Printable Version of Topic

Posted by: ElkGroveDan Feb 11 2009, 09:35 PM

Two satellites collide in orbit
BY WILLIAM HARWOOD

Posted: February 11, 2009

In an unprecedented space collision, a commercial Iridium communications satellite and a presumably defunct Russian Cosmos satellite ran into each other Tuesday above northern Siberia, creating a cloud of wreckage, officials said today.

Iridium satellite
An artist's concept of an Iridium satellite orbiting the Earth. Photo: Iridium

The international space station does not appear to be threatened by the debris, they said, but it's not yet clear whether it poses a risk to any other military or civilian satellites.

"They collided at an altitude of 790 kilometers (491 miles) over northern Siberia Tuesday about noon Washington time," said Nicholas Johnson, NASA's chief scientist for orbital debris at the Johnson Space Center in Houston. "The U.S. space surveillance network detected a large number of debris from both objects."

MORE.....


http://spaceflightnow.com/news/n0902/11iridium/

Posted by: OWW Feb 11 2009, 09:58 PM

http://www.spaceflightnow.com/news/n0902/11iridium/

Ouch. How long before LEO becomes a dangerous thick cloud of debris?

Posted by: nprev Feb 11 2009, 11:25 PM

Holy crap!

Man, I knew it was getting crowded, but still amazed that it happened; the odds have to be pretty long. Article quoted Johnson from NASA as saying that there's no 'ATC' in space; maybe it's time to start thinking about establishing one.

Posted by: Zvezdichko Feb 12 2009, 09:12 AM

This is bad. I hope it's not too late...

Posted by: ugordan Feb 12 2009, 09:25 AM

What puzzles me is how come this collision wasn't seen coming. Usually the threatened satellite performs a collision avoidance maneuver.

Posted by: nprev Feb 12 2009, 10:59 AM

I wondered that too, G. The Cosmos seemed to be dead already, but the Iridium's condition wasn't stated. All I can think is that the Russians weren't keeping tabs on the Cosmos anymore, and maybe the Iridium wasn't being tracked because it's a private spacecraft...?

There's obviously some sort of disconnect here, but not sure what it is.

EDIT: It seems that the Iridium was in fact still active, so this remains puzzling.

Posted by: rlorenz Feb 12 2009, 11:27 AM

Since this was a defunct satellite that hit the Iridium, it may not be quite the first
event of its type - a catalogued object (Ariane fragment) impacted the Cerise
microsatellite in 1996, severing the gravity gradient boom.

Posted by: nprev Feb 12 2009, 12:17 PM

True, Ralph. I think that the 'first' here is the collision of two previously intact spacecraft, though. Seems as if the LEO environment is getting crowded enough to justify some sort of regulation in addition to intensified tracking efforts.

People have been saying this for years, of course, but it's probably time to do something about it, esp. considering the fact that there are an increasing number of nations and private concerns beginning to launch stuff in addition to the existing players. Somebody's billion-dollar baby is gonna get whacked at some point, and whoever it is will doubtless scream bloody murder; best course of action is to start formulating policies & procedures to mitigate the risk.

Posted by: Geert Feb 12 2009, 01:35 PM

Do the Iridium satellites have own maneuvering engines? The space shuttle and ISS have several times dodged potential close encounter situations, but other craft simply do not have the capacity for this (Hubble for instance...).

What surprises me is that this happened at 800 km, as far as I know there is a lot more traffic further down (ISS altitude) and/or higher up (stationary orbits). Also, what kind of orbit was the other satellite in, in other words how fast did they hit?

Worst case is a kind of chain-reaction, this has resulted in a large cloud of debris, which will slowly spiral down through all the other crowded orbits below, and it might hit other satellites in turn. Big satellites like Hubble, which can't steer themselves out of the way, are the most vulnerable, hope it stays clear!

Regards,

Geert.

Posted by: ugordan Feb 12 2009, 01:39 PM

I believe so, given that one of the 8 "spare" satellites will now be moved into Iridium 33's position.


Roughly polar orbits. They hit at almost a 90 degree angle and that comes out to something like 11 km/s relative velocity. Ouch.

http://spaceweather.com/swpod2009/12feb09/deak1.gif

Posted by: AndyG Feb 12 2009, 01:52 PM

There's a trillion cubic kilometres of space in LEO, up to 2000km. Less than ten thousand objects above 10cm across.

