Jeffrey LewisCrashing Glider, Hidden Hotspring

Don’t miss our podcast on the failed Chinese and US hypersonic launches!

Analyzing China’s August 7, 2014 Hypersonic Glider Test

James Acton, Catherine Dill and Jeffrey Lewis

September 3, 2014

By a lake in an Inner Mongolian desert, about 200 km south-east of Ordos—the oft-described ghost city that hosted the Miss World contest in 2012—lies a Chinese resort called the Bulong Hu Hot Springs Resort (布龙湖温泉度假区). On August 7, at about 11am, tourists in the resort were presumably doing what tourists at a lake-side spa do. Maybe a young couple from Beijing was soaking in the hot springs, enjoying a luxurious end to a hot and dusty trek around Inner Mongolia. Perhaps a retiree from Ordos, bored of watching Miss World highlights on Good Morning Ordos, was enjoying the relative excitement of fishing on the lake. Maybe a shepherd was grazing his sheep in the cultivated land just outside the resort. What we can safely assume is that none of them knew what was, almost literally, about to hit them.

The noise—a thundering crash—must have been the first terrifying indication of what had happened. Fortunately, we don’t have to speculate about what they saw because some of them photographed it: huge clouds of red smoke billowing up from the desert. Someone even got near enough to the crash site to take photos of it. Even to his or her (presumably) untrained eye, it must have been clear that the debris littering the area was from some sort of a rocket.

These cell phone images appeared online almost immediately. However, they seem to have been suppressed and quickly vanished from the Chinese websites where they first appeared.  But this is the internet, so nothing can be deleted.

Almost immediately, Chinese internet sources connected the rocket with a test of what the Pentagon calls the WU-14—a hypersonic glider, launched by a rocket, that China is known to have tested at least once before, in January 2014. (Technically, the term “WU-14” probably refers to the whole package of booster and glider, but it’s become the glider’s de-facto name).

Bill Gertz, of the Washington Free Beacon, picked up on these rumors and on, August 19, published a somewhat alarmist article, which appears to have been largely based on Chinese internet sources—although he also reported that two anonymous U.S. officials had confirmed that the test did involve the WU-14. Three days later, the South China Morning Post reported that the test was a failure. Chinese internet sources had said the same thing but Gertz did not, implying that such debris was to be expected.

It turns out that there is a wealth of open-source information about the August 7 test. It has allowed us to find the exact location of the crash site, and to make several important observations about what happened that day in a remote part of Inner Mongolia: the WU-14 hypersonic vehicle was almost certainly tested, but the test was probably a failure. More generally, our analysis indicates that the Chinese hypersonic glider program is probably significantly less ambitious than the U.S. Advanced Hypersonic Weapon—a U.S. hypersonic glider that was tested 18 days after the WU-14 and also failed.

Geolocating the Crash Site

The rocket debris appears to have fallen in an inhabited area of Inner Mongolia.  Yes, there are people there.  Why are you laughing?

The debris clearly shows a Chinese booster marked with “China Aerospace” (中国航天).  Some debris appears to be an engine, possibly a YF-22 that one might find in the second stage of a Long March booster. Moreover, the plumes of red smoke are probably the N2O4/UDMH propellant used in the first two stages of all Long March rockets (See 3:13-3:18). They indicate that there were substantial amounts of unburned fuel on board when, as planned or otherwise, gravity got the better of the rocket.

Chinese social media postings provided good descriptions of location of the crash site, provided one is familiar with the lower-order levels of Chinese administrative units in Mongolia.  Or has access to Wikipedia.

中文 English Administrative Level
内蒙古 Inner Mongolia Province
鄂尔多斯 Ordos (E’erduosi) Prefecture
鄂托克旗 Etuoke Qi County
布龙湖 Bulong Hu Local place name

“Hu” means “lake” – there aren’t so many in the Gobi Desert.  (Ok, it’s technically the Ordos Desert. The Gobi is nearby.)

