the home of online investigations

You can support the work of Bellingcat by donating through the following link:

New MH17 Photograph Geolocated to Donetsk

October 20, 2017

By Bellingcat Investigation Team

Translations: Русский

On October 19, 2017, the Joint Investigation Team (JIT), the Dutch-led criminal investigation into the downing of Malaysian Airlines Flight 17 (MH17) over Ukraine, published a new photograph Buk 332, the Russian Buk missile launcher that downed MH17 on July 17, 2014. Buk 332, previously identified by Bellingcat as ‘Buk 3×2,’ is an anti-aircraft missile launcher belonging to Russia’s Kursk-based 53rd Anti-Aircraft Missile Brigade. The image shows the right side of the Buk, loaded on the red low loader. In publishing the photograph, the JIT wrote that the photograph was “probably taken on July 17, 2014 in the town of Makeevka, Ukraine,” but we can now conclusively say that the photograph was actually taken at 78 Prospekt Ilycha, at the intersection with Shakhtostroiteley, in Donetsk.

The Hunt

Using Check, a crowdsourced verification platform, we asked our readers to contribute tips regarding the location of this photograph so that we could collectively geolocate this key piece of evidence concerning the downing of MH17.

One of the strongest candidates was 78 Prospekt Ilycha in Donetsk, due to previous witness accounts identifying a Buk missile launcher with its Russian/separatist convoy at this site.

Source (Archive)

“Bad news. Around 9am, a hauler was going along the Makeevka highway from Makeevka in the direction of Donetsk. On the platform was a BukM1-M2? This AAMS proceeded to the intersection with Shakhtostroiteley Boulevard. The system was accompanied by a convoy that was composed of 1 gray Rav4 SUV, a camouflaged UAZ, and a dark blue Hyundai van with tinted windows. As of 9:15am, the vehicle was located at the intersection of Shakhtostroiteley and Ilycha. The militants got out of their cars, blocking 2 of the far left lanes. Obviously, they were waiting for logistical guidance.”

Numerous features in the photograph were also visible in Google Street View and Yandex Panorama imagery of the same location, including the two trees in the foreground, a gate beneath the Buk’s low-loader, a rock next to the nearest tree, and notches in the curb blocks.

Taking these tips together, the following comparisons show how some key details match between the two photographs, making the site worthy of  further investigation to confirm the geolocation.

Left: A part of a gate visible in an open space through the red low loader, just under the Buk missile launcher.
Right: the same type of gate visible on October 2011 Google Street View of Prospekt Illycha in Donetsk, Ukraine.
Left: The new Buk 332 photograph with a thick tree, a slanting branch of another tree, a little rock and notches in the curb, all marked in red.
Right: the same landmarks in Google Street View marked in red.

Conclusive Evidence

Today, at least one local in Donetsk snapped two photographs from this same location, attempting to confirm or refute a geolocation to Prospekt Ilycha in central Donetsk. These photographs were shared online by Christo Grozev and Rudy Bouma.


With this geolocation, we now have photographic evidence of a scene that was reported on by numerous locals: Buk 332, originating from Russia’s 53rd Anti-Aircraft Missile Brigade, arrived in Donetsk from Makiivka around 9am on July 17, 2014, parked at the intersection of Prospekt Ilycha and Shakhtostroiteley, and then eventually turned towards the east-bound Prospekt Ilycha towards the Motel roundabout in Donetsk. The movement is shown in the map below, with the photographer of the original Buk photo facing south from building number 78.

Bellingcat Investigation Team

The Bellingcat Investigation Team is an award winning group of volunteers and full time investigators who make up the core of the Bellingcat's investigative efforts.

Join the Bellingcat Mailing List:

Enter your email address to receive a weekly digest of Bellingcat posts, links to open source research articles, and more.


  1. ttb

    It is a truly awesome piece of deduction and sleuthing. I do sincerely hope though it doesn’t lead to the photographer being publicly identified. If they sent it anonymously to the JIT I’m sure they’ve their reasons.

    • Aric Toler

      Thankfully, the photograph was taken from outside of a beauty salon on a very busy sidewalk, and not an apartment or a house. The photo was clearly cropped on the sides when it was submitted, so the witness may have already removed identifying information about other people (or him/herself) in the photo.

  2. Tom Wonacott

    Robert Parry at Consortium News writes in his September 2017 article “The Official and Implausible MH-17 Scenario” that the circuitous route taken by the Buk anti-missile battery was implausible yet he never addresses the geolocation of the Buk by Bellingcat or the Dutch investigation unit. He could have investigated the same sources as Bellingcat yet did not even address the issue as if that visual evidence is just going to disappear. Parry simply obfuscates and deflects while accepting Russian accounts. Parry writes:

    “…….After the MH-17 shoot-down, which killed 298 people, I’m told the Russian government did fear that somehow one of its field operatives might have been responsible and conducted an intensive investigation, including an inventory of its equipment, concluding that all its Buk missiles were accounted for……”

    Parry accepts the narrative of the Russians without question using an unidentified source(s) (“….I’m told….”). This is typical of the journalistic standards of Parry who treats the Russian and Syrian governments as clients. He then promotes an unsubstantiated conspiracy theory using a single source and the familiar “……I was also told……”:

    “……..I was also told that at least some CIA analysts shared the doubts about Russia’s guilt and came to believe that the MH-17 shoot-down was the work of a rogue and out-of-control Ukrainian team with the possible hope that the airliner was a Russian government plane returning President Vladimir Putin from South America……”

    Pure bunk – and Parry knows it. Parry is an advocate. He provides zero evidence for his theories on MH17, but the goal isn’t the truth, but to promote doubt in the investigation to protect Russia.

      • Tom Wonacott

        Thanks. It’s a big error. I confused the date with the Consortium article on September 7, 2017 “A New Hole in Syria-Sarin Certainty”.

  3. Andrea

    I find it AMAZING how in a matter of hours an image can be reliably located (even tough you obviously had a specific route to check).
    This image is really good cause it shows the Buk and his details, and the blue van too!
    A timestamp would have been great too…. but if you are able to do 1+1 you probably don’t need the timestamp…

  4. The Bird

    While this video did not draw too much attention (yet) I think it contains a nice piece of information:
    Jump to 8:30. The DNR soldier was asked to tell what he believes what happened on the day of downing MH17. He did not say: “There was a Ukrainian fighter jet.” He said that there were Ukrainian soldiers that sent a rocket towards MH17. This is the first time I see DNR people saying that a surface-to-air missile hit MH17. Of course he would not admit that it was done by DNR or Russian people – and so far I don’t know who to blame for the downing (although I have an idea). Anyway, the important piece of information is that even DNR people say that MH17 was brought down by a BUK missile.

  5. Ilya Kuprik

    Any evidence, that photo was really taken on 17th July? Any evidence, it is not a digital manipulation?

    • KimmoK

      There is no evidence (or eyewitness) of any BUK TELAR being there on any other day.
      There is even less evidence of that specific Russian BUK being there on any other day.

      There is no evidence of any digital manipulation (not yet, not beyond cropping and resaving). So far all 9+ BUK TELAR photos/videos seem authentic.

      • Walter Mitty

        It is my opinion as an animation and compositing worker, that there is strong evidence that the Snizhne video, upon which part of the Buk launcher narrative is based, is likely a fake. On frames 55 – 72 (around 4-5 seconds) there is a hard, straight, vertical edge video artefact which develops on some foliage in front of the Buk launcher, and propagates above the foliage and launcher into the road, and down below the launcher too, cutting away some of the foliage. It is my contention that this artefact is very likely a product of where the foliage in the footage has been isolated and duplicated to create a foreground layer for the Buk launcher to pass behind.

        The masking process for creating a foreground works like this: Duplicate the video layer, and using ‘masks’ cut out the area you require to be a foreground layer, in this case the foliage on the near side of the road. You can then place your middle ground object, in this case the Buk launcher, behind the foreground layer. In the case where the original source footage is video and the camera FOV is moving, it is often necessary to manually animate the mask position and shape of the foreground layer to ensure the correct area of the video remains in the foreground, a time consuming process. Done meticulously, this can be flawless, but since this was uploaded not long after the MH17 tragedy, corners seem to have been cut (literally and figuratively), and instead of perfectly cut out foliage we see a hard, straight mask edge emerge between frames 55 and 72, culminating in a large vertical cut line at 72 frames (Just shy of 5 seconds into the footage) of the video. I work in After Effects almost every single day, animating and compositing, and I have never seen video compression produce such a video artefact, and this artefact is unique in this footage in both it’s scale and nature, and located exactly where we’d expect to see a line caused by a poorly animated mask layer – no other encoding artefact of the video produces a line so straight, nor so long. I am familiar with the difficulties of masking foliage on video footage, and of the appearance of poorly masked layers, and this video artefact looks to me to be an archetypal example a poorly executed foreground layer mask, due to it’s straight edge, it’s emergence with the camera movement, the lack of ‘partial’ pixels (pixels around the edge of an object that share the colour of an object, and of the object behind it) and the fact that it extends well past the object that is being masked: it is entirely consistent with an animated mask edge.

        There are further interesting elements of the Snizhne video that raise questions too – each time the Buk passes behind a post (a relatively simple foreground layer masking task as compared to foliage, due to the straight edges of a post), it’s angle visibly changes, as if it is being refracted by the post. Interestingly, this ‘bending’ of the Buk launcher actually occurs in a fashion incompatible with it’s change of position relative to the camera: The bend is to the right, whereas an object moving away from the camera along that motion path would, if anything, bend slightly to the left, but imperceptibly over time: the right hand side of the launcher would grow less visible as the distance from the camera increases, whereas a rightward angle turn as we see in the video would in theory show us more of the right hand side of the launcher. The effect is both incorrect in it’s direction, and far too pronounced for the distance travelled behind a narrow post. I’ve tested this theory in 3d software using map data of the location and the upper edge surface angle shifts to the left as the object moves away from the camera. This can be seen at it’s most pronounced at 0-1 seconds, and again at around 17 seconds. The straight edge of the Buk launcher chassis above the tracks is bent rightwards as it passes behind each of the posts. Foreground elements which momentarily hide the object (Buk launcher) provide a compositor a good opportunity to shift the angle or perspective of an object in compositing software, especially if working with a static image of the object question, in this case the Buk launcher. From experience, achieving accurate perspective shifts on a 2D layer object in calculated 3D space is difficult, and it appears to me that the compositor in this instance has wrongly interpreted the perspective, and shifted the Buk launcher’s perspective in the wrong direction under the cover of the roadside poles. Interesting too, that the video stops just before the Buk would have to make a turn up the hill – this would be very tricky to do without a different angle image of the Buk, and despite the presence of a post to hide the turn, I do not think it would be convincing: it would require considerably more time, a different angle image of the launcher consistent with the other image, and the more profound angle shift of the bend and incline would risk additional errors. Of additional interest, is that the video itself doesn’t appear to be OF the Buk launcher, rather its presence appears to be incidental to the video footage: One would expect that if the intention was to film a Buk launcher driving along a road, the Buk would largely be located in the centre area of the video shot – people tend to keep the focus of a video towards the centre of the FOV. That it is not further suggests that the Buk launcher was added later to a video shot out of an apartment building window. This last point is very circumstantial, I’ll grant.

        This all casts doubt on the veracity of this video, would provide an explanation for the rush to take it down from Youtube on 17th July 2014 (it’s creator may have noticed their mistake): Even a relatively easy video compositing job such as this would likely require several hours to create and render; one must find the assets – video of the location, image(s) of a Buk at the correct angle(s), 3d motion track the footage to place the required object in calculated virtual 3D space (using After Effects 3D camera tracker feature – FYI – I was able to to successfully perform a 3D camera tracker ‘solve’ on the footage, which is not always possible – After Effects needs to be able to compute 3D depth and is not always capable of ascertaining sufficient data from footage. This would be far and away the easiest way of placing and moving a Buk launcher into 3D space given the amount of camera movement), successfully composite the elements, create foreground layers and mask the foreground elements (the step which appears to have been bodged due to the camera movement), colour correct the elements to ‘glue’ the aesthetic, add lens blur/Depth of Field shifts, and brightness & contrast changes – necessary in this video – and finally render and upload the video – I anticipate this would take, in the absolute best case scenario, a couple of hours for a competent After Effects user, but much likely more, even given the serious masking error in it. I’d be very interested to know if we know it’s exact upload time on 17th July.

        I’m not claiming to have answers, and there remains a possibility I am wrong and that there is some other logical explanation for what has every appearance of a poor foreground masking job and incorrect perspective changes, but I believe these issues deserve further attention and refutation, lest this key piece of evidence is later challenged. I highly recommend you download the footage yourself and go through it frame by frame to see what I refer to.

  6. muchandr

    Isn’t it convenient how black van is blocking the special roller location. So there is really nothing but the white text connecting this to anything else. (Disregarding preconceived notions)

    This one seems to be unusually well legible and say “H-2400” You have presented not a single sample of this before, only “H-2000” and “H-2200” Moreover, the cases where there seems to be nothing but a smudge of paint (disregarding hopeless case of low resolution, which is most of your evidence) Specifically cases where it should be visible, but is not, is it really overpainted with non-specific white smudge on the vehicle or Photoshopped? Because that one got to be the easiest image edit around, don’t you think, just a bit of blur? Does not have to be Photoshop, probably even easier to hint the JPEG compression to be extra diligent at the spot with a command line tool.

    Also still interested in anyone substantiating the claim that there only one special Buk that’s got those white marks / text / smudges on both side’s skirts. You’ve got to have an entrance point somewhere to the logic chain that leads to connection of numbered sides/Russia to blank sides/Ukraine. I don’t see any of that. If you don’t, you have seen not one Buk in Ukraine that you’ve also seen in Russia.

  7. Rob Heusdens

    “Probably taken at 17.07.2014”
    What is the value of that picture if the date can not be verified with certainty? Is the author even known?
    And why would witnesses been cut of from the photo? Other then being able to trace them and asking them their witness report (which could conclude wether the photo is fake or not, probably it is just fake, because the whole case “Russian BUK downed MH17” is fake from the beginning).

    • muchandr

      In order to be in line with the intercepts, they need “this” Buk to to be in Ukraine for one day only. The separatists talk about only expecting it in the morning on 16th around 7pm, and the morning of 18th it was supposedly sighted on that intersection in Luhansk. The reconstruction of the track to/from Donezk from/to Russia via Luhansk has been shown to take 4:27 minutes or something equally ridiculous! “This” Buk also got to spend ~10am-12am parked near Torez gas station. This whole thing as presented requires the thing to be everywhere at once and I don’t care anymore about specifics of image evidence. There is too much of it, yet none is any good in quality.

      Consider that Pervomaisk by Snizhne is some 23 min to Russian border, but going strictly South, opposite direction from Luhansk Donezk. In about 1:44 you can get to closest Russian airbase in Taganrog, where they have Buks too. Why take one in Kursk in the North, but than drive it all the way around Ukraine before entering?

      There is a railroad station in Torez (known for being where they loaded #MH17 victums onto refrigerated train) which get you to South of Russia probably faster still than 23 mins. There is however, a straight railroad connection from Torez/Gorlivka to Kursk too, only takes longer. Why using the public roads? It is not even obvious that you can take a vehicle so badly oversized through the center of Luhansk even if did make any sense, so this all looks like a show staged for the benefit of Bellingcat.

      If anybody was really interested in investigating, they would have addressed the missile nosecones and how they correspond to the alphanumerical types. Is there 9M38 or 9M38M1 among these? The “separatist Buk” has been sighted using white cone missiles only, the “Kursk 53rd” red caps on its TELs. What are those? Why and how would they interchange for “Ukranian type” missiles and is it even possible to reload the TELAR vehicle without a TEL?

      Note the JIT is now showing a nose cone colored zinc galvanized bucket the likes of which no one has seen before. It is supposed to be 9M38 types (that came with original pre-M1 Buk system from the 70ties neither Russia nor Ukraine don’t really use anymore) 9M38M1 they claimed all along, including in their 2016 preliminary was supposed to be last built in 86, so this should be even older. Did they lie before or did they just found the missile now?

  8. gupu

    IF BC would know who published all these videos and images they would never use them as evidences. But who cares? The russians who made these fakes with all these foolish forgeries laughed out loud, they had their fun everytime they hit the council button.


Leave a Reply

  • (will not be published)

You can support the work of Bellingcat by donating through the following link: