Geoff FordenThe Future (of Space) Is Now

Iridium Constellation. Note how the orbits cross each other near the poles

Iridium and Cosmos Satellites Collide!

Some of us have been warning about a catastrophic chain reaction of space collisions that could render near Earth orbits unusable for tens or hundreds of years. Unfortunately, it is very likely that we are going to see a key factor in that scenario tested over the next year: would a collision in one orbit seriously affect satellites in similar orbits but with different inclinations? How about different mean anomalies? (Basically the slot in the orbit each satellite is assigned.) The collision between an Iridium satellite and a Cosmos satellite could act as the test we need to examine this possibility. No one is happy, however. Look to these pages for over the next week as more debris is cataloged to see what is happening!

Jonathan McDowell reports on the amateur satellite observers electronic bulletin board that it was Iridium 33 (ID=24946) and Cosmos 2251 (ID=22675) that collided at 16:56 UTC on the 10th of Feb. 2009.

click on the image for a larger version

The collision took place at about 760 km altitude, one of the most populated altitudes!


  1. Josh

    Geoff, at first reading, I thought this was a hypothetical. In case anyone else is subject to the same confusion, here’s the story:

    Not good.

  2. Azr@el (History)

    Place a global mooring fee on every gram of dead material in orbit. The fee could be proportional to the orbital belts the debris crosses. Thus every dead satellite or booster would incur a yearly mooring fee to it’s operator. Thus there would be incentive to get rid of this stuff or boost it into stable low fee mooring belts. Increase the mooring fee every year ‘til the growth of the debris field begins to subside. Use the proceeds from the mooring fee to offer a bounty on filtering the more trafficked swaths of the field.

  3. Maggie Leber (History)

    What a great idea, Azr@el….who do you suppose will be in charge of collecting such a tax? Do you imagine that the Russian Federation will be willing to pay for items orbited by the USSR?

    Sounds rather like you’re accustomed to the London congestion fee. 🙂

  4. Muskrat

    If both of these objects were known and catalogued, how could they collide? Doesn’t Iridium have a plan to deal with possible collisions? Wouldn’t NORAD have warned them if a collision were imminent?

  5. Geoff Forden (History)

    So far, no debris has been cataloged (as far as I can tell). But this is hardly surprising. It takes a while for both the debris to separate from each other so that they can be uniquely identified but also it will take a few passes before reasonable orbital elements can be determined. It took a week before the the first debris from the 2007 Chinese ASAT were cataloged, though I think it took considerably less for the first batch from the USA 193 shoot-down were posted.

    On a perfectly techno-wonk level, I wonder how they are going to be listed in the catalog? Will those that are going in roughly the direction of one satellite be assigned as debris from that satellite? What about those in the middle? (They are likely to be the first cataloged since they will be the first to separate out from the swarm.)

    I take that last part about the debris in the middle being the first to be cataloged. Its possible that they will be the last cataloged, if they are cataloged at all. Its going to be hard to find them!

  6. Wang Ting

    To Forden,

    As the addition velocity of most cataloged fragments will smaller than 2 km/s, and the angle between two satellite orbit is 80 deg, I think it is not difficult to know the source of the fragments.

  7. Geoff Forden (History)

    Hello Wang Ting,
    Funny things happen in hypervelocity collisions. For instance, almost all the highest energy debris pieces (that ended up going in the direction of the target FY-1C satellite) were from the interceptor that coming at a nearly head-on collision. But I agree that they will certainly be assigned to the satellite in whose direction they appear to be going.

  8. Brian W (History)

    Also the satellites were in two different inclinations (86.4 and 74 deg). That plus what Ting pointed out should help the sorting process.

    But I agree, it’s going to be messy and the real question is whether there is anyone left at the JSpOC that can handle this magnitude of breakup.

  9. Tal Inbar

    interesting animation of the collision:

  10. Wang Ting

    Things will be much more complicated if the cross angle of two orbit is small, but in this case, it is not difficult to determine the source.

    In FY case there is no cataloged fragment get addition velocity greater than 2km/s.

    If that is true for this case, the simplest way to determine the source of debris, is to calculate delta velocity of debris for both satellites. Satellite corresponding to the smaller delta velocity can be regard as the source of debris.

  11. David Wright (History)

    I’ve just posted a short analysis of the debris consequences of the collision:

  12. Tal Inbar

    AGI uploaded a details clip of he events and orbital parameters of the satellites:

  13. krepon (History)

    Can someone answer the following questions: Did Iridium receive a warning notice? If they had, could they have done anything about it?

  14. Gridlock (History)

    Headline in satirical newspaper I read:

    “Drunk Russian Satellite Hits Fat American Satellite”

  15. Geoff Forden (History)

    I think I should include a link to an ACW post I made about 6 weeks ago that discusses how space is getting really crowded:

  16. Allen Thomson (History)

    A bit of speculation, but I notice that Cosmos 2251 was in a period of negative decay, perhaps due to the 14th-order resonance that affects the Tselina-2 ELINT satellites.

    Maybe the orbital models underlying the collision prediction programs don’t adequately this effect?

  17. Allen Thomson (History)

    > don’t adequately this effect?

    Make that “don’t adequately take this effect into account?”

  18. Muskrat

    From a story by Joel Achenbach just posted on the Washington Post Website:

    “Iridium, the Bethesda-based satellite phone company, had received a report that its satellite …would pass within 300 meters of the non-operational Cosmos. Instead, the Iridium suddenly went silent.

    ‘If you look at statistics, every single day there are close approaches between some satellites, and typically four or five satellites which are within about 300 yards’ of each other, [a NASA scientist] said.”

    So there are 4-5 passes per day where the margin of separation is about the same as for this collision? Yikes. I mean, either somebody made a math/radar error, or one of the satellites changed orbits. And if it happened once, it can happen again.

  19. Geoff Forden (History)

    I checked that number quoted in the Washington Post article (which was spotted by Muskrat) and it turns out to be false. I suspect that Nicolas Johnson is being misquoted or at least quoted out of context. Johnson is quoted as giving the impression that SATELLITES approach each other within “300 yards” four or five times each day. That is not true! What actually happens is that three or four satellites (including nonoperational satellites) each day are approached closer than 800 m by SOME PIECE OF CATALOGED DEBRIS each day. (This is calculated from the average flux cross sectional area Donald Kessler sent me about a year ago.) Space is filling up, but satellites are not whizzing past each other that often.

    Iridium should have taken action given the catastrophic consequences of massive satellites colliding.

  20. yousaf

    The WPost story no longer has the quote that Iridium was notified (?) — and the author corrected the other quote also.

  21. Wang Ting

    for thoese who are interested in historical conjunction events distribution. I upload a pdf file at

  22. Wang Ting

    I think the “satellite” in “Washington post”, includes nonoperational satellite.

    After Chinese Test collision risk in LEO is significant increased, 300 yeards for four and five satellites make sence.

    I also checked the conjunction history of Iridium, After FY event, every 13 days, there will be a object has distance smaller than 1 km, every 50 dyas, there will be a boject has distance smaller than 500 m. Consider there are 66 satellites in the Iridium constellation. They actually get such report every single day!

    According to TLE provided in space track website, the minimum distance of two satellites is roughly 800 m.

    The point is the accuracy of TLE data is not good enough for satellite maneuvor.

    NORAD has more accuracy data called SP, such data is used to support ISS, space shuttle. But I don’t know whether they provide that data to Iridium.

  23. Allen Thomson (History)

    Wang Ting said,

    > The point is the accuracy of TLE data is not good enough for satellite maneuvor.

    Yes, this might be getting to the heart of the matter. Maneuvers are expensive in terms of fuel, constellation maintenance, and, consequently money, satellite lifetimes and effectiveness.

    So why would an operator do a maneuver if it weren’t really, really necessary?

    Generally speaking, I’ve found that calling warning is a pretty futile exercise.

  24. Gridlock (History)

    “NORAD has more accuracy data called SP, such data is used to support ISS, space shuttle. But I don’t know whether they provide that data to Iridium.”

    More than likely, or if not then they should – Iridium isn’t exactly your average “commercial operation” (headquartered in MD – not surprising..) and there’s numerous dependencies within .gov that they wouldn’t want to put at risk..

  25. Geoff Forden (History)

    If you haven’t seen it yet, the New York Times has a very informative and insightful editorial about the collision.

    The NYT assertion that making such calculations is too time consuming is surprising but it appears to be true, at least if you use desk top PC’s. (And if things are done in a brute force sort of way.) 19,000 objects in space corresponds to 36 millions distance comparisons each time you take a step forward in the orbit propagations. If you did that in 60 second intervals (during which time average low Earth orbit objects move 440 km), then you would need to make 5×10^11 distance calculations each time you simulated a day in orbit. Your average intel Core i7 desk top processor would need about 7 million seconds to simulate a single day. A trillion instructions per second super computer, however, could quickly handle this problem even if done in the most stupid, brute force sort of way. Isn’t this problem worth spending a little supercomputer time on?

