Saturday 30 January 2016

MH17: On the Resurs-P1 image with "BUK's near Zaroshchens’ke" (twice updated)

(Note added 31 Jan 2016: Max van der Werff brought to my attention that he is "kremlintroll" and that the pdf report on the blog was not written by him, but by someone else. That was not clear to me. I have added some comments in the text below to reflect this)
[this post was editted by the addition of a (marked) new section on 17 Feb 2016]

Since my contribution to the January 22 Dutch Parliamen hearing on flight MH17, a number of people have reached out to me with satellite-related questions about MH17.

One of them was Dutch blogger Max van der Werff, who in a tweet from 28 Jan 2016 asked me if I had anything to say about this analysis (pdf) by a blogger named "Kremlintroll" [edit: this turns out to be Max himself: but the pdf analysis on his blog was by someone else]. It concerns a report with purported Russian satellite imagery of what was claimed to be BUK installations in a field near Zaroshchens'ke, on the Ukranian side of the front. It in addition includes claims that a particular US spy satellite overflew the area at the time the missile was fired.

I will answer Max' question here, but note that I do so without necessarily condoning Max viewpoints on MH17. There are a lot of bloggers expressing their (often very partisan) views on the case, and I prefer not to take sides with any of them at this moment.

At the same time, when satellite observations are brought up, I would like to see to it that the information presented is correct. So I will provide some of that information below, as it pertains to this case.

My initial intent was to keep it focussed at information about satellite passes. This is factual, objective information. While checking pass data for Resurs-P1 I did however note something that appears off with regard to one of the images. I will discuss that as well, with some reluctance: but given that I mentioned my doubts on twitter, it is fair that I should present the reasons for my doubts.

The imagery comes from a Russian government website and was earlier the target of a "photo forensics" analysis by the Bellingcat collective. The latter analysis has drawn fire from professional photo-forensic analysts, who say the analysis was flawed. This critique comes from knowledgeable people on this matter, so should be taken seriously imho.

click image to enlarge

The image (reproduced above) purports to show Ukranian BUK installations in a field near Zaroshchens'ke in Ukraine, 47 deg 59' 00" N, 38 deg 27' 05" E. Text on the image states it was taken on 17 July 2014 at 11:32 (no timezone given, but presumably standard Moscow Time, UTC +3).

The pdf in the blog by "Kremlintroll" discusses the image as part of a larger narrative. In addition (and this was my initial focus), the author of the pdf claims that a US spy satellite actually overflew the MH17 disaster area at the time the missile was fired:

"According to our data from 17:06 till 17:21 Moscow time on July 17 over the South-Eastern territory of Ukraine flew a US space satellite. This is a special device of the experimental space system designed to detect and track various missiles launches"
[edit: the quote above is purportedly from a Russian officer, Lt-Gen Makushev]

US spy satellite fly-over?

Let us examine the latter claim first, as this is squarely within my field of expertise. As we will see, the claim appears to be false.

The description seems to refer to one of the Space Tracking and Surveillance System (STSS) demo satellites in Low Earth Orbit (note the given pass time of 15 minutes in the quote above, indicating a satellite in Low Earth Orbit). They are experimental satellites for tracking missiles, serving a role initially projected for the (cancelled) SBIRS LOW segment.  A pass of one of these satellites indeed takes about 15 minutes.

These particular satellites are tracked by our network. I checked their positions for 17 July 2014, 13:20 UT (the time of the MH17 disaster). They did not have sight of the Ukraine at that moment.

STSS demo 1 and STSS demo 2 (2009-052 A and 2009-052B) move as a pair at a fixed distance of each other: STSS demo 1 passes a given location 10 minutes after STSS demo 2 does, along a similar track.

STSS demo 2 passed over the MH17 crash area around 12:35 UT, and STSS demo 1 did so around 12:45 UT. This is more than half an hour before the MH17 disaster took place. STSS demo 1 lost view of the area around 12:55 UT, still 25 minutes before the disaster. At 13:20 UT, the time of the disaster, the satellites were south of Australia, well away from the Ukraine (map below).

click image to enlarge

STSS-ATRR, a third satellite connected to this program, was just east of Sakhalin at that time, so nowhere near the Ukraine either.

Another US spy satellite, USA 161, did pass over the area near 13:20 UT (see my previous post here). This is however a KH-11 optical reconnaissance satellite, not one designed to track missiles as claimed in the pdf analysis hosted by Kremlintroll. It also did not pass over the south-eastern territory of the Ukraine, but well west of the area in question. See my earlier post for a discussion of what this satellite might, or more likely given amongst others that the area of interest was very periferal in the footprint, might not have imaged.