A miss is as good as a mile in vacuum: I'm utterly stunned that this could occur. Paint flakes and the like, impacting in lower orbits, yes - but to quote http://en.wikipedia.org/wiki/Harry_Hill...

"What are the chances of that happening?"

Posted by: dvandorn Feb 12 2009, 02:08 PM

I don't know, it seems to me that, with the number of objects in orbit, this kind of thing was bound to happen sometime. I'm a little surprised that the event wasn't predicted, since there are several agencies across the globe that skin-track everything in orbit. (Where do you think those predictions come from that result in all those collision-avoidance maneuvers?)

-the other Doug

Posted by: algorimancer Feb 12 2009, 03:19 PM

Yes, considering that they (apparently) routinely track fragments the size of bolts and coins (at least in LEO), I find this a tad suspicious. Just how "dead" was that Russian satellite? Not to be going off on a conspiracy rant or anything, just finding my belief a little strained here -- particularly in light of the recent Chinese and US ASAT operations.

Posted by: tasp Feb 12 2009, 03:40 PM

Not to make light of this occurrence, but tracking of the resulting debris' orbital evolution could be interesting and relevant to several phenomena in the solar system. Laplace might find this to be a fascinating (serendipituous) experiment . . .

Posted by: ngunn Feb 12 2009, 04:35 PM

And now at last the truth about that equatorial ridge - it's the remains of all the satellites the Iapetan's had in orbit before the first accidental collision started the Laplace cascade.

Posted by: stevesliva Feb 12 2009, 05:40 PM

I'm curious, too. But if you expand your timescale to infinity, perhaps the chance is pretty close to one. What are the chances in any given year? Any given decade? And given the choice of orbits, is the space being considered really so large?

Posted by: stevesliva Feb 12 2009, 05:43 PM

If it was less than dead, you'd hear about its divergence from its expected orbit.

While it's Iridium's fault, I'm surprised there was no prior warning. I would have thought the US trackers would anticipate any two objects--dead or alive--occupying the same space.

Posted by: Geert Feb 12 2009, 05:43 PM

That picture says it all, the more you think about it the more perplexed you are that this could happen, such a tremendously small chance...

Two satellites, each over 1 ton in weight, hitting at 11 km/sec, must have been quite some fireworks in the Siberian sky, some of the fragments might even have made it into solar orbit... Shape and progression of this debris cloud will indeed be an interesting mathematical experiment...

Regards,

Geert.

Posted by: ugordan Feb 12 2009, 05:48 PM

How exactly is this Iridium's fault?

As for advanced warning, I'm puzzled as well.

Posted by: djellison Feb 12 2009, 05:50 PM

At 7.5km/sec - it takes about 0.00013s for a spacecraft to cover the, say, 1 metre width of an Iridium sat.

I wonder if anyone was watching it do an iridium flare at the time


Doug

Posted by: stevesliva Feb 12 2009, 05:50 PM

Same way it's the skipper's fault if a boat runs aground on a charted shoal.

Posted by: ugordan Feb 12 2009, 05:56 PM

Bad analogy. Iridium doesn't have assets to track every conceivable, potentially harmful object in the sky. Other organizations specialized in that do. It's very likely these other organizations already know Iridium 33 is/was a functioning satellite and as such would benefit from collision predictions. Or are Iridium guys expected to go to those same organizations every day and ask "I know we asked this yesterday, but could you see if there's any possibility of a collision today?"

The skipper is blind and a third person needs to inform him if/when he gets into very shallow waters.

Posted by: dvandorn Feb 12 2009, 06:06 PM

The only way in which I think you could say that the collision was Iridium's fault is in the fact that, of the two satellites, Iridium was the only one that was still "live" and had any capacity for collision avoidance.

However, just because Iridium was capable of maneuvering doesn't mean that its controllers were aware of the collision threat. I think if there is any "blame" to lay here, it's with the agencies that track the satellites, who could have sounded a warning and given Iridium's controllers the opportunity to make a collision avoidance maneuver.

-the other Doug

Posted by: mcaplinger Feb 12 2009, 06:38 PM

Wasn't the Cosmos stage in the published NORAD two-line element sets? I'd say that Iridium should have been looking at those.

EDIT: sure it was: COSMOS 2251, NORAD ID 22675.

Posted by: djellison Feb 12 2009, 06:45 PM

They should have been - as MC cites, the details of the defunct satellite were available, along with most other chunks up there.