Many of the pictures of the debris and cloud appear to have been taken by a single user who was near the crash site and then apparently relocated to the Bulong Hu Hot Springs Resort.  Rocket propellant is extremely toxic so it’s not surprising that, according to social media accounts, authorities moved to relocate residents and tourists from the crash site and to the resort.  More surprising is that locals were able to take images that were later posted online.

The Bulong Hu Hot Springs Resort is a recently constructed eco-tourism site boasting thermal baths and other amenities.  It’s well-described online, including in the local news, and the architect’s website. Here is an image of the resort, which shows the distinct architecture of buildings.

Schematics of the resort also show plans for boat docks, although most of these had not been constructed at the time satellite images were taken.

Other images of the Bulong Hu Hot Springs Resort.  No, we have no idea what’s with the dinosaurs.

The three images of the plume were clearly taken at this resort – one outside the front gate, one from the grounds looking east across the complex, and a series from the lake – perhaps by the same person.  In all three images, the shape of the red plume is similar suggesting they were all taken from approximately the same vantage point.  The images are consistent with accounts that suggest the photographer or photographers, may have been evacuated from near the crash site.  Perhaps someone took additional photographs of the plume when entering the resort, then again on the grounds, and finally then went down to the boat dock to take a series of images not obscured by the resort buildings.  Or perhaps there were multiple photographers.

The first image is possible to match with moderate confidence.  The paved highway is relatively unusual in the area.  Moreover, along the right side of the road, vegetation has been planted in straight lines running along the road.  In satellite imagery, this feature is visible in front of the main entrance to the Bulong Hu resort.

We can match the second image with high-confidence. The unusual design of the buildings—especially the domes–is a signature of the resort. Moreover, the relative location of the buildings in the image matches the layout of the Bulong Hu Resort.

The final series of images, taken from the dock, can be matched with moderate confidence. The docks were not constructed at the time satellite images were taken, although schematics show plans for them. The shape of the docks in these schematics is a good match for the shape in the ground-truth images. The dock in these images, moreover, appears to be a temporary structure – the side of the dock is lined with tires to soften the impact of boats against the dock. Some images of the resort show what may be temporary boat docks.  Still, the presence of a lake large enough for boats in the desert is a relatively unusual feature.

Taken together, the images suggest that the crash site is a few kilometers east of the Bulong Hu Resort.  We can further use the line of site from three slightly different images to triangulate the likely crash location.

The first image has a line of sight directly down the paved road.  The second image looks directly east across the compound, facing the westernmost building.  The third series of images are taken from the  lake, with the resort buildings just off frame.  The approximate location of the crash site is roughly 6 kilometers east of the Bulong Hu Resort or 38°55’16″N, 107°34’54″E.


Analyzing the Drop-zones

The Notifications to Airmen that China issued to warn pilots (or at least the male ones) of falling debris are another source of information on the test. There’s no official source for cancelled international NOTAMs, but we obtained them from two independent sources: KKTT’s blog and In addition to closing part of an international air route, China declared two keep-out zones, shown as yellow boxes on the map below.

First off, it’s clear that the launch site was the Taiyuan Space Launch Center (known by the US intelligence community as Wuzhai, which is where the WU in WU-14 comes from) and not the Jiuquan Space Launch Center, as originally reported. Subsequent media reports corrected this error, but the correct launch site is immediately obvious from the location (and orientation) of the keep-out zones. For good measure, there are also pictures on Chinese websites of a space launch from Taiyuan at the right time. Besides, an eastward launch of a glider from Jiuquan makes no sense since the glider would be aimed towards heavily populated areas.

Drawing further inferences is complicated by the fact that there were almost certainly other areas—in addition to the two keep-out zones shown in the picture—where debris was intended to fall. If the launcher was a two-stage rocket, as seems likely and has been reported by those who track launches, then you’d probably expect four keep-out zones: two for the rocket stages, one for the shroud, and one for the glider. Hell, even if we’re wrong and the launcher was a one-stage rocket (a retired DF-3 painted in civilian colors, anybody?), it would still most likely result in three keep-out zones.