In other words: the claim about a US missile-tracking satellite in Low Earth Orbit overflying the Ukraine at 17:06-17:21 Moscow daylight saving time (13:06-13:21 UT), is factually incorrect.

Of course, we should realise that three SBIRS satellites for missile detection in higher orbits (HEO and GEO) with (semi-) permanent view of the area did cover the relevant area, as I reported to Dutch Parliament, and in that sense this discussion is slightly academic: early warning satellites for missile launches did potentially observe the Ukraine at that time. This SBIRS high component in HEO and GEO however clearly is not the low orbit (fly-over time listed as 15 minutes) experimental missile tracking satellite system the Russian officer was referring to. So, this specific information is incorrect.

The photograph

So what about the satellite photograph distributed by the Russian government, showing purported BUK systems just south of Zaroshchens'ke and discussed in the pdf hosted by Kremlintroll as evidence for Ukrainian BUK deployment in the area.

I was, and am, reluctant to discuss this photograph.

First, I am knowledgeable about satellite orbits, satellite positions and satellite system specifics (the kind of factual information I provided to the Parliament committee during the hearing of Jan 22): but I am not a photo interpreter.

I want to emphasize this. Analysis of this kind of imagery preferably should be done by a professional photo interpreter.

Second, following my role in the Parliament hearing, I prefer to refrain as much as possible from making any statements that might be misconstrued as 'taking sides'.

Nevertheless, while looking at satellite pass information I noted something about this photograph, that I think is relevant. With some reluctance, I provide these observations here in this blogpost. I want to make clear that this only concerns this particular photograph: I am not saying anything about other evidence presented by various parties, and importantly: I have no desire to analyse more images. Again: that should better be done by professional photo interpreters. In a way, given the way this quickly developed on twitter, I regret mentioning my doubts about the satellite image in public, even though (as discussed below) I stand behind these doubts. Given that I mentioned my doubts on twitter, it is fair that I should present the reasons for my doubts below.

1. the satellite

First, the satellite that took the image. The image lists a time, 11:32 on 17 July 2014. Resurs-P1 (2013-030A), a Russian imaging satellite with 1-meter resolution, passed the area shown in the photograph at 08:32 UT (11:32 Moscow Standard Time, or 12:32 Moscow Daylight Saving Time) - an identification earlier also made on a Russian webforum. Other high-resolution Russian imaging satellites do not match the pass time listed. The assumption that the image was made by Resurs-P1 therefore appears a valid one and will be the starting point of the discussion below.

click image to enlarge

Resurs-P1 culminated at 57.5 degrees elevation at 08:32:46 UT (July 17th 2014 - that is 11:32:46 Standard Moscow Time, 12:32:46 Moscow Daylight Saving Time) as seen from the field imaged in the photograph (which is at 47 deg 59' 00" N, 38 deg 27' 05" E). The satellite subpoint was some 281 km west of the location shown on the photograph at the moment of culmination. The location on the photograph is within the satellites known image swath width of ~950 km (~475 km to each side of the satellite nadir).

It is therefore possible that Resurs-P1 did take an image of this field at the listed time.

That does however not necessarily mean that the image is genuine. Something isn't sitting well with me regarding this image, raising my suspicion. I will discuss this below.

2. the photograph: suspicions

(again, I want to emphasize that I am not a professional photo interpreter. Keep that in mind)

So far, what I have written above is all straightforward factual: where were what satellites at a specified moment? What I am going to write below has a larger factor of interpretation.

As mentioned, Resurs-P1 culminated at a maximum altitude of 57.5 degrees for the location in question (that is a factual observation).

This immediately made me uneasy about the featured satellite photograph. For this image seemed to be a head-on image, taken with an angle to the horizon of near 90 degrees (straight down), rather than under an oblique 57.5 degree angle.

The image should, with a culmination altitude of 57.5 degrees for the satellite, instead show the installations obliquely (i.e. skewed, i.e., not just show the tops but also show parts of lateral sides of the purported BUK installations). It should also show the landscape somewhat obliquely.

Resurs-P1 passed west of the area in question. This is towards the left side of the image. The image should therefore show the BUK installations somewhat skewed with something of the lateral sides visible on the left hand side.

I used sketchup to make a simple block-model with the dimensions of a BUK (roughly 9.9 x 3.3 x 3.8 meter), and then rendered it under a 57.5 degree angle with the horizontal. The image below shows the result as an inset in the original image.

To me, it appears that the photograph indeed should show more of the lateral sides of the purported BUK installations than it does: the installations shown in the image appear to be seen too-much head-on compared to the modelled view under a 57.5 degree angle.

There is also something off in the positions of the shadows. They should be slightly displaced to the left, compared to what the image appears to show (the sun was at azimuth 150.7 degrees, i.e. southeast, and elevation 60.6 degrees at 8:32:46 UT. The shadows should point slightly to the left in the image, even more so if the image is taken from a point to the west of the area under a 57.5 degree or smaller angle).

Taken together, it suggests that the photograph does not show a scene compliant with the geometric situation at 8:32 UT on 17 July 2014 as seen from Resurs-P1.

As a caveat, I should emphasize that we are talking about structure at the edge of the satellite's image resolution here, with objects only a few pixels wide (and again: I warn that I am not a professional photo interpreter). There might also have been done some unknown image processing, including resampling, as part of preparing the photo for publication. Still, I feel there is reason to be very cautious about these images.

[Edit: extra material added Feb 17, 2016, now following:] 

The previous discussion was from the viewpoint of an image taken around culmination of the satellite for the area in question, 08:32:49 UT. In a follow-up discussion about this post on Twitter, it was brought up that the image might have been made before culmination when the satellite was still north of the location. That would better match the listed time, 08:32.

I therefore did a new model simulation, based on the satellite-to-location angles for 08:32:00 UT (azimuth 335 degrees (i.e. N-NW), elevation 45.1 degrees). It is given below. Again, it does not appear to match. It should be noted here that from known specifications, 45 degrees is near the limit of what the satellite needs, in terms of elevation angle, in order to image a location.

click to enlarge

In the other imagery provided by the Russian MOD, there are discrepancies too indicating that the images were not taken at the claimed dates and times. A very clear one is the one below (source), that purports to be an image taken on 14 July 2014 11:40 Moscow time (08:40 UT). The Russian optical reconnaissance satellite Kosmos 2486 (Persona 2, 2013-028A) culminated at 08:40:54 UT at an elevation of 83 degrees for this location.

It is however interesting to compare the configuration of vehicles on a parking lot in that image . These vehicles do not seem to be used much. Using Google Earth imagery from April, May and July 2014, a clear long-lasting change in the configuration happens before July 2nd: one vehicle is lightly replaced, and a new vehicle is parked in what previously was an empty space.

click to enlarge

Comparing this to the vehicle configuration on the Russian MOD image from "14 July 2014", shows the first vehicle still in place and the empty space in reality filled in before July 2nd, still an empty space. So I am in agreement here with the observations made by Bellingcat in June 2015: the Russian MOD image claimed to be from "14 July 2014" in reality is from before July 2nd, 2014.

"But how about the Ukrainians?", some have asked me. Well: the vehicle pattern does positively fit with the "July 12" image (upper right image above) published by the Ukrainians (and later ostensibly 'rebutted' by the Russian MOD here: but we now note it are in fact the Russian MOD images that are doctored).

This latter image appears to be commercial imagery obtained from Digitalglobe, shot by the Worldview 1 satellite (which isn't anything unusual: several governments use the services of Digitalglobe). It made a pass culminating at 55 degrees at 08:11 UT (11:11 local time) on July 12.

Note that the Russian MOD claim here that:
"At the time specified in the images, the American electro-optical reconnaissance satellite of the Key Hole series was flying over the crash site area, so the source of the images for Ukrainian Security Service is obvious"
This Russian MOD statement about the Ukrainian images is factually incorrect (as was their claim for STSS passes around the time of the MH17 shootdown, as we have earlier seen): the dates and times listed for the Ukrainian images do not match with passes of  US KH-11 optical satellites.

But they do all match passes of three Digitalglobe satellites: Worldview 1, Worldview 2 and Geoeye 1. Evidently these satellites made these images, not US military KH-11 satellites.

So the Russians are making a lot of factually incorrect claims in this case.

[end of section added 17 Feb 2016]

A few last words on my position in this

I want to note that this is the first and last time I will, with regard to MH17, foray into the domain of satellite photo interpretation, also given the very strong reactions this has already spawned on media like twitter. Agitprops from all sides have jumped on it.

I prefer to keep my further involvement with MH17 focussed on providing factual data on satellite positions and satellite system specifications, in line with the role I played in this month's Parliament hearing. That is a position only people with an agenda could take issue with.

With all the things that have come up over the past 1.5 year, and with multiple parties involved clearly unwilling to provide pertinent data, I have taken a step back, opinion-wise. In the current situation, my aim is to not point fingers until verifiable evidence is put on the table. The report of the Dutch Safety Board is a first step into that direction, but there are still many aspects of the case that are very unclear. Clarifying these was the main point of the Parliament hearing of January 22.