Not keeping track of your satellites and comparing them to the published NOARD elements, is like sailing along a coast without charts.

Posted by: ngunn Feb 12 2009, 06:59 PM

A question - could a defunct satellite suddenly spring a leak and thereby start to deviate from it's predicted path?

Posted by: stevesliva Feb 12 2009, 07:29 PM

Or bump into something smaller first...

Posted by: mcaplinger Feb 12 2009, 08:20 PM

One has to keep in mind that satellite paths are not predictable for more than a couple of weeks under any circumstance due to non-gravitational effects like drag, light pressure, etc, especially for LEO orbits. That's why NORAD regularly updates their element sets.

Unless there's reason to think that Cosmos 2251 suddenly changed course or that the data in the NORAD sets was wrong, I'd place the blame for this totally on Iridium. If they weren't looking for collisions, it was basically a hope that they would stay lucky. With 90+ satellites in LEO, they were really in a position to pay more attention.

Posted by: nprev Feb 12 2009, 08:33 PM

Mike, you're right; even the GPS constellation (which is in MEO) downloads ephemeris updates to ground receivers fairly often.

Seems like the question is who's responsible for predicting such collisions? From the discussion thus far, I suspect that the US & Russia watch the ISS like a hawk, of course, and probably also all their government-owned assets (at least those that are active). Could it be that Iridium & other private spacecraft are expected to do this on their own with nothing provided to them but updated ephemerides? If so, this seems unwise & definitely not in the best interests of all stakeholders for the sole reason that nobody wants to deal with a debris cloud.

Posted by: AndyG Feb 12 2009, 08:48 PM

I need to delve into the statistical probabilities, but given the apogee/perigee variation for the satellites was ~ 26km for Cosmos 2251 and ~ 15km for Iridium 33, that there was complete overlap, and that the inclinations were similar (74 versus 86.4 degrees) that's still a huge amount of sky to exactly meet one-on-one in that tenth of a microsecond that counted.

It's beyond lottery-winning "unlikely".

...But maybe it'll raise a few eyebrows and result in positive decisions for the future.

Andy

Posted by: AndyG Feb 12 2009, 08:51 PM

A decade is infinitesimally smaller than infinity. Which would suggest the chance is 0.

As to "who's to blame", the answer, as Mr Newton would have to say, is "both of them".

Andy

Posted by: mcaplinger Feb 12 2009, 08:53 PM

That is certainly the case now, and frankly, it wouldn't have been that big a deal for Iridium to have done this. The argument could be made that it shouldn't be up to them, but absolving satellite operators of any responsibility for collision avoidance also seems like the wrong way to go.

Posted by: nprev Feb 12 2009, 09:02 PM

Thanks, Mike. Wow. Think we might have found the disconnect.

Liability issues aside, spacecraft collisions present hazards to everything operating in orbit; it's a lot more than a two-party issue. IMHO, it looks like there really does need to be an orbital 'ATC' of some sort, and this doesn't have to be more than a continuously updated model; maybe 3 or 4 software engineers keeping it happy, plus a few more people to broadcast alerts.

Worth doing, and maybe a good small business idea for any of you computationally gifted types out there. The root data's provided for free...

Posted by: imipak Feb 12 2009, 09:04 PM

What are the chances that a deliberate attempt to set up a collision between orbits with those orbital characteristics would succeed? Statistics and frequency probability are deep and frequently counter-intuitive waters.

Posted by: stevesliva Feb 12 2009, 09:30 PM

Tenth of a millisecond 5000 tries per year... but I getcha. Not only do the orbits have to exactly intersect, but they have to be at the point of intersection at the same time.

Making the wrong assumption that two orbits will always intersect at exactly one place, with a 100 minute orbit, the chances of being at that particular place are 6000/.00013= 1 in 46 million per orbit? And then say it does 50000 orbits/decade-- a 0.108% chance of it happening in a decade. (Times 70 satellites)

No idea what the chances are of two orbits intersecting, though. And I know they probably don't, for long. Brings the chances way down. So if they intersect once per day, divide by 3650, for a .0000296% chance, about 1 in 3.4 million.

That does get you up into winning-the-lottery range.

Posted by: OWW Feb 12 2009, 09:31 PM

Speculation: Maybe Cosmos 2251 very recently suffered an (undetected) fragmentation and the debris cloud intersected with Iridium's orbit.