    But I see plenty of room for improvement by simply thinking about which distances need to be calculated! I still come down on the side that this disaster was avoidable and it should have been avoided.

  26. Wang Ting

    To forden,

    I have developed an algorithm for that. Usually, it takes only 10-13 minutes for a one day conjunctions prediction for all cataloged objects against itself, in my laptop.

    If you only want to know a specific satellite against the cataloged objects, it only takes several seconds.

    In the pdf I gave, I actually calualte all conjunctions from 1998-2008. I only use my laptop to do that calculation.

    Clearly, it is not a problem.

  27. Allen Thomson (History)

    > then you would need to make 5×10^11 distance calculations each time you simulated a day in orbit.

    There are many comparisons that don’t need to be made on the basis of orbitology. For example, for pairs where the perigee of one satellite is well above the apogee of the other.

    And there are creative ways to get to lotsaflop computing power. Graphics processors, it turns out, are built around very powerful FPUs:

  28. Cosmicray (History)

    Don’t you get the feeling that somebody is trying avoid responsibility for this thing? First, a NASA scientist says it happens so often its not even worth noticing? (Close flybys, I mean.) Then someone (probably NASA again) says that its too time consuming to calculate this sort of thing and nobody ever does it for more than 10 satellites. (Even though both Forden and Wang Ting say it can be done easily and cheaply.) Whats going on here?

  29. Muskrat

    In what my brother called the “coolest weather alert of all time,” the National Weather Service has warned of falling debris:

    Not to get all politically correct, but note that it says the Kosmos “crashed into” the (impliedly peaceful, innocent) Iridium.

  30. Geoff Forden (History)

    That “weather report” is well worth reproducing!

    NOUS43 KJKL 140445

    1145 PM EST FRI FEB 13 2009





    But they forgot the plague of locusts!

  31. Geoff Forden (History)

    Here is the Iranian briefing to COPUOS about their Omid satellite. There are some interesting details about the satellite.

  32. Warren Platts (History)

    Allen, you say Cosmos 2251 was in a state of “negative decay”. Does that mean it was gaining altitude? According to a Johnson Space Center official, Cosmos 2251 drifted down from a higher orbit.

    I’m not saying you’re wrong. But maybe you could expand on how you arrived at that conclusion.

  33. Allen Thomson (History)

    >Allen, you say Cosmos 2251 was in a state of “negative decay”. Does that mean it was gaining altitude?

    Temporarily, slightly, yes. The long-term trend was decreasing altitude under the influence of atmospheric drag, but there were periods when the presumed resonance effect pumped a bit of energy into the orbit. The days before the collision were such a period.

    > But maybe you could expand on how you arrived at that conclusion.

    Just by looking at the mean motion and decay terms in the two-line element sets for C2251 at Space Track. The mean motions were decreasing slightly and the decay term was preceded by a minus sign.

    I’d post a couple of TLEs to show what I mean, but the Space Track terms of use prohibit distribution.

    If you’re interested, you can go to, register, and download the TLEs to see what was going on.

    A particularly striking illustration of resonance is provided by the orbit of Cosmos 1833 (NORAD 17589) over the first decade or so of its life.

  34. Allen Thomson (History)

    On resonances messing up prediction accuracy:

    Note that I’m not claiming that this was what caused the Iridium 33/Cosmos 2251 splat, just raising it as a possibility.

  35. Warren Platts (History)

    Thanks for the info Allen. Do we know for sure that the Cosmos 2251 was incapable of maneuver? It’s rated lifetime was 24-36 months, yet it was only used for about a year before it was superceded by the Strela 3 constellation.

  36. Allen Thomson (History)

    > Thanks for the info Allen. Do we know for sure that the Cosmos 2251 was incapable of maneuver?

    I’ve never seen any indication of such capability, but will check with others who would probably know better.

  37. Allen Thomson (History)

    > > Thanks for the info Allen. Do we know for sure that the Cosmos 2251 was incapable of maneuver?

    > I’ve never seen any indication of such capability, but will check with others who would probably know better.

    OK, Those Who Know (aka the Great Old Ones) say “no propulsion.”