NOTE ADDED 4 March 2016,  10:30 UT:

- sigh.....deep sigh.... -

I wish it would not be necessary to get into this, but some spin deserves a rebuttal when the spin in question really is too outrageous:

As an example of how some propagandists are trying to spin unwelcome analytical results (such as those of this analysis), I bring you the twitter troll Deus Abscondis.

For reasons only known to him/herself, (s)he tries to (falsely) insist that I initially ignored the Persona 2 satellite. The goal, of course, is to try to suggest that my analysis should be dismissed and my expertise questioned.

Here is the flaw in his/her logic: the fact that I did not mention Persona 2 until Deus Abscondis named that satellite on twitter (and I answered that Persona 2 appeared to have made some of the other images) simply does not mean I didn't consider Persona 2 earlier. I did, matter of fact.

It is important to realize that my blogpost initially (and at the moment Deus Abscondis posted his twitter question) discussed only one image: that with the "BUK's in a field near Zaroshchens’ke".

I wrote about this image, in the original blog-post (emphasis added):

The image lists a time, 11:32 on 17 July 2014. Resurs P1 (2013-030A), a Russian imaging satellite with 1-meter resolution, passed the area shown in the photograph at 08:32 UT (11:32 Moscow Standard Time, or 12:32 Moscow Daylight Saving Time) - an identification earlier also made on a Russian webforum. Other high-resolution Russian imaging satellites do not match the pass time listed.

Take a look at that last sentence.

The sentence highlighted in black makes very clear that I considered other satellites than Resurs P1 as well. The latter category of checked satellites not fitting the pass time for the Zharoshchens'ke image actually included Persona 2.

In addition to Persona 2, the ones I checked but rejected also included Resurs DK1, Kanopus V1 and the Kobalt-M photoreturn mission Kosmos 2495 (still on-orbit at that time).

When Deus Abscondis days later asked about whether I had considered Persona 2,  my answer to him was that Persona 2 could have made some of the other image, i.e. I told him/her that I was well aware that some of the other imagery could be from Persona 2. I also informed him that this other imagery was not unproblematic either (and updated my blogpost next to show why):

Note that nowhere does my answer implicate that I had not considered Persona 2 in my original analysis, of the Zharoshchens'ke image. I simply point out that Persona 2 is of relevance only to some of the other imagery, not to the image discussed at the blog at that moment. Discussion of that other imagery was added later, as an update of the post.

The twist that Deus Abscondis is now trying to give to that conversation, is therefore extremely insincere.

It is a good example of the kind of spin some propagandists are producing when trying to suppress unwelcome analytical facts.

Note added 10 June 2020:

In the MH17 trial yesterday (June 9 2020), the prosecution tore apart the validity of this imagery as well, based on an analysis by ESA.


Thomas said...

I am neither an expert on satellites, nor on image intelligence, my only asset in this case is my open mind, but please allow me to comment on this.

As I understand it, the Resurs-P is the civilian cousin of the military Persona high-resolution imaging satellite, pictures of which apparently also were used in the image war surrounding MH17.

So are the raw data from this satellite commercially available, or does the Russian DoD have privileged access to that? That the published images from the military satellite are not raw data should come as no surprise even without being aware of the Secrecy Fetish of the Russian military. But why should the Russian DoD "photoshop" images from the Resurs-P? Maybe your observations provide the answer: generally satellite images are associated with pictures taken from directly above, aerial photographs from oblique angles could also be taken from drones or airline passengers, which would make the dating uncertain. There are certain algorithms in programs like photoshop to "de-distort" pictures taken at oblique angles.

So maybe both sides are correct: the picture was indeed taken by the Resurs-P satellite, but photoshopped so as to appear being taken from directly above?

Unknown said...

As far as I understand, "de-distorting" of ground surface is routinely done, but it cannot "re-distort" objects above the ground such as buildings or vehicles. Because only single 2D photo is available, neither 3D data nor a stereo pair.

Unknown said...

>> But they do all match passes of three Digitalglobe satellites: Worldview 1, Worldview 2 and Geoeye 1. Evidently these satellites made these images, not US military KH-11 satellites.

You are wrong. Russian do not say about KH-11. They called all optical reconnaissance satellites by "KeyHole type" (e.g. GeoEye or WorldView). That's why they write about that on mh17 report.

SatTrackCam Leiden said...

"Keyhole" or KH-11 or ADVANCED CRYSTAL is the name of a series of US military optical reconnaissance satellites. Commercial imagers from Digitallobe are optical satellites but they are not Keyhole satellites. If the Russian MoD refers to these as "Keyholes", then that is simply stupid.