I guess they aren’t rare for Israel. But Israeli launches are themselves rare. They launched once in 2023 and zero times so far in 2024. That’s compared to 116 US launches in 2023 and 107 so far in 2024.
Here are two satellites at the time of collision:
This is what the debris field looks like 10 minutes later:
And 180 minutes later
Thanks for the STK illustrations, @DPRK.
Stranger
Huh, I’d have expected much greater dispersal than that, basically a uniform explosion around the center-of-mass trajectory.
Was that an actual event or a simulation, and if a simulation, what’s the software?
It’s an actual event. A fact sheet here:
And the wiki page:
Only a relatively small portion made contact. Which should be expected since satellites have antennas and solar panels and various other bits that increase the radius. So a main body to main body collision is less likely than a glancing blow.
There’s still an enormous amount of energy involved, so both satellites were broken apart into lots of pieces. Though there were some intact pieces left.
You can see from the second image (just post-collision) that in fact the imparted velocities relative to the trajectory of the original body are broadly distributed. But the components of the debris field post impact still have virtually all of their original momentum, plus or minus whatever exchange happened at impact, so you essentially get narrow stream of debris still mostly following the same orbit, as showing in the following images. Some of the pieces will end up going slightly faster or slower depending on how the body broke up, and therefore changing the eccentricity of the orbit, which is why in the C+180 minutes image you see the field spread out with the faster elements somewhat above the original orbit and the slower ones slightly below, but they just don’t fly off sideways to the original azimuth because it takes an enormous amount of energy to make an azimuth change. As time goes on, these fields will spread out a bit more, virtually eliminating self-collisions.
That is also why just a few debris fields like this aren’t a big concern, because they can be tracked and the evolution of the distribution predicted within statistical bounds, and so spaceflight trajectories can be selected to avoid them and large spacecraft and space stations can perform collision avoidance maneuvers. However, the more of this debris you have in orbit, the more likely these fields are to interact, or to collide with some defunct satellite or expended upper stage, and once you get to a certain volume of a debris along a certain azimuth and altitude it essentially becomes self-perpetuating, at least at human timescales, which is the phenomenon described under the label of Kessler syndrome.
As @Dr.Strangelove noted, this is from a real satellite collision in 2009 between an Iridium (commercial satcom bird) and a malfunctioning Kosmos (Russian telcom) satellites. The data likely came from the then-US Air Force Joint Space Operations Center (JSpOC), and the visualization software is Satellite Tool Kit (STK), which is widely used for trajectory and orbital simulations. It also has some capabilities to do trajectory studies and models, but doing any kind of collision simulation would be done by a dedicated simulation tool, likely using Monte Carlo methods to produce an array of results to determine statistical bounds.
Even a “glancing blow” will do enormous damage and produce a large amount of debris. I think people have the notion of two cars colliding and chucking off a few parts and broken side windows, but this is more like to glass bottles being smashed together; at orbital speeds, which are an order of magnitude greater than the fastest rifle-fired bullet, structural materials such as aluminum and steel will just shatter from the impact (or in the case of direct solid-to-solid impact be heated to incandescence and turn into a spray of molten droplets) and delicate ceramic components of solar panels, radiators, and optics will essentially just turn into giant fields of pebble or sand sized debris too small to track. In fact, the objects shown in this Iridium-33/Kosmos 2251 are only the components above the tracking threshold (probably ~5 cm or ~2 inches) which may only be a fraction of the total mass of the spacecraft.
Stranger
Just to give a sense of scale, the two satellites here were 950 kg and 560 kg. The relative velocity was 11.7 km/s. Even if each satellite contributed “only” 60 kg to the collision, that’s still equivalent to a 1 ton (of TNT) bomb in each satellite’s rest frame. A little over 4 gigajoules each.
All I have to add is that credit where credit is due— that analysis using STK software was done years ago [not by me] and all I had to do was link to some of the images on the Celestrak web site.
It was a real event where the satellites impacted at nearly a right angle. Out of the thousands of pieces tracked, one can see that a few did pick up some significant out-of-plane relative velocity and flew off at some slight angle, but nothing close to 45 degrees.
Video evidence suggests that the Iridium satellite was struck near the top and some of the Main Mission Antennas at the bottom survived relatively intact; that is probably what people mean in this case by “glancing blow”. Relative velocity was on the order of 10 km/s, which puts it firmly in the realm of a hypervelocity collision.
Thanks for the illustrations.
Some thoughts - 10km/sec is what, about 10 times faster than the best rifles? Note how the debris spread out even after 180min (about 2 orbits?) Presumably in a few days this will be a ring of random debris - plus some is obviously departing above and below the original orbit, meaning it will become more like a slim donut of debris. Also it seems to me that at 490mi altitude, the decay will not be much of a factor for a long time. (years? decades?) As it does, the less dense vs crosssection will fall inward first creating a ring of debris.
I can see why tracking this is a really good idea.
That was awesome thanks
Not quite; the reason debris is spreading out is because some of it has gone into a more elliptical orbit, but if you watched an animation the debris train would appear to contract and expand. It would take a very long time (hundreds of years) for the tidal influences from the Moon and Sun to stretch it out into a ring around the Earth. At 800 km (490 nmi) above mean sea level it will have a residual lifetime on the order of 1,000 years (most of that above 600 km), although the lighter particulate debris will fall into lower orbits and burn up in the upper atmosphere long before then.
Stranger
I just queried space-track.org and there were 855 Kosmos pieces and 186 Iridium pieces currently listed.
The decay time for any individual piece is going to depend on the dimensions and mass of the piece and the exact orbit it ended up in, but large pieces are going to stay in orbit for decades.
I think the debris cloud is going to spread out (globally, eventually, due to the different orbits?) but not into a ring in any reasonable amount of time. NB someone with a bit of energy can download the current orbital elements for all the tracked pieces from this collision and create a nice up-to-date graphic 
If I have free time I might try it but no promises…
It’s already quite spread out. I whipped up a 3D renderer (using AI!) with the current TLE data downloaded from Space-Track:
https://scottcutler.net/projects/debris/debris.html
It’s rendering at 100x speed. Red is Cosmos, green is Iridium.
Looking at it again (sorry, this was a quick hack), I’m almost certain two coordinates are flipped. There’s a kind of polar vortex over… Central America. Fairly sure the overall satellite rendering is correct, though. I’ll try to fix this later. Should probably make Earth rotate as well…
Ok, made a few fixes. Earth rotates, the coordinate system is fixed, and I tweaked some positioning. Bad AI! Pick a coordinate system and stick with it!
The lighting is still wonky and I’m sure the phase of Earth’s rotation and the satellites is wrong. Still, there you go. Kinda neat to watch. You can actually see that the Iridium sat had a greater inclination than Cosmos.
There’s some satellites that are going around almost exactly backwards of the others of the same color. That doesn’t seem right.
The debris pieces have had about 15 years to disperse. Due to the oblateness of Earth and the fact that they’re all in slightly different orbits, it means they will precess at slightly different rates–which means that over time, you’ll have some satellites at X degrees longitude and some at X+180. So those pieces will be going in effectively opposite directions even though they came from the same object.
BTW, here’s a (static) pic from a slightly less janky renderer:
That was (you can see from the timestamp) ~2 years after the event, and the debris was already well distributed (though if you squint it looks like the blue particles don’t cover the full 360).
Looking at your animation all I can think of is Ambassador de Sadeski mournfully telling President Muffley about “A doomsday shroud that will envelop the Earth for 99 years!” The actor (Peter Bull) really aced those lines.
Seriously though, that is really amazing that an expert can whip up something like that quickly on just a whim and with reasonable effort. My hat’s off to both modern tools and your own skill in your field. Thank you!
AI tools are great for this sort of thing. On one had, I am an expert in 3D graphics as it’s been my career and hobby interest for my adult life. So recognizing things like coordinate system bugs is trivial. Plus I dabble in space.
On the other hand, I’m a total rookie when it comes to web programming and in particular the various available APIs. There are just thousands upon thousands of them and I can’t hope to remember them. A project like this would take several hours just to research the appropriate components and get some basic understanding of the APIs, which all have their own idiosyncrasies.
So it’s like having a junior programmer available that can do the annoying grunt work for you. Then I come in at the end and fix a few things and maybe clean up the code a bit. Took like 10 minutes of back and forth with the AI and another 20 or so to fix the bugs and make some visual tweaks. Might have been a tad less but I ran into the history length limit with Claude and had to take over a bit early. I should probably start paying for some of these tools.