Posted by: AndyG Feb 12 2009, 10:02 PM

It helps the odds immensely. And a tiny lumplet, at 11km/s, would do the job.

Andy

Posted by: nprev Feb 12 2009, 10:19 PM

Thing is, all the press reports seem to imply a head-on collision. Were there any hints that the Cosmos had a pre-existing associated debris cloud?

Lots we don't know; a very odd event.

Posted by: ngunn Feb 12 2009, 11:08 PM

A very good point. One would expect the first event to lead to others. Would we necessarily detect the first one, or would we catch on a little later in the process?

I'm not so sanguine as some of the the statistical assessments above. I think we have a serious problem here.

Posted by: djellison Feb 12 2009, 11:12 PM

For any one satellite on any one orbit - yes.

But 3000 satellites, orbiting 14 times a day, 365 days a year, for, say, 10 years - and infact, it's not beyond the realms of possible in any way shape or form.

Doug

Posted by: dvandorn Feb 13 2009, 02:10 AM

Statistics are what people play around with while awaiting actual empirical data.

We now have empirical data. Put enough satellites and associated pieces of crap in orbit and, eventually, things start colliding. That's not a statistical analysis -- it's the empirical description of an observed phenomenon.

Also remember that, for everything that actually occurs, the statistical likelihood of it having happened is exactly 1 in 1. And as for the "lottery-winning" odds, please keep in mind that, at least in the United States, lotteries with odds of a single given person winning that work out to something like 100 million to one are won by *someone* every few weeks. So, while the odds that one given satellite may impact another given satellite may be very, very low, the odds that *some* satellite out there will collide with some *other* satellite are obviously a lot higher.

-the other Doug

Posted by: kwan3217 Feb 13 2009, 04:58 AM

From the http://www.satobs.org/seesat/Feb-2009/index.html SeeSat-L list: Closest predicted approach of the two objects was 800m plus or minus whatever the error in a two-line element is. There are six predicted close approaches within 100m in the next five days. Closest one is something like 52m. http://celestrak.com/SOCRATES/top10minrange.asp . Probably none of them will collide. One of the guys on SeeSat says that these two didn't even make the top 10 hazard list.

You pays your money and you takes your chance. Just because something bad happened doesn't mean that someone is at fault. If you are running Iridium, maybe you have a conjunction like this once every couple of days with one of your satellites. You have to maneuver to dodge, and maneuver again to get back into the proper slot. Do that enough and you will run out of gas a lot faster than just doing nothing.

And, I highly recommend SeeSat. It is for Earth orbiters what UMSF is for the rovers.

Posted by: ilbasso Feb 13 2009, 04:59 AM

They have moved the ISS orbit - what, 8 times? - because of a satellite coming within a few miles of the predicted location. That's a pretty large area of uncertainty. I'm trying to imagine a scenario where they require every active satellite that might pass within a few miles of another piece of debris change its orbit. How many satellites would be changing their orbits every year, and how badly would that complicate the job of those who are trying to keep track of them all? Is it worth the hassle to do all that moving for an incredibly low probability that any one of those "near misses" might be a collision? Yes, for a manned vehicle. It depends, for an unmanned one - depends on criticality of function, etc etc.

Posted by: PFK Feb 13 2009, 07:52 AM

Large Hadron Collider, who needs it?!

Posted by: nprev Feb 13 2009, 11:40 AM

The only thing to be sure of is that this event is gonna increase costs for private LEO flights. It's a pretty safe bet that the insurers are going to factor this in for all future policies.

Posted by: ngunn Feb 13 2009, 11:58 AM

I don't get this (from spaceweather):

"This injection of debris substantially increases the population of space junk at altitudes near 800 km. Collisions are now more likely than ever. Fortunately, the International Space Station orbits Earth at a much lower altitude, 350 km, so it is in no immediate danger. The Hubble Space Telescope is not so safe at 610 km. In the days ahead, researchers will carefully study the make-up and dynamics of the debris cloud to estimate when bits will begin to drift down to lower altitudes."

If you collide two (equal) objects at 90 degrees you'd kill one-over-root-2 of the velocity right there. Far from expecting the debris to spread out at the same altitude and only drift down slowly I'd expect a lot of it to fall down pretty quickly.

Posted by: ugordan Feb 13 2009, 12:03 PM

You're assuming a totally inelastic collision. What appears to have happened is that two distinct debris clouds were created that roughly follow original objects' orbits.

Posted by: tedstryk Feb 13 2009, 12:53 PM

That math isn't really fitting, since satellites are not distributed randomly. Based on the launch site and purpose, certain types of orbits are used far more than others.

Posted by: ngunn Feb 13 2009, 02:34 PM

I know - a VERY crude approximation just to make the general point.

Kinetic energy was lost (a lot of it since both satellites were completely destroyed) so the mean velocity of the fragments must be less than it was before the collision. Most will therefore be moving too slowly to remain in circular orbit and so must drop down to a lower perigee at the antipodal point.

Posted by: remcook Feb 13 2009, 03:54 PM

And that's the reason why they expect most bits to burn in the atmosphere relatively soon. Yet the apogee stays at the same altitude in that case. And some bits will probably gain a bit of momentum (for instance, if stuff like fuel explodes or just being ejected in the right direction) and move a bit in one direction and/or their orbital plane will change a bit and their orbital periods will be slightly different, so things will spread out eventually. The amount of energy lost during the collision is probably not so high as you imagine relative to the enormous amount of energy they had, since the material of the structures must not have given much resistance to such an impact.

Posted by: dvandorn Feb 13 2009, 05:38 PM

The dynamics of the impact also affect the kinetic energy transfer. Remember, you don't just "lose" energy, it's translated into other forms of energy, other vectors, etc.

If the two bodies hit pretty much directly, then much of the "lost" orbital energy would be used up in melting, shattering and vaporizing the structures of the bodies. If the collision path would have only a half or a quarter of one of the bodies intersecting the other, less kinetic energy is used up in vaporization and the resulting debris' orbits are altered less than in a direct impact. If only a very small physical interaction occurs (i.e., if an antenna on one vehicle snags an antenna on the other vehicle), the two bodies will still break up (especially if the "just barely" contact comes at 11 km/sec!), but they'll break up due to the extreme rotational rates imparted by the "brush-by". This last scenario changes the orbits of the resulting debris the least.

I don't know if our "security assets" are able to image satellites at 800km range well enough to characterize the resulting debris from the Cosmos-Iridium collision, but the extent of debris and its orbital characteristics (as tracked by radar) ought to give us some idea as to how direct a hit they endured.

-the other Doug

Posted by: mcaplinger Feb 13 2009, 05:41 PM

As noted by kwan3217, there is already a very nice service to look at collision probabilities: http://celestrak.com/SOCRATES

If you look at today's top 10, the highest probability is 2.785E-03 (1 in 359). This is certainly a lot higher than I would have expected. I wonder if Iridium was using this service, and if so how, and if they were concerned at all before the collision.

Posted by: ugordan Feb 13 2009, 10:53 PM

Via NSF.com - a couple of simulations showing debris evolution: http://www.agi.com/corporate/mediaCenter/news/iridium-cosmos/

Posted by: Geert Feb 14 2009, 02:42 AM

That is indeed a extremely interesting service, and I'm equally amazed at the results, didn't know also there were regularly so many encounters at such close ranges!

If the Iridium collision didn't even make it into the top 10 I'm not surprised that no prior avoidance action was undertaken, as has been mentioned already if you start maneuvering for each of these close encounters you will run out of fuel quite soon, and indeed it causes problems with your commercial service as satellites will be drifting out of their slots, etc, etc.

Still, now the question opens, if the Iridium collision did not make it into the top 10, then how accurate where/are the details of the orbits? Did either of the satellites recently manoeuvre ( I know the Cosmos satellite was dead, but it could have been losing fuel or some pressurized gas, causing a slight propulsion), how accurate is the tracking really and how often are those parameters updated?

At such speeds and with almost 90 degree intersection, you need only a very, very, minor tracking error and the situation completely changes...

Posted by: Vultur Feb 14 2009, 02:47 AM

It will be interesting to see how (or if) regulatory bodies will respond. (For that matter, what regulatory body would have jurisdiction? The US government could make laws requiring the US companies to take precautions, but they have no authority over non-US companies, or the space agencies of other nations. It seems unlikely that the UN will respond, but you never know...)

Posted by: nprev Feb 14 2009, 04:22 AM

Well, the right answer IMHO would be at this point to add it to the duties of the International Civil Aviation Organization (ICAO); they're sort of like an international version of the FAA. Any sort of regulation's gotta be multinational, obviously.