In analyzing the limited information that we do have, it’s perhaps easiest to start by pointing out an obvious interpretation that’s probably wrong: that the eastern keep-out zone is the first-stage drop-zone and western keep-out zone is for the second stage. Drop-zones for rocket stages generally get larger the further downrange they are. Since the western keep-out zone is significantly smaller than (and a different shape from) the eastern one, it seems unlikely they correspond to the drop-zones for the two stages.

Instead, we think it’s likely than the western keep-out zone is the intended target, and the eastern keep-out zone is probably—but not definitely—the intended final resting place of the second stage.

There are two reasons for associating the western keep-out zone with the intended target—as others before us have done. First, it’s relatively compact in size (especially perpendicular to the flight path), suggesting that whatever was supposed to land there was guided. Second, it doesn’t make sense for the target to have been much further downrange because the landscape rapidly becomes mountainous (which would severely complicate ascertaining the exact landing place of the glider).

The association of the eastern keep-out zone with the second-stage drop-zone is more tentative. First off, the crash debris shows what looks like an engine for a second, not first, stage, although this is far from conclusive. Moreover, when the United States has conducted boost-glide tests, the first stage of the booster has been launched almost vertically, resulting in a first-stage drop-zone less than 100 km from the launch site. The rocket is then pitched over rapidly to flatten the trajectory. (“Why?” I sense you are wondering. Lisbeth Gronlund and David Wright’s discussion of the aerodynamic loading for “shaped” versus “symmetrical” depressed ICBM trajectories may contain the answer.)

If the Chinese booster followed a similar trajectory, then the eastern keep-out zone, which is about 600 km downrange, is probably the second-stage drop-zone. That said, we can’t entirely rule out the possibility that this keep-out zone is actually associated with first stage or the shroud.

The center of both keep-out zones and the launch site lie, almost perfectly, along a straight line—the red one in the picture—which is the most probable intended flight path. The crash site lies about 5 km south of this line, suggesting that the rocket lost control and deviated from its intended trajectory before crashing. However, another possibility is that the flight path was curved (after all, one of the big purported benefits of gliders in their midcourse maneuverability) and the crash site, in fact, lies on it. The blue line shows a notional representation of what such a flight path might look like.

Beyond that, it’s difficult to say much about the rocket’s trajectory; there are too many unknowns to model it. One intriguing possibility suggested to us is that the rocket’s second stage could have been used to drive the glider downwards to increase its speed—much as Jonathan McDowell tells us the Russians apparently do when testing new ICBM re-entry vehicles (more). This maneuver would help explain why a normally reliable booster, such as the CZ-2C, might fail (as, for reasons explained below, it appears to have done). But it bears emphasizing that the available evidence is also consistent with more conventional trajectories and doesn’t allow a conclusion to be drawn about the rocket’s trajectory during the second-stage burn.

A final observation is that the crash site lies someway outside the one declared drop-zone, raising an important question: Did the rocket crash short of this drop-zone (implying some sort of a failure), or did the rocket debris fall where it was supposed to, in an undeclared drop-zone? While we can’t rule out either possibility, it seems unlikely that the debris fell in an undeclared drop-zone. An air traffic route (the green line on the map) passes within 40 km of the crash site, and a search for NOTAMs within just two nautical miles of the desert spa generates hits related to that air route. This suggests that if rocket debris was supposed to land where it did, China would have declared a drop-zone there.


So what?

Here are three conclusions and some food for thought.

The launch’s purpose was, almost certainly, to test a hypersonic glider. This is hardly news—but we can present concrete evidence for this conclusion. First off, space launches, with one notable exception, are oriented eastward to take advantage of the rotation of the earth. China follows this practice and its westward launches are generally missile tests. Moreover, we can be pretty sure that China tested a hypersonic glider on January 9, 2014. A senior U.S. official has said so unequivocally. And, the Chinese government has acknowledged it too—albeit more equivocally. Thanks to we have the keep-out zones for that test. Actually, I should say keep-out zone, because there was only one—but it was identical to the eastern keep-out zone for the August 7 test. This strongly suggests that the two tests were of the same thing, that is, the WU-14. (Why there was only one keep-out zone for the January test is a mystery.)

The August 7 test was probably a failure. There are three pieces of evidence that support this conclusion. First, it’s unlikely that rocket debris would be intended to fall inside a zone featuring several development projects within a few kilometers, including a holiday resort. Second, there was a lot of fuel left in the rocket stage (or stages) that crashed near the desert spa. Third, the crash site lies outside the declared drop-zone (and it seems likely that China would have declared a drop-zone around the desert spa had debris been intended to land there).

None of these reasons is, by itself, conclusive. But together they would seem to confirm what “two sources close to” the Chinese military reportedly told the South China Morning Post –  the test was a failure. Moreover, if the test did indeed fail, it appears very likely that the failure was caused by a booster problem and that the glider was probably not deployed. (Remind anyone of any other recent hypersonic glider tests?)

The WU-14 appears to be significantly less ambitious than the U.S. Advanced Hypersonic Weapon. The location of the probable target zone for China’s August 7 test implies that the intended range was about 1,750 km and provides the first real evidence about the capability of China’s hypersonic glider. To put this range in perspective, the U.S. Advanced Hypersonic Weapon was successfully tested over a range of 3,800 kilometers in November 2011. The plan for the more recent failed American test, on 25 August, was to test the Advanced Hypersonic Weapon across a range of over 6,000 km and for it maneuver hundreds of kilometers cross-range. By contrast, the planned flight path for China’s most recent test appears close to completely straight.

It therefore appears as though the United States has a distinct lead in hypersonic glider technology, which is not surprising given the long history of U.S. efforts in this field. But, it does run contrary to a growing media narrative of China surpassing the United States. Exhibit A: the unsubstantiated claim that China’s glider can travel at twice the speed of the Advanced Hypersonic Weapon.

Finally, some food for thought: what is the significance of China’s use of a liquid-fueled booster?

Given China’s increasing use of solid-fueled ballistic missiles, the use of a liquid-fueled rocket for a boost-glide test came as a surprise—at least to us. Here are two possible explanations. First, if China’s glider is a bit overweight, liquid fuel may be necessary to achieve the required speeds. (Because liquid fuels generally have a higher specific impulse than solid fuels, a given quantity of liquid fuel can typically accelerate an object to higher speeds than the same mass of solid fuel).  Russia uses a liquid fueled booster, for example, to achieve the speeds necessary to simulate an ICBM reentry for its nuclear warheads.

A second possibility—which is not mutually exclusive to the first—is that China may be planning to deploy the glider on liquid-fueled missiles. This would be interesting because China’s liquid-fueled missiles are used exclusively to deliver nuclear weapons —possibly suggesting that China’s glider is intended to ensure that China’s nuclear weapons can evade US missile defenses. Indeed, at a hearing of the U.S.-China Economic and Security Review Commission on January 30, 2014, Lee Fuell from the National Air and Space Intelligence Center testified as much. He didn’t give a reason—but perhaps he was thinking about propellants.


  1. John Schilling (History)

    Wow; that’s one of the more impressive BFRCs I’ve ever seen in the wild. Agree that there is no way the test was planned to end that way, and I’m wondering about possible casualties on the ground. A folly-fueled Long March second stage has over twenty tons of NTO, which is roughly phosgene-equivalent in toxicity.

    I don’t know that I would read too much into the use of a liquid-fuel rocket, or the range profile of the test. The Long March series are still China’s workhorse space launch vehicles, and DF-31 deployment means there are probably a bunch of recently-decomissioned DF-4s that need to be disposed of somehow, so that could just be simple economics.

    And the 1750 km range may have been a test-range limitation; as you note things get mountainous farther West. Since China doesn’t have the vestigial colonial empire or the blue-water navy to easily conduct long overwater tests, I would expect even their very long range weapons development to be characterized mostly by reduced-range testing, though they would obviously want at least a few proof shots at full range before deployment.

    • George William Herbert (History)

      I was too busy yesterday to mention the BFRC responding to the podcast, but yes, that’s impressive.

