Did North Korea Conduct a Secret Nuclear Test in 2010?

What does the evidence tell us? 

Zhang and Wen, working from the second De Geer paper, examined a wider range of dates. They identified a tiny event on May 12 using a method that looks for tiny events by matching (or cross-correlating) very small deviations at multiple regional seismic stations. This method should produce a lot of spurious correlations and the Zhang and Wen paper is a little vague about how many standard deviations the event in question represents. But in one study that looks at earthquakes using this method, even nine standard deviations above the mean resulted in something like one spurious correlation a day. This is some serious data mining.

That said, the Zhang and Wen event is still interesting. It occurred in the morning of May 12. This event is consistent with Ihantola et al’s estimate of a later event around 16:00 UTC on May 12, and is just inside DeGeer’s confidence interval, too.

Now, here is a problem. The event identified by Zhang and Wen probably did not occur until several hours after the DPRK released its statement about fusion. Moreover, the original DPRK announcement indicated that the fusion event had occurred on April 15. On balance, if the May 12 event was a DPRK nuclear explosion, it does not appear to be related to the announcement of a successful fusion event earlier in the day and referring to an event in April. In other words, the fusion announcement that De Geer emphasized so heavily in his paper was a coincidence. And, if there is one theme that I keep coming back to over and over again, it’s that we have to be cautious about building a case by collecting coincidences.


There is another problem that is worth pondering. The Zhang and Wen paper posits an Mb of 1.44. It is not straightforward to convert this to a nuclear yield for an explosion, although Zhang and Wen use a formula that yields (sorry about the pun) an estimate of about 3 tons. That’s a very small event. It isn’t clear why North Korea would conduct such a small test.

Although De Geer did not stipulate the size of the event necessary to produce the radionuclide signatures, others suggest the explosion must have been on the order of several tens of tons, if not more. That has lead proponents to argue that North Korea might have conducted a May 2010 explosion in a giant cavity that decoupled the seismic signal from the size of the explosion. Decoupling factors in hard rock could be as large as a factor of 40, transforming a 3 ton event into a 120 ton event. Constructing a cavity in hard rock large enough to decouple a 120 ton explosion would be quite an engineering achievement in hard rock. Moreover, construction of such a large cavity would surely have been noticed. This is yet another complication in the story, one that is plausible yet also unlikely.


There is some circumstantial evidence. To me, though, it just doesn’t hang together. The scenario in the original De Geer paper has been completely abandoned. If there was a test, it was one event, not two. And if there was a test, it occurred much later than De Geer initially thought, making the DPRK announcement a coincidence.

What we are left with are some interesting radionuclide readings, but it is possible to imagine alternative explanations for them. And, to my frustration, we haven’t seen a careful examination of those alternatives. Instead, there has been a tendency to build a case against the DPRK. Collecting coincidences makes me nervous.

I still think it is possible that the DPRK did, in fact, conduct a nuclear test in May 2010. But proving it requires more than just collecting data that corroborates the event, while ignoring alternative hypotheses and data that doesn’t fit.

Writing about the September 22, 1979 event, Pief Panofsky concluded the best description of the evidence was the so-called Scotch Verdict. In Scotland, juries may make one of three findings rather than two—guilty, not guilty and not proven. Panofsky was not quite prepared to acquit Israel or South Africa of having conducted a nuclear test, but nor did the evidence point conclusively to their guilt.

“Not proven” seems to be the right verdict in the case of the May 2010 event as well. It is worth noting, by the way, that there was—or rather could have been—a simple way to determine whether North Korea conducted a low-yield nuclear test in May 2010. If the CTBTO had been in force with North Korea as a member, the United States or any other State Party could have requested that the CTBTO conduct an onsite inspection. Radiochemists are undeniably proud that the CTBTO’s network of radionuclide stations detected hints of a possible nuclear explosion. But the system was never intended to function with sensors alone. The ability to conduct onsite inspections is an essential element in the regime envisioned to verify the worldwide ban on nuclear testing. In November and December 2014, the CTBTO conducted Integrated Field Exercise 2014—a simulated onsite inspection in Jordan. Absent such an inspection—or better evidence than has been found to date—the events of May 2010 remain interesting, but ambiguous.

Jeffrey Lewis is Director of the East Asia Nonproliferation Program at the James Martin Center for Nonproliferation Studies (CNS), Monterey Institute of International Studies.