Posted by: ngunn Feb 14 2009, 01:41 PM

From SpaceWeather:

"LISTEN UP: The US Air Force Space Surveillance Radar is monitoring the skies above Texas for echoes from satellite fragments. Try listening on Saturday, Feb. 14th between 10:45 and 10:55 am CST (1645 - 1655 UT). That's when Iridium 33 would have passed over the radar intact had it not been shattered."

On the spaceweather site 'Try listening' is an active link (I think).

Posted by: Leither Feb 14 2009, 08:23 PM

Well we might not have to wait too long……in July 08 the UN General Assembly recommendation that the Scientific and Technical Subcommittee of the Committee on the Peaceful Uses of Outer Space should consider space debris as an agenda item at it next session. The next session (the 46th) started on 9th Feb 09 in Vienna and according to the agenda (published in Dec 08) it’s due to discuss space debris this Monday/Tuesday (16 -17 Feb 09). What timing…no guesses as to what they will be talking about! It will probably be a closed session but the daily report might be interesting.

The GA also ”.. noted with appreciation that some States were already implementing space debris mitigation measures on a voluntary basis, through national mechanisms and consistent with the Inter-Agency Space Debris Coordination Committee Guidelines and the Space Debris Mitigation Guidelines …..(and) invited other Member States to implement, through relevant national mechanisms, the Space Debris Mitigation Guidelines.” Space Debris Mitigation Guidelines cover both Prevention of on-orbit collisions and Post-mission disposal.

Posted by: leper Feb 15 2009, 12:30 AM

I'm surprised this happened in my lifetime! The future is now...
Does anyone know if de-orbit tethers have been / will be added to any existing/future payloads?

I wonder how many more collisions like this there can be before the debris situation becomes unmanageable....

Posted by: Juramike Feb 16 2009, 02:11 AM

CNN reporting that a fireball was imaged and sonic booms were heard over Texas: http://www.cnn.com/2009/US/02/15/texas.sky.debris/index.html

Posted by: nprev Feb 16 2009, 02:46 AM

Hmm. The speed of the object is fairly slow, which is consistent for a debris re-entry, but it could still be a natural meteor. Either way, if I was in central/eastern Texas right now I'd start lookin' for chunks!

Posted by: PhilCo126 Feb 16 2009, 07:29 AM

Another weblink with the video footage:
http://www.news8austin.com/content/top_stories/default.asp?ArID=232081

Posted by: Exploitcorporations Feb 17 2009, 03:44 AM

I've been following this story with astonishment and dismay. Such infinitesimal odds, and so much potential for harm! Just a few weeks ago I was watching WALL-E with my son and explaining why I thought it was funny when they ploughed through the mass of satellites when leaving Earth. Doesn't seem so funny now. I stumbled across this lovely image while working on another project today, and it seems appropriate.

Posted by: mcaplinger Feb 23 2009, 02:48 PM

http://www.thespacereview.com/article/1314/1 is a very good summary of this event.

Posted by: stevesliva Feb 23 2009, 04:42 PM

My bold. Ha. Given that supercomputers and salaries are an order of magnitude cheaper than satellites, perhaps that will change. (Although the Iridium constellation is odd in that is was essentially gotten for free after bankruptcy.) The conclusions in the article are a little more nuanced. Good article!

Posted by: helvick Feb 23 2009, 06:12 PM

Nice post on the collision by Diandra over at http://twistedphysics.typepad.com/cocktail_party_physics/2009/02/when-satellites-collide.html.

She mentions two things that I hadn't come across before - firstly that the predicted closest approach of the two satellites prior to the impact had been around 600m and secondly that the increased atmospheric drag caused by the recent Sunspot minimum is a prime suspect in explaining why the predicted orbits were sufficiently incorrect to turn that 600m into 0.

600m would still seem like far too close for comfort to me but does anyone know what the normal error in such things would have been expected to be?

Posted by: Tom Womack Feb 24 2009, 01:39 PM

http://celestrak.com/SOCRATES/top10maxprob.asp almost always shows at least one expected pass at 0.1km or less, with an estimated probability of collision more than one in a thousand; since we don't see a collision every few years, I suspect their model is not perfect.

From a 500km circular orbit, a change in velocity by 1 centimetre per second changes the radius of the orbit by 18 metres, so you have to know the velocity very accurately indeed.