      Even better is that it seems like the close-up cameraman may have walked through the cloud at ground level taking the pictures. I am going to hazard a guess that this was at least seriously injurious if not fatal. 8-(

      My guess on the use of liquid propellant boosters: They are retiring a lot of older liquid propellant missiles and replacing them with solid propellant missiles. That means that liquid propellant vehicles are coming surplus to active military needs, and solids are key for active military needs. While a solid would likely be used for a production vehicle, working out the design on a hypersonic weapon using a surrogate launcher that’s cheap and surplus makes sense to me. The US is using old Polaris missiles, for example, which would never see the light of a sub or silo again for safety reasons (cough).

    • George William Herbert (History)

      As an aside, whatever that stage was, it’s a reasonably well done orthogrid aluminum structure.

      Question for the masses – what’s the correct reading orientation top/bottom for the characters on the side of the tank section?

    • Jeffrey (History)

      It reads top down, but you are looking at the top in the foreground. The first character of “China” has been torn apart.

  2. RAJ47 (History)

    Well done.
    We need to answer one more question.
    What happened to the payload?
    The glide weapon made to survive Mach 10-20 plus speeds, logically should have survived the crash. Someone taking pics from 8-10 feet distance of burning 2nd stage should have been able to get the pic of the glider too.
    Mark also has a nice article.

    • Jonah Speaks (History)

      Yes, Mark Gubrud’s article, “The argument for a hypersonic missile testing ban” (09/02/2014), makes some excellent points.

      He argues a) deployment of hypersonic missiles would be destabilizing, b) a ban on testing would be easily verifiable, and c) lack of testing precludes deployment. He argues that a hypersonic missile has no significant military or civilian use and is distinguishable from existing missiles. A ban would prevent an expensive arms race and prevent development of a destabilizing weapon that could increase the odds of nuclear war.

  3. Brian Weeden (History)

    Great work, although I think you let Gertz off too easy. This most definitely does not “accelerate China’s timeline exponentially” or pose a “near-term threat” as stated in his article.

  4. SQ (History)


    • Jeffrey (History)

      I forgot to give one of my students, Phillippe Mauger, credit for identifying the rocket engine.

  5. RAJ47 (History)

    IMO the crash site is West of the resort not East.
    I have requested SOC for a second opinion.

    • Jeffrey (History)

      That’s a good idea.

  6. Jonathan McDowell (History)

    I am open to the idea that it’s a CZ-1 (CZ-1D?) variant rather than a CZ-2 derivative – there have been rumours of these being used for suborbital tests from Taiyuan/Wuzhai in 1995 and 1997. And it’s certainly plausible from your analysis that it’s a single-stage vehicle with the
    E drop zone being stage 1 and the W zone being the gliding payload.

    I guess the debris is too mangled to estimate a tank diameter…

  7. Allen Thomson (History)

    What would be expected for the cross-range capability of a hypersonic glider with a nominal 1750 km range? I ask because there’s an extensive set of targets 200 – 250 km to the north of the path between Taiyuan and the western keep-out zone. (See 40.474 N, 93.515 E and 40.404 N, 93.637 E for examples.)

    • RAJ47 (History)

      The second one is a HAARP like experimental facility and not a target.

    • Allen Thomson (History)

      > The second one is a HAARP like experimental facility and not a target.

      That’s — interesting.

      What would you say the area 4.5 km to the west, at 40.414 N, 93.583 E, is?

  8. RAJ47 (History)

    That one is a target area to measure blast effects of ammo on various objects at varied distances and heights.

  9. krepon (History)

    ACW at its best.

  10. Bradley Laing (History)

    “More surprising is that locals were able to take images that were later posted online.”

    —-A few weeks ago I found an article online that claimed that study had been done of Chinese internet censorship. The article said that criticism of the Chinese Communist Party is not the first priority of internet censors. The first priority is preventing online organizing of protests.

    —Maybe the delay in censoring the photographs happened simply because the “medium level office worker” could not find a reference in the manual telling him to delete the photos, so he had to wait until it went up the chain of command.

    —Where, of course, this man was in charge:

    • Bradley Laing (History)

      —A follow up thought: someone was complaining about a company, and said of the lower level executives “They have the power to say no, but not the power to say yes.”

      —Do Chinese internet censors face getting fired if they delete something that they have not already been told they can delete?