Andreas PersboGreetings from Geneva: More on Mb to Yield

As always when the DPRK tests, I’m in a seminar somewhere else. This time, I was discussing FMCT verification with the good people of SIPRI and the diplomats of the Conference on Disarmament.

The plot is based on the Mb to yield estimates for dry-rock, close coupled, underground nuclear explosions. I’ve used the USGS estimate for the yield calculation. The graph can only be used as a general indication because the exact geological conditions of the area is not known. It is known to have a shallow water table, which could explain why the Russians always get their yield calculations on the high end. We also don’t know how well this test was coupled to the rock.

However, as indicated, today’s North Korean test seems to be significantly bigger than its previous test. What is also interesting is that the centre of today’s test, according to the USGS, is about 5.5 kilometers away from the old test site. The IMS puts the test closer to the first site, but within the USGS margin of error. The USGS sets the error to +/- 3.8 kilometers, which strongly suggest that we’re looking at a second test site in close proximity to the first.

Interestingly, a couple of years ago I learned that the South Koreans were looking for test site preparations on the “northern side of the mountain” relative from the first site. This, to me, means that the second site has been known to the South for some time.

I’m now off to have some dinner.


  1. Tom B.

    Is there a reference for the equation underlying the graph?



  2. Azr@el (History)

    I wonder if they timed this to coincide with my holiday; If only the mountain had wifi. It seems the DPRK is keen to restore both “face” and deterrence in the wake of orbiting Unha-2 below sea level. No surprise, everyone knew this was coming…what is surprising is the complete lack of U.S. engagement on the matter; how many 100 day blocks does Obama need to transform his rhetoric of engagement into tangible diplomacy.

  3. Major Lemon (History)

    The yield is significant but it was reported somewhere that the physical size precludes it’s use as a deliverable missile warhead.

  4. Andreas Persbo


    1 mb = 4.10 + 0.75 log Y

    Ref. Terry Wallace, The May 1998 India and Pakistan Nuclear Tests, Seismological Research Letters, September 1998

  5. JF (History)

    Actually the equation is
    M = a + b log Y
    where a and b are empirically determined constants

    for Novaya Zemlya, the eqn would be
    M = 4.45 + 0.75 log Y
    this gives a yield Y of 2.2 kt

    for Nevada Test Site, it would be
    M = 3.92 + 0.81 log Y
    this gives a Y of 9.2 kt

    Nobody has calibrated the North Korean test site and so the empirical constants can only be guessed (or chosen to give the yield you wish to claim)

    Andreas: Which was my point exactly.

  6. Daniel Pinkston (History)

    The ROK knew about the second tunnel entrance or test site at least around the time of the first test in 2006. Chŏng Hyŏng-gŭn, a former Grand National Party (GNP) National Assemblyman (2004-2008) who had been a high-level official in the National Intelligence Service (NIS) when it was still called the “Agency for National Security Planning” wrote about it on his website in 2007. Chŏng claimed that after the 2006 test three buildings near the “eastern” tunnel entrance were demolished and the North Koreans dug a 95-meter long tunnel near the entrance and filled it back in. Chŏng wrote that the excavation could have been conducted to remove cables that were used to monitor the first test, but he speculated that the work possibly could have been related to preparations for a second test.

    I think there might be some confusion over the orientation of the tunnel entrances. I believe the actual orientation is north and south, but the South Koreans usually reference them as being east and west. But hey, if you look on a map, the Ch’ungch’ŏng Provinces are actually east and west, but they are called North Ch’ungch’ŏng Provonce and South Ch’ungch’ŏng Province, so go figure…

  7. Andreas Persbo

    Just to clarify. I chose Wallace’s estimation simply because I believe that conditions in DPRK are closer to Pakistan than to Nov. Zem. But as JF pointed out, it’s mostly guesswork since the DPRK site has not been calibrated. And this, of course, applies to ALL yield estimations that you’ll see in the coming days.

  8. JF (History)

    Indeed, if one uses the MOS reported magnitude M of 5.0 instead of the M of 4.7 from the USGS, the constants from NZ give a yield of 5.4 kt while the NTS constants give a Y of 21.5 kt.

    So the choice of M and the two empirical constants can give any yield from 2 all the way to 21 kt

    I would be cautious about assuming Pakistani constant values (non-empirical and themselves assumed from Kazakh constants). After all, Pakistan lies on a different tectonic plate to the rest of Asia. Why choose Kazakh rock instead of Nevada rock? Who has studied the geology of the test sites?

  9. Andreas Persbo

    Well. JF. I don’t want to state the obvious, but the DPRK site is obviously not sand, right? You’ll also notice that most hard rock estimates fall out in the same yield range (as you yourself pointed out in the NZ calculation).

    I had my reasons for choosing the Pakistani formula. But, anyway, I think it’s fair to say that the yield was somewhere between 2-3 kT. I would not rule out 4 kT.

    However, I would be quite surprised if they achieved 5 kT and gobsmacked if it were 21 kT (you simply cannot use the NTS formula – ground conditions are way different).

    Goodnight from Geneva, everyone.

  10. Pedro (History)

    3Kt,5Kt or 20Kt is a moot point if they have Tritium. With a few grams of Tritium you can fairly easily make a ‘feeble’ 3Kt bomb into a 20Kt bomb. Do our NORK friends have Tritium? (I’ve never seen that question answered)

    Andreas: I totally agree. But to the best of my knowledge, the DPRK does not have any facilities capable of producing significant quantities of tritium. Anyone else have anything to add?

  11. Anon.

    Is the tritium boosting as “trivial” even if it is a solid pit desing, e.g. Fat Man clone? Is it even possible at all in that case? Is it possible to have that knowledge and be allowed to talk about it?

  12. FOARP (History)

    Well, the amount of Tritium-deuterium gas needed to boost a single weapon is in the single-digit grams. Can commercially produced/used Tritium (which is produced in the 100s of grams per year) be easily adapted for this purpose? How easy is it to get hold of commercial tritium?

  13. George William Herbert (History)

    Facilities for producing tritium?

    You mean, a reactor, and a “fuel rod” full of Lithium?

    It’s not that hard… And they had an operating reactor for years…

  14. George William Herbert (History)

    Fat Man was only semi-solid pit.

    It had that small spherical hole in the middle, to fit the Urchin Polonium-Beryllium neutron generator.

    One could easily use that as a reservoir for boost gas instead and use external neutron generators.

    I don’t know why you’d think that would be so sensitive.

  15. sun bin (History)

    1) yes the equation is only a log fit. the constants depends on local rock structure/etc. and even then it is an approximation.

    in theory the constants are really fixed(universal) is x=energy, in practice the energies (Ritcher scales) are “measured”. so you are talking about equipment calibrated in the US (with US averaged geological data) to measure on particular location (in this case NE coast of DPRK).
    an analogy will be comparing temperature form a merucry thermometer and that of a “water” thermeter by plotting an approximately linear (log-linear in this case) fitted graph.

  16. sun bin (History)


    so what is really meaningful is only the relative magnitude from the same source. eg USGS’s 4.7 this time vs 4.2 last time => the energy is 3 times (10^0.5=3.1) larger.

    the difference of the 2 blasts in these 3 sources should be about the same. can you check if this is the case?

    in fact, in theory the russians might have more accurate/consistent data!


    regarding the location. as i recall, they dug a tunnel into the middle of a mountain (instead of drilling deep like nevada/white sands — which also explains why we cannot fit the data with formula calibrated with ‘true’ underground test, plus the wave pass through ocean to reach USGS probe/etc), since the mountain they used previously was contaminated, it is natural that they use another tunnel in a nearby hill.

    i suppose a satellite photo of certain characteistic radioactivity spectrum should tell the exact location? (and to know whether they faked it with TNT)

  17. sun bin (History)

    globalsecurity was comparing the average richter numbers, which IMHO is wrong. what they should do is to compare the data for each monitor station and look at the difference of each data pair. (then maybe plot it on a global to spot pattern)

  18. Andreas Persbo

    Sun Bin. Excellent comments. Yes, using my formula puts the first nuclear test at around 1 kT. This fits NORSAR’s 2006 estimate, which I think is very credible.