2010: Assessing the Evidence with Atmospheric Transport Models and Xenon Activity Calculations,” Science and Global Security 21, no.1 (2013): 3-52.

[7] Panel on Basic Research Requirements in Support of Comprehensive Test Ban Monitoring, National Research Council (1997) Research Required to Support Comprehensive Nuclear Test Ban Treaty Monitoring, Committee on Seismology, Board on Earth Sciences and Resources, Commission on Geosciences, Environment, and Resources, National Academies Press, Washington, DC,; and Radionuclide experts from six US Organizations: Lawrence Livermore National Laboratory, Los Alamos National Laboratory, Pacific Northwest Laboratory, Air Force Technical Applications Center, Environmental Measurements Laboratory, and National Oceanic and Atmospheric Administration, Source Term Review. A Report of the Peer Review of the Conference on Disarmament International Monitoring System Expert Group, CD/NTB/WP.224 Part II, 1996.

[8] See figure 2 in Lars-Erik De Geer, “Radionuclide Evidence for Low-Yield Nuclear Testing in North Korea in April/May 2010,” Science and Global Security 20, no. 1 (2012): 1-29.

[9] Miao Zhang and Lianxing Wen, “Seismological Evidence for a Low-Yield Nuclear Test on 12 May 2010 in North Korea,” Seismological Research Letters 86, no. 1 (January/February 2015): 1-8.

[10] Christopher M. Wright, “Low-Yield Nuclear Testing by North Korea in May 2010: Assessing the Evidence with Atmospheric Transport Models and Xenon Activity Calculations,” Science & Global Security 21, no.1 (2013): 2-52; Gerhard Wotawa, “Meteorological analysis of the detection of xenon and barium/lanthanum isotopes in May 2010 in Eastern Asia,” Journal of Radioanalytical and Nuclear Chemistry 296, no. 1 (2013): 339-447.

[11] Gerhard Wotawa, “Updated meteorological analysis on a possible nuclear test in DPRK in May 2010, New insights into the May-2010 RN detections,” Presentation at the MARC X conference in Hawaii, April 13, 2015.

[12] David P. Schaff, Won-Young Kim, and Paul G. Richards, “Seismological Constraints on Proposed Low-Yield Nuclear Testing in Particular Regions and Time Periods in the Past, with Comments on ‘Radionuclide Evidence for Low-Yield Nuclear Testing in North Korea in April/May 2010’ by Lars Erik De Geer,” Science and Global Security 20, no. 2-3 (2012): 155-171.

[13] Miao Zhang and Lianxing Wen, “Seismological Evidence for a Low-Yield Nuclear Test on 12 May 2010 in North Korea.” Seismological Research Letters 86, no.1 (2015): 1-8

[14] Sean R. Ford and William R. Walter, “International Monitoring System Correlation Detection at the North Korean Nuclear Test Site at Punggye-ri with Insights from the Source Physics Experiment,” Seismological Research Letters 86, no. 4 (2015)

[15] US Department of the Interior, US Geological Survey, Geologic and engineering constraints on the feasibility of clandestine nuclear testing by decoupling in large underground cavities, by Leith W, Open-file report 01-28, US Geological Survey (Reston, Virginia, 2001),

[16] 1 TBq barium-140 corresponds to 0.2 EBq xenon-140.

[17] T. B. Cochran and C. E. Paine, “The Role of Hydronuclear Tests and Other Low-Yield Nuclear Explosions and Their Status Under a Comprehensive Test Ban,” Nuclear Weapons Databook (Washington, DC: National Resources Defense Council, 1995).

Stay informed about our latest
news, publications, & uploads: