the home of online investigations

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

Video Apparently Showing Flight PS752 Missile Strike Geolocated to Iranian Suburb

January 9, 2020

By Bellingcat Investigation Team

Translations: Русский

On January 9th, a video spread online after it was posted onto a public Telegram channel showing what was apparently a mid-air explosion. The New York Times has contacted the person who filmed the video, received it in high resolution, and confirmed its authenticity. Below, an annotated version of the video created by Jake Godin (Newsy) highlights the events of the brief clip. (Uploaded directly here)

We have geolocated this video to a residential area in Parand (coordinates 35.489414, 50.906917), a suburb to the west of Imam Khomeini International Airport, from which Ukraine International Airlines Flight 752 (PS752) departed to Kyiv.

The camera here is facing northeast, towards the flight trajectory of PS752. A number of elements in the source video can be found in satellite imagery, including a series of apartment blocks, a number of light fixtures, and a construction area to the immediate left of the camera position. In new satellite imagery posted on Terraserver dated 15 November 2019 (not included in this article due to copyright issues), the buildings visible in the video are constructed and in the same position.

By measuring the time that it took for the camera to hear the explosion (~10.7 seconds), we estimated the straight-line distance of the event from the camera at approximately 3.6 kilometers. By using the Pythagoras theorem, we calculated the land distance from the camera to the event to be approximately 3.3 kilometers. We then cross-referenced this distance with our tentative geolocation of the video, and also plotted the flight trajectory of PS752 (taken from from FlightRadar24, found here). The land location of the event aligned with the trajectory of the plane as extended based on FlightRadar24’s data.

[Note: We made a slight adjustment to the vertical side of the triangle, resulting in an extension of the radius by 500 meters (2.8 to 3.3 kilometers). The graphic below is adjusted accordingly. Thank you to our readers who helped notice a necessary change in the elevation.]

It is unclear why the person holding the camera was filming at the time, but it is possible that there were two missiles, prompting the decision to start filming for the second strike. The New York Times reported that the person filming started doing so after hearing “some sort of shot fired“.

Unresolved questions around missile fragment

While the footage showing an apparent missile strike has been geolocated, two photographs apparently showing part of a Tor M-1 missile have yet to be verified, despite claims from a number of sources. The warhead is located midway on the missile, meaning that its nose may not be destroyed in an explosion. A number of similar photographs of Tor missile fragments have been taken in eastern Ukraine, but none have been discovered to be the same as the ones attributed to the recent incident.

The origin of these photographs is still being determined, as the people who snapped the images have not come forward publicly. This object is likely located in a residential area near Parand, not near the crash site. Both of the images show the same location and object, as seen with many of the same rocks and damage patterns on the curb.

If you make any progress in geolocating these two photographs of the missile fragment, please leave your ideas in the comments, or tweet at us.


Update: New photographs and videos from this location in Parand confirm the geolocation.

Edit: A previous version of this story had a typo in the title for “PS572” instead of “PS752”. This has been fixed, but the typo persists in the URL.

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.


      • Joey

        Did you stop calculating angels, speed of sound, analyzing environmental objects, what was flight altitude. Wanted to calculate it myself. 2800 meters. Speed? I can’t get the radar data.

    • Joey

      Second 25. Did you Analyse the video itself. Why does that seem like a unnatural frame jumped. Didn’t come from the shaking of the explosion.

    • Clark

      Second 25. Did you Analyse the video itself. Why does that seem like a unnatural frame jump. Didn’t come from the shaking of the explosion.

  1. CF

    With the latest video, I’ve established a timeline (a/c altitude interpolated from data blocks):

    0242:47z – PM752 first data block PM752 .25nm off dep end rwy 29R – IKA 3776’MSL

    +01:35.0 (from first data block) PM752 becomes visible to hypothetical launch site from behind terrain PM752 position 6250’MSL N35°27’46.69″ E50°59’45.33″

    +01:51.8 first missile launch PM752 @ 7054’MSL N35°28’31.75″ E50°58’28.22″ (range ~11.2km to launch site)

    +01:58.0 last data block received PM752 @ 7349’MSL N35°28’48.27″ E50°57’59.95″

    +02:10.1 first missile impact PM752 @ 7924’MSL N35°29’27.17″ E50°57’12.56″

    +02:22.5 second missile launch PM752 @ 8518’MSL N35°30’06.97″ E50°56’24.04″ (range ~7.2km to launch site)

    +02:33.8 (0245:21z) second missile impact PM752 @ 9052’MSL N35°30’44.02″ E50°55’38.86″

    I worked out an animation on Google Earth that fits the videos and the timeline.

    Note – I still haven’t worked out the exact position of the launch site, so launch range is estimated. Working on that next.

    • It doesn't add up...

      The flight times of the missiles allow the range of the launch site to be estimated. Allow a second for the initial vertical launch and ignition. Then 5 seconds acceleration to 850m/sec maintained until motor burnout. a=850/5=170m/sec/sec. 0.5at^2 =25*85=2125m. Some slowing in the pure ballistic phase of flight of first missile. Perhaps acceleration is faster, but it’s easy to do sensitivities. Unless the trajectories follow the same bearing where the range arcs cross, drawn from the intercept locations, should give the launch site.

      • Mark Keogh

        I’ve “run the numbers” on the missile time-to-range based on a maximum velocity of 850 to 900 m/s with a decay (after 4-second boost phase to max velocity) of between -15 to -18.75 m/s/s. This is based on the performance envelope data from what appears to be a Russian-language Tor system manual (scan submitted by a Twitter account!)

        Funnily enough, the average velocity of 650 m/s cited in Wikipedia is sufficient for most purposes!

        The ground-range-to-target I got from this basic analysis are between 11.736km to 12.921km for first missile and 7.34km to 7.9km for second missile.

        Trajectory is low, so we can assume gravity losses for the missile were negligible, although the performance chart would already take aerodynamic and gravity drag into account.

        Assuming this *was* a Tor system, based on a maximum velocity of 850 m/s and steady decay of 15 m/s/s after 4s boost phase we get ranges from impact point as follows:

        Missile 1: 11.74km to 12.11km (11.4s)

        Missile 2: 7.34km to 7.45km (18.1s)

        Happy to share any of this data with anyone if they want to check it for errors.

        Unfortunately this can only tell us the launcher (if only *one* unit was involved) was somewhere around the northwest of the facility at 35°33’43.7″N 50°53’55.4″E  and south/southeast of the facility at 35°35’15.8″N 50°53’17.8″E.

      • It doesn't add up...

        If it weren’t, it wouldn’t be the last data block, since the missile destroyed the transponder!

        The best way is to work backwards from the time of the second interception which is at a reasonably known location, and to account for the distance travelled at 275knts airspeed from the flight data in the time in between intercept explosions (23.7 seconds) – which gives a point 3.35km earlier in the flight path, just ~1 second after the last transponder data transmission.

        • Mark Keogh

          This is spot-on. First missile hit only a few hundred meters after the last data transmission: We know the missile time(s) of flight pretty well now from the CCTV footage and from the audio in the Parand video we know PS 752 stayed on course after the first hit. Working backwards with known flight characteristics of Tor missile the first impact occurred very soon after the last ADS-B update.

      • CF

        Not 100% sure, but looking at the data blocks, they were updated at 15-second intervals, with a few missing blocks. This is consistent with other aircraft in that region currently. 8-10 second delay between updates. So the next update simply never came.

        Also, I just watched a THY 737 depart IKA on Flightradar24 and there was a large data gap from almost the same position over Parand that PS752 disappeared until they were past the military area. I don’t want to read into that too much, as it could simply be due to terrain between the a/c and the ADS-B receiver that’s feeding Flightradar24.

        In the US, sometimes the lack of ADS-B equipment on either end means that normal secondary radar/transponder updates drive the data blocks that go to the commercial sites. That update frequency is dependent on the RPM of the radar dish, terrain masking, and a few other things. I doubt the radar in Iran feeds the commercial providers, so it’s likely purely dependent on ADS-B.

        • It doesn't add up...

          You may want to recheck your assumptions

          Once airborne, transmissions are about every 6 seconds, with a couple missing from the record. Ideally, you should also check on earlier flights the same night, for which you would now need a flightradar24 subscription.

          Note in passing that times are only given to a 1 second resolution, which gives about 140m of uncertainty at 275 knots: the aircraft is about 40m long.

          • CF

            Good to know. The resolution on the kml I used only had data points every 15 seconds. I do have a fr24 sub. I’ll update with the more granular data and I might be able to tighten up the timing. However, I don’t think it will change the timing of the launches and impacts. Just perhaps shed more light on the ADS-B loss

          • Mark Keogh

            I have a previous reply buried in the comments somewhere, but as 275KIAS is 319TAS or 164 m/s. True airspeed is greater than indicated airspeed. Any pilots out there might be able to explain better.

            **Anyway, this means PS752 could have travelled 3.9km in the 23.8s between impacts**

            I had hoped the time format was down to hundredths of seconds, ie. 2020-01-08 02:44:55Z.146 meant 02:44:55.146 but this is not the case?

            What do the last 3 digits represent?

          • It doesn't add up...

            I should have paid more attention to your comment. I didn’t realise that Flightradar had more recently posted an additional csv file with granular (=raw?) data. In addition to showing millisecond timing, it shows many more data points. If you scatter plot the time since the previous point against the timeline, it becomes clear that as the aircraft passes through 4800 ft AMSL the frequency of points increases sharply and is maintained until the end of the record. The most likely reason for the increased frequency is that the aircraft was being pinged more often by radar, triggering a response. The missile radar completes a sweep at roughly 1Hz according to data on the Tor system. The airport ATC radar is likely much slower. This looks like evidence that the missile radar was tracking the aircraft rather earlier than claimed. That in turn raises other questions.

    • Ever considered microwave weapons technology from USA?

      There is more to this. And not merely a coincidence.

    • It doesn't add up...

      From 10km underground? No, it’s an earthquake near the major fault between the Arabian and Eurasian tectonic plates. Look up the Bam 2003 earthquake for information on that.

  2. Andre

    On a new video, we see two missiles hitting the plane. However, we can clearly (to my point view) feel that the plan has almost managed a U turn before being hit by the second missile (would match previous statement that pilot attempted to return to airport, a manoeuvre not seen on Flightpath). The video above described seems to be the 2nd missile, the first explosion (10 sec earlier) prompted the user to film.

    Can someone work on this and figure out explosion point 1, explosion point 2, crash site? It seems no news outlet bothers.

    • It doesn't add up...

      Please re-check with NY Times. They have updated the article and accompanying video to show the locations and timings of the missile intercepts, and also the location of the missile launch site, all of which are discussed in recent comments here. The crash site is also well established.

      Things awaiting confirmation include when and where the aircraft was when it was picked up by the missile radar (we only have the IRGC commander’s word on that as being a range of 19km); what was the damage done by each missile; the track of the aircraft from the second missile intercept to the crash site; how the aircraft flew after the pilots were killed, since it was not circling back to the airport under pilot control with the pilots killed by shrapnel from the first missile; what caused the sudden lurch and change of course just before the plane flared up into an orange fireball, and driving steeply to crash in the last few seconds of flight. The latter might include some catastrophic failure of the autopilot, being under further attack from the ground, or the loss of or failure of part of the aircraft making it unstable. Much of this will depend on analysis of the black box. It is possible that a reason for Iranian reluctance to release it is because it may reveal some other action on their part that is not already in the public domain.

      • stargazer

        This is just suggestion that [all] pilots were killed before the plane hit the ground and the flight was not controlled at least partially. Upper front part of cabin, the windows doesn’t look heavily perforated with shrapnel. It’s really depends on what damage was caused by the 1st and 2nd missile. The bottom of the fuselage, bearing structures might shield shrapnel. The crash site makes me wonder that the pilot was looking for the field without buildings for emergency landing.

        • It doesn't add up...

          I presume that the Ukrainian investigators who said that the pilots were killed instantly by shrapnel had examined their bodies to remove any doubt about the cause of death. Shrapnel injuries are quite distinctive.

          • Jeroen

            Yes it might be so but shrapnel wounds do not simply tell if it was the first or the second missile that killed the crew or did they?

          • It doesn't add up...

            The far greater probability is that it was the first missile that killed the pilots. Reasons include the transponder being knocked out by the first missile: it is immediately below the cockpit area; the fact that the plane appears to have maintained its course until after being hit by the second missile (which caused the engine and wing fuel tank fire, and therefore probably hit further back).

  3. stargazer

    TSN-Ukraine published recordings of communications with the Tehran airport air traffic control: a pilot of Aseman Airlines, flight ER3768 approaching Teheran airport for landing saw a missile hit the UIA plane. After that, the dispatcher tried to reach UIA 752 many times.


Leave a Reply

  • (will not be published)

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