Jeffrey LewisDF-31 Launch Site at Nanyang

I often observe that I am a lousy imagery analyst. Sean O’Connor, it seems to me, is a very, very good one.

After an earlier post on my blog, he’s gone out and, in a very nice piece of research, identified what appears to be the DF-31 deployment area near Nanyang, China:

Nanyang is located in east-central China, approximately 850 kilometers southeast of Beijing and 180 kilometers south of Luoyang. Analysis of the surrounding area has led to the location of six DF-31 launch sites approximately 30 kilometers northwest of Nanyang. The six DF-31 launch sites can be found at the following coordinates:

33°16’33.80“N 112°20’06.49“E
33°14’23.78“N 112°17’50.82“E
33°12’46.96“N 112°18’55.43“E
33°12’02.03“N 112°18’49.17“E
33°09’44.19“N 112°15’47.01“E
33°11’48.95“N 112°18’38.92“E


Further evidence that these locations are DF-31 launch positions can be found in the characteristics of the DF-31 TEL itself. The DF-31’s Hanyang HY4301 TEL consists of a missile launch canister mounted on a trailer which is towed by a four axle tractor. This configuration is clearly not off-road capable. That being the case, DF-31 deployed launch sites would have to be near major roads in order to facilitate dispersal of the TELs. This fact, when combined with the information that Brigade 813 is based in Nanyang, supports the conclusion that the sites identified above are in fact DF-31 ICBM launch positions.

There are at least two policy implications of his research, if this is the right area.

  • First, the road-mobile DF-31 may not be truly survivable because the launch positions appear to be pre-surveyed. (If you think about inertial guidance, errors in the initial position of the missile will degrade accuracy.)
  • Second, Nanyang is located in Central China. It’s 7,250 km range would not be enough to target the continental United States (Even Juneau, Alaska is out of range.)


  1. Lao Tao Ren (History)

    “First, the road-mobile DF-31 may not be truly survivable because the launch positions appear to be pre-surveyed. “

    Geeze, I just about said that in a previous post about how there are few places for this thing to be deployed.

    See comment by ‘none’:

  2. Sean O'Connor (History)

    The site does hold all of the former USSR at risk. That was the initial purpose of the DF-31, to target Russia. That may be why the first sites were located here, indicative of an old strategy.

  3. Allen Thomson (History)

    > HY4301 TEL

    Are these rail-transportable?

  4. Lao Tao Ren (History)

    Within range:

    Japan, Taiwan, Koreas, most of former USSR, India, Vietnam, Philippines, etc.

    It is vulnerable to a nuclear attack only from Russia and India on that list.

    The catch is that it is also vulnerable to a conventional strike by Japan and the US for now, and more in the future.


    There is a rail mobile version. Given their terrain and infrastructure, it is probably the ideal mode of basing.

    The survivability of the Chinese deterrent has been deteriorating for quite sometime.

  5. peter Zimmerman (History)

    If there are more pre-surveyed launch sites than missiles, perhaps 2x-5x as many, then barraging the system with missile warheads becomes very expensive. It’s also expensive since catching a missile at the launch site is chancy. This was pretty much the ‘shell game’ the US was going to play with the MX missile except that movement was from shelter to shelter. Surveyed sites are cheap, and could even be not much more than a marker along a highway for secondary positions.

    Anyway, the missile’s own inertial navigation system should be able to provide launch location information sufficient to the deterrent role. That is, it should get the system to within a few hundred meters of true, even if there has been significant time to drift after departure from the home garage. Given a map, the launch crew should be able to eyeball it to that accuracy.

    This doesn’t add to much to the CEP if the target is soft, as a city or even a military airfield. Small CEP is only needed for attackin hard targets.

  6. Martin Dirksen

    Dear Sirs,
    This always interesting blog confronted me heavily with my deficencies as a “Remote Image Analyzer” . So I looked up the Internet a bit and found this interesting Online-Tutorial ( at the NASA-website, produced by Dr. Shorts. It does not cover especially military objects but is good as an general overview.. I would be thankful for further tutorials worthwhile reading. Maybe like the ones at FAS:
    Well, there are very few things where one can be excellent anyhow.
    Yours Sincerley
    Martin Dirksen

  7. Lao Tao Ren (History)


    Concur if the attack is by nuclear warheads delivered by ballistic missile with a relatively long (15min+)cycle time between recon and strike.

    The cost / benefits calculation changes favorably for the attacker if you can,

    a) continuously monitor the site with drones or like sensors with near real time reports.

    b) attack with smart conventional munitions with a very short lag time – ideally catching the missiles en route to their launch pads when they are most vulnerable.

    A cycle time of 5 minutes or less would be nice.

  8. Geoffrey Forden (History)

    Without disparaging Sean O’Connor’s tremendous effort, I remain unconvinced, especially of the so-called garrison sites which show none of the wide turns that the supposed launch sites all have in common. The supposed garrison sites also show no signs of nuclear warhead bunkers etc. The launch sites could, in deed, turn out to be what they are conjectured to be but I need more convincing. One thing that bugs me is their width, about 7 meters, while the roads leading up to them are about 4 meters wide. You could, perhaps, argue that the wider pads are needed to enable the TEL to make the turn but does that explain the launch site at 33.1623° N 112.263°E? That site doesn’t provide the TEL with a way out and should, perhaps, be referred to as a ‘J’ site. (There is one other site that has this J aspect but that does provide a way out for the TEL.)

  9. Lao Tao Ren (History)


    It would be entirely in character for decoy sites to be set up.

    Given how much hardening a missile / warhead storage facility should have in order to survive a near- direct hit by a nuclear strike, it ought to be quite difficult to hide during construction.

    I am with you —- rather uncomfortable about how vulnerable this site is.

    But then, think back to the Cuban Missile Crisis, when the Soviets set up precisely the same, highly visible, obvious and vulnerable launch sites as they did in the USSR. As I recall, that was blamed on bureaucratic logic.

  10. Sean O'Connor (History)

    Gentlemen, I have located, with the help of Hans Kristensen, what should be a storage bunker associated with the support facility It is likely that this site houses the missiles and/or the warheads. The garrisons are likely meant to house only the TELs and the administrative functions (i.e. the HQ, the barracks, etc). At any rate the DF-31 is not off-road capable, so all of the infrastructure has to be easily accessible by road, making it much harder to hide. I’ll be updating my article to include this new information and hopefully clarify some things.

  11. peter Zimmerman (History)

    I am posting something I sent Jeffrey privately on April 1 before Geoff Forden’s comments appeared.

    I am convinced that the ‘garrison sites’ are not garrisons for mobile ICBMs.


    I have read and re-read O’Connor’s descriptions of the two garrison sites. Several comments:

    * Neither site has any large radius roads (like the shepherd’s crooks), and neither has any roads that look as though an overall 17-20m articulated vehicle could enter and leave w/o having to make numerous back and forth movements (like a K-turn in your car). The roads aren’t very wide. See shepherd’s crook sites for what such roads might look like —- or see a site I commented on a couple of weeks ago.

    * It’s possible that the taller building could be a place where a TEL could be erected, but… there aren’t any appropriate roads into our out of it.

    * There appear to be walls around parts of the sites, but no security entrance. I can’t believe the PLA operates without a guard gate to check passes of people entering and leaving.

    * He hasn’t flagged the operating industrial facility 2 km due south of the western garrison site. You’ll recognize it by the very high stack (inferred from the length of the shadow) and the two cooling towers to the right of the stack. I would be tempted to identify it as a power plant, but I can’t see transmission lines — but the resolution could be just barely too low to see the towers or their shadows. OOPs: I found the shadows of two obvious electrical transmission line towers @ the SW corner of the site. One has its base inside a little circular area with a berm around it. Very big power plant! I can trace the power lines quite some ways northwards.

    * There does appear to be a coal storage yard at the northern edge of the plant, and some big construction cranes and a lot of work going on.

    * The power plant is located near the intersection of two large roads, but there’s no railroad spur into the site — though there is a main line RR N of the power plant and between the plant and the ‘garrison.’ I think it’s for electric-hauled trains because I can just barely make out the towers to support the wires. But I ain’t sure. I didn’t think the Chinese used electric-hauled trains in the boonies.

    * It also has a wall around it, with a security gate at the southern entrance; something the ‘garrisons’ lack.

    Almost all high res images are fascinating; this one surely is, even in a little bitty 2.5 km square near the N edge of the swath. But I think O’Connor has misidentified the ‘garrison.’


  12. peter Zimmerman (History)

    I am posting something else I sent Jeffrey privately on April 1 before Geoff Forden’s comments appeared. These go to the “shepherd’s crook” and “J” sites that Geoff discussed.

    1. They are very much smaller than DF-21 sites, but the DF-31 is a bigger missile, requiring a larger radius of the turn to get the TEL’s trailer around.

    2. There is no security of any kind around the sites.

    3. The smooth ‘roadway’ is only 5-7 meters wide, smaller than a standard 2-lane road.

    4. There is no concrete or asphalt surfacing, and the connections of the characteristic shepherd’s crook ‘roads’ to the main roads are not bit or strong enough to handle a heavy truck.

  13. blowback (History)

    But then, think back to the Cuban Missile Crisis, when the Soviets set up precisely the same, highly visible, obvious and vulnerable launch sites as they did in the USSR. As I recall, that was blamed on bureaucratic logic.

    If you regard the “Cuban Missile Crisis” as a successful attempt to remove Jupiter missiles from Turkey, then building highly-obvious missile sites in Cuba makes an awful lot of sense. BTW, the Soviets fought much of WW2 without air superiority so the Soviet military were obsessed with camoflage and deception. Those sites were seen because the Russians wanted them to be seen!

  14. Lao Tao Ren (History)


    If you take Graham T. Allison’s book on the Cuban Missile Crisis as a starting point, the most plausible explanation is that the Soviet missile sites in Cuba were set up in exactly the same manner as in USSR without camoflage. That counter the “intentional” explanation for not camouflaging the sites as those sites in USSR are just as vulnerable to air strikes from Europe, which presumably the Soviets care about.

    The Turkey / Cuba missile “trade” hypothesis is widely regarded in most conventional accounts of the Cuban Missile Crisis as a face saving gesture for the Soviets as those missiles were widely known by the Soviets to be “on their way out”.

    If you have first hand information or accounts based on them as to what are the Soviet’s real intent in placing missiles in Cuba, by all means share them with us.

    The Chinese are even better masters at disguise than the Soviets. Look at the facilities they have hidden quite successfully for decades.

    Given the issues that Professor Zimmerman raised earlier and the possible set of intended targets of this site, one or more of the following might be interesting explanations:

    a) there is no need to camouflage it because it is not aimed at a first rate power with the means to take it out, namely the US.

    b) it is intended to be a visible deterrent to places like India, Russia, Japan, and Taiwan which are its primary targets.

    c)it is a decoy site.

    In the absence of the alternative of an “on site” inspection by a qualified technical team, or what you might call, “national technical means”, hard answers are hard to come by.

    Your guess is as good as mine.