September 19 Donetsk Attack Crater Analysis

October 5, 2022

(rough, incomplete - edits 10/6, 10/7 - updates 10/15-16)

I feel bad I haven't tried my hand at analyzing any of these ongoing attacks in the Donbas and notably in Donetsk. A terrible one occurred September 19, killing a reported 16 people at least. I had seen the heartbreaking images of people dead, a few of them after something tore off their legs or whole lower bodies ... a woman severed at the pelvis who seems to have dragged herself a bit before she expired - a severed hand left laying on the street - a heavy scrap of flesh hanging off a metal bar ... 

It seems this was one of 2+ impact scenes in Donetsk that day: teleSUR English "The Ukrainian armed forces attacked the Baku square, where the various explosive devices hit a bus stop." 13 and then 16 were reportedly killed. Only 8 fatalities are visually noted at the scene in question, which is next to a sort of square and marked as a bus stop. While another 8 could easily die in the hospital, there's a photo with the article that shows a burning van and truck in another spot not at this square, which I couldn't easily find, There's also video of stalls in a public market that are torn apart. There may be a few fatalities at these other attacks, but so far, we have a good view of just the one scene. 

Otherwise, I haven't followed the story or caught up - Skipping context like who's said just what about the incident (someone else should take this and work it in with that). But generally, as usual, the locals and the Russians blame Ukrainian forces for another terrorist attack to demoralize and kill their enemies, while Kiev denies it, either implying or explicitly stating the Russians or Donestk forces deliberately shell their own people just so they can blame Ukraine.

I and 2 others were consulted by Ayaz K on Twitter to double check some "crater analysis" work presented in a video by Ryan McBeth, claiming to absolve Ukrainian forces for this attack with forensic science. This was to help inform his article Fact Check: Ryan McBeth claims that the DPR shelled themselves in Donetsk – Telegraph. Some discussion on Twitter with he and I and Michael Kobs, and this blog post - which is evolving daily ATM - seek to inform that and everyone, as possible.

The video in question: Donetsk City 9/19/22 Attack Crater Analysis - YouTube. I'm glad I was invited and checked. I knew nothing about who that is and who cares, but over 194,000 views now, 16,000 net likes, and several positive comments from people who claim to also know their stuff, suggests it's all relevant enough to bother with. Skipping other things I could say on McBeth's method and reasoning and motives ... it seems he doesn't really understand the processes involved, and got his science wrong. That's how he decided on a firing direction that's roughly opposite of the real one. Although I'm still not a true expert, I do get it and I can help set things straight here. 

The actual crater analysis takes about a minute of his 14-minute video. He brings up a common but weird (to me) diagram from an old US Army manual and the scant supporting text. See at right. It's been shown to me by many people, most of them hacks of the Bellingcat school who don't seem to understand it any better than I do. Now I have seen this almost Batman logo shape appear with shell impacts in a dirt field. I guess that's dirt drawn. But there usually are no stakes in the images we have questions about, and a compass is no help with a video. There's often no pile of dirt but instead dead bodies and fragmentation marks on pavement, waves of smoke stains, and various things getting damaged in their unique ways. If you look at it and think about it long enough, considering what weapons would do at various angles, you can get the answer that keeps being right. 

Ryan McBeth did not do that. He noted the manual's explanation the blast effect (and thus soil or debris) moves "forward and to the sides" - NOT to the rear. And so he decided at 4:55 - "there should be "more debris on THE REAR of an impact," (emphasis his). He doesn't notice that makes no sense, but the graphic is confusing; he took a stake in the middle as the impact crater, and pointed to the actual impact point behind it as "the rear." Dirt seems piled up to a point behind the crater-stake, but how and why? I suggest that's just the point of their triangle, the start of spread forward and to the sides FROM the point where shit was blowing up - not from a point a bit ahead of that. 

To clarify: I think his arrow points to the unseen crater or point of impact. The rear of it would be above this arrow, and there's nothing there. The "stake" in the middle is just that. I don't know why they seem to have put in a black hole. That's confusing.

Next, he mistook some debris from the wall to the left - that must have deflected back to land right in front of the crater - as debris blown forward from the crater. I'm pretty sure there were no bricks, nor an entire window frame embedded in the pavement. But McBeth sees these things a few inches from the crater, ...

Add 10/6: That's fun enough to just leave, but I assume he meant the broken asphalt piled up at the crater's south edge, tucked partly underneath that window frame. I marked this orange below. It seems to be a valid indicator. End 10/6.

So he decides ... that's the rear. "The arrow of debris" - an illogical arrow he just drew between the crater and the biggest visible stuff - "points south," so the rocket came from the south before blowing this stuff slightly backwards. 

Then he took nothing in particular that was on the perpendicular to that, called it "side splatter," and drew 2 arrows to show where there should be something but wasn't. He even showed these first as if it meant anything, before showing how it proves a southern origin. South is where is where his illogic wants to point. We can't be certain why that is. It's one of several directions that tends to implicate DPR or Russian forces (below: from ISW map 9/19). 

He also notes that windows to the north were broken because the blast moved "forward" - which it would do, and a bit less in all other directions - and he decided that forward was to the north. And then he mentions how glass across the street was broken as well - because the blast moved to the north and to the south? 

And so he decides the shell came from the south, "behind DPR lines." Having dropped that "truth bomb," he then shrugs like Atlas in dramatic silence and sips some water before continuing to explain why that only makes sense. He doesn't suggest the separatists attacked their own on purpose, as some will have decided. Rather, he concludes they meant to hit Ukrainian forces further north. So it was an accident - or at least 2 of them, considering the multiple strike points.  He goes on and on about the kind of shell the DPR guys probably used, where to the south they fired from and why, and what went so (tragically) wrong. 

But Skipping all that, I need to show why his central analysis underpinning all that is unsound. He has a ton of fans who've said he's right, but I think they're all wrong, and I can only guess just why. We now go to the visuals, and we can move quickly on text.

Cameras Not Lying, but Needing Some Translation

Scene geolocation: McBeth did the work, and it seems sound. A few possible issues probably resolve like he says; I've seen the same with things looking different from space, or street-level views in Google Maps being many years old, missing later changes. Location: on Shaposhnikova St at Bohdana Khmel'nyts'koho Ave, in the Kuibyshivs'kyi district of Donetsk - coordinates 48.0101689, 37.7699298. The marked-up wall is of the shop on this southeast corner, with impact crater just a bit northwest of that, in the street near the curb's corner. It seems the opposite corner to the west was the bus stop, where I guess people were gathered waiting for the next ride.

Body geolocation: Eight fatalities all near the crater, and mostly just north of it. Again, McBeth did the work, and it seems sound. All mapped on an earlier image from Google Earth. Body placement was just copied over, and a bit approximate.

Looking to the SW corner shop, awning and windows gone from both sides, and the clearly visible crater at the soldier's feet, plus dead bodies not terribly visible:

When you don't have that army manual, some triangular dirt splash or even "stakes," just fragmentation marks and smoke stains on pavement and wall, and variable bodily injuries, what do you do? Consider it all together to see what provides the best single explanation.

Pavement Markings

Michael Kobs used an existing view for this analysis. I'm still not sure that's quite it - marks are clear to the right, not so much at the bottom or on the left. To get: clearer view of an angle like this, maybe after debris cleanup.  

That's a splash/rose/sunburst pattern on pavement. In most impacts, it will be like below - an arc of intense marks less than 180 degrees - from the underside closest to the pavement - the middle of the arc and often the longest and sharpest line is the one that best points back to the firing position. The next image helps show why - the fragments radiating from the back hit the pavement at very close range and after very little spreading, making marks that sometimes blur into etched lines. Along the sides the impacts may be more glancing, then further out, none at all. And it all depends on impact angle; if it comes straight down, there will probably be no pavement marks in any direction - it will all go into walls, people, etc.

Another crater view - Eva K. Bartlett photo. She blocked me years ago, probably for saying "Assad regime" as shorthand or something like that. So I can't ask directly, but if anyone there can get us (or has gotten us) a clear view of this pattern before it's gone, that will help with the record.

Enhanced, likely frag marks painted red, maybe missing some and including some other things, and getting less sure what is or isn't right in the middle. I see here and in the wall marks 2+ frag sizes - tiny ones punch in and burn the pavement? They seem to get bigger and heaver to the right, chipping away pavement? That seemed like the back direction I was looking for, but it didn't fit the rest of the damage so well. not sure from these views, but this might be more what we're calling side-spray. That would be similar to what McBride labeled, and what he meant, and he was right ... but he had it backwards. All told with the opposite wall also marked ... 

Here may be where and how the lowest fragments, among those sent backwards, impacted the ground - concentrically from the middle: the lowest fragments dug the back side of this crater - the next ones up removed the pavement a ways out - the next layers up marked the pavement heavily for a span - the next ones more lightly and loosely. 

The pattern looks odd to me - maybe foreshortened so we hardly see back spray, and instead see more to the sides, because the munition came in abnormally horizontal, following an unusually long-range flight. Or maybe it's because that is just the rose spray behind impact, and it points back to the northwest, but I added too many dots in other areas. This analysis should not be taken as gospel, but as a point of reference for further consideration.  

Back to the shell - Fragments facing down will hit the pavement quickly, the middle ones most directly, and the outer ones up to the sides at more oblique angles. Those packed in the side and forward will all radiate above ground, increasingly high as one looks in the forward direction. 

Any wall to the side of such an angled disc of damage may be marked with fragments in a sloping line that's roughly perpendicular to the incoming munition's vertical flight path. If it came straight down, its damage band would be horizontal.  With such an angled pattern, up = the forward direction of flight. Add 10/7: Michael Kobs modeled the "doughnut" of fragments to help illustrate why up = forward. I labeled this frame at right. 

See the opposite wall above, where the burning frags and the detonation fireball hit the wall, sloping up in a direction that is SOUTH. Therefore, the projectile came at least partly from the north. Above: two possible estimates - I drew the red one first, but the magenta one looks better.

It probably won't be from due north parallel with the wall, but it's far from perpendicular into the wall. Consider that we don't see a line, really, but an arc of marks that would roll off to horizontal at the front, while angling more on the sides. This is clearly more side than front. That angle would seem to be a bit into the wall, not away from it. So this is probably some kind of northwest, but one leaning well to the north aspect. But it could be from due north, and come to think of it, it could be from a tiny bit east of that as well, but even then, simple "north" is the best way to put that option.

Next: the nature and pattern of casualties tells the same story. 

Out in front and to the sides at mid-distance, people would face a scattered but body-wide danger over a decent span - they might be hit in the head, torso, or lower body, or all of them, depending. Closer in, the frags will have spread out less, making intense machine-gun-like contact, that tears off body parts, more likely. 

Ahead of the blast, it's more likely the head that will suffer, but far enough out, the fragments might angle totally above people well before they would fall short of them. add 10/7: crude sketch. 

Behind the blast: those quite near might have fragments passing just above the ground and intensely, puncturing and even shredding away feet, legs, and whole lower bodies, besides more scattered hits to the middle and upper body. Further behind, the intense frags are used up, and the danger zone of lighter ones might be horizontal for a while, or might angle down into the ground, besides having scattered out wider. Severe injuries here remain possible, but probably not dismembering or fatal ones.  

Add 10/7: the mapping of bodies above is just approximate, based on a quick copy of McBeth's mapping. I MIGHT have a closer look at who was injured where on their body and where on the map more exactly, to improve the usefulness of these clues. But so far I haven't. End 10/7

Ahead, to the south, and along the side: intact people with legs have fallen from hits to ... mostly unclear to me. I saw one that was hit in the belly and lower leg, at least. Likely the head too, causing him to die like that.

Behind, to the north - low fragments hitting a pillar - the expansive and high-rising detonation fireball wrapped around the pillar above that, but not very far above. This isn't forward. people with legs and lower bodies torn away, a stray shoe, scraps of flesh, river of blood, all within a few meters of the crater, some probably after crawling a bit or being moved - most likely further away than they had been. 

Further back: no one who fell and died and was left there, anyway. As McBeth noted, the severed hand was near the furthest body out on the west side of this area. A few others may have been left with dangling limbs, massive bleeding, punctured lungs, and other fatal problems, but they weren't just mowed down like this. This view shows where they fell, vs. the only fragmentation marks I can mark out on this pillar. It's marked to about halfway up, not to the top. At the bottom are some of the heavier frags. Seen closer, there will be more marks than this, but this is a low band of marks compared to the others, here a bit north and east of the impact.

And this too fits the overall, multi-faceted picture of a munition - be it small rocket or mortar shell or whatever - that was fired from the north or the NNW - the one narrow range of directions in which Ukrainian forces were still operating within a 10km radius. They've always terrorized the people of Donetsk and denied it, blaming the separatists or the Russians for shelling their own. Thay also could miss their targets - 2 or 3 times - OR they could more logically have done it intentionally. And while the forensics here don't point exclusively to them, they sure as heck don't point exactly away from them like McBeth strangely concluded.

McBeth Response:

(added 10/6) I'm hoping for some non-hack response, including to my aggressive tone. I'd prefer that over a weaselly evasion with that tone as an excuse. It's difficult to admit you're wrong, but still it's important. I'm ready to do that myself, it I can actually see how I was wrong on any point here. 

But ... (later) it seems he's blocked Michael, has muted or is ignoring me and Ayaz. He had three things to say to or about us before that. To Ayaz: "Please read Army FM 6-50" (the manual). Like that can explain what went wrong in his brain. To Michael, who mentioned a Grad rocket or a mortar shell as examples of what might have been used, he said: "You really need to think about this. Why would you launch one grad rocket? Gran Rockets are launched in multiple waves. This could not have been a mortar round because A 120 mm mortar would’ve only originated from inside DPR lines. This was a short round, unintentional but tragic." It came from behind DPR lines, he thinks. But still, it couldn't be a 120mm mortar shell. I don't have an opinion what it was except fired from the north-NW. Ayaz and I both asked him for comment on our findings, but he never replied. Instead,  he mused in a tweet "It’s pretty cool when people are so afraid of you they say you are wrong in a “fact check.”" People do that when they think you're wrong, not when they're afraid of you. Afraid people will try not to provoke you, like with a fact-check. Or they'll ignore, mute and block you. Ayaz replied "Yea the person who was asking for answers before writing this article and didn't get a clear answer is totally the afraid one." Gaslighty McWheelspinner, KJohnson,  came to McBride's aid, explaining to me "He is ignoring you. His analysis was far better than yours. Harassing him to change his opinion & take down his video is pure desperation on your part." As I replied "I was giving him a chance to not be a lying hack like you. And giving him extra chance too. WAS. Past tense. He's just a lying hack who covers for mass murder on political lines just like you, Bellingcat, a bunch of corrupt governments, and millions of other soulless people. Sad." Apparently, anyways.

Add 10/16: was I too harsh? A quick review of McBeth's "investigation": "I believe that the dead bodies in the video are #Russian #propaganda" (here) in part because "the clothes of the dead don't match up with the weather that day & mortar rounds don't cut people in half like that." (here with 1st video). in a full thread with the later video (here) he cites an ancient lithograph that I for one can't decipher, and so he decides "debris are highest at the rear of the round," then drew an arrow pointing south where there's a bit more piled debris, pointed out how it points south, etc. "So in summary, I think this was an accident, but it was almost certainly caused by Russia or the DPR." Because it came from the south. But also for good measure because "I think we can effectively rule out Ukraine randomly shelling cities" - even if his analysis did get it backwards. (here)

Updates October 15: 

I finally got a grip on the "splash" pattern that seemed to point west while other clues seemed to point more north. It has different parts I can't fully explain, but the overall shape seems to fit all the more visible bits into a logical whole that includes some invisible parts. The left/north side (bottom here) was unclear compared to the right. This view, color-enhanced, revealed that besides the smoke stains across the area, a different, almost greenish shade appears, in a shape which I traced in pink. Whatever caused that, it fits well for the front edge of the splash pattern, encompassing the few visible marks in the pavement here (which I popped out in red as I think I see them), and it gives them a logical shape. The marks on the right were already visible, and they mirror that shape. The (pink) forward edge isn't visible here under the window frame, but I traced it roughly to match. Quite roughly. Ejecta in orange - piled rubble still seem off center to the right/south, but there's some on the left and more hidden under stuff.  I'd bet it's a bit right of center all told, but pretty symmetrical.

Note: that sharpens a video blur, giving everything a melted-in-tar look. But I think the asphalt also may have been extra soft, leading to more rounded, less visible impacts than usual. 

Using that with other views I tried to trace the same basic pattern and use it to trace the trajectory. From the crater roughly to the former corner (vertical line here), where the wall bends to a 45. See 2011 street view below for comparison. I think the low wall to the right is intact, just defaced, punched through, and rounded, and lost the window frame it was holding up. To the left, the wall looks fairly intact as well, but detached at the bottom and pushed back? Those panes used to meet, but now one seems to hide behind the other. Or is that just the corner and my eyes trick me? 

Crater placement: in the images above, I marked a grid point in yellow-green, with a line west from where the 45 angled wall ends, and another line running north, set the car's width from the sidewalk. Note how they intersect about at the window frame's corner. The crater is just a bit east and 1 or 2 bits north of that grid point. The trajectory is from there to where the 45 wall starts. 

This would mean I was wrong about damage angling up on the (main) wall. I connected too many dots on irrelevant lines. Ahead of the blast at that corner, the band of damage should be highest and level, and then actually sloping down to the right. I was just noticing marks fall pretty top to bottom all across, and thus the pattern is mostly unreadable. However, the lower marks on the pillar to the north are still relevant, and would fit this angle about the same way, being more on the side of the arc. So the bands I drew above from the low-marked column - where 2-3 died in the canopy area - are still pretty applicable, but those closest to the impacted corner should suffer more widespread injury, being ahead of the blast. Any such might be blown into the shop where we don't see them. 

I'd expect the legs-gone zone behind the blast to be north of the doorway, in the canopy area, out into the street, and at the opposite corner, mainly around the arc in purple at right. There would be roughly 0 people standing in the street, one to the west but a ways out, where frags would angle higher and wider - no longer likely to sever parts. Around the covered area and pillars north of the doors, any person getting their legs torn off will also be moved with the same force, maybe winding up under the canopy or up to the doors like we see, maybe after deflecting off the pillar or the roast chicken booth (dark green).

The line I drew to firing spot measures 304° on the compass. That's an estimate, but a good one. I'd venture it's actually a hair lower, and to give some room for further error, I'll suggest 303° +/- 3 degrees, or a trajectory of 120-126°.   

Finally a bit on who has said what: Donetsk Mayor Alexei Kulemzin on Telegram: (translated) 

"at 12:15 and 12:23 a total of 9 shells of 155 mm caliber were fired in the direction of the Kuibyshev district ... from n. n. Netaylovo."

Google maps has this town as Netailove, almost exactly 20km from this shell impact, at bearing 302 to the middle of town. My 303 reading says the same, as does 304 mapped here, specifying northern half or outskirts are more likely included in the line of fire. I wouldn't say if it originated inside the town, closer on the line (SE), or further out (NW) from the strike. But it will be on that line or very close to it.

I won't have a lot to say on how the scene lines up for a 155mm shell, except to note that McBeth doubted it, and while his judgment seems to have no value, maybe he's right anyway and a different kind of shell was fired from around Netailove. I won't have a lot to say on what fired the shell and how the range compares, or who exactly was operating there. But whoever it was, they might realize the shells they fired at 12:15 and 12:23 killed at least 13 innocent civilians at a bust stop, maybe 16 or more, besides terribly injuring many others. No one's holding them to account, and they probably cheer the fact their work that day sowed more terror than usual. The mayor said "Do not go outside unless absolutely necessary! Take care of yourself and your loved ones!" The targeted bus route was suspended for safety. That's the idea. Daily life needs to be made unlivable - and sporadically ended - for the people of Donetsk, as long as they resist their "liberation" from a "temporary Russian occupation." It's been over 8 years so far.