Iridium 33/Cosmos 2251 Collision

Coverage started 2009 March 5

Updated 2009 July 15

In the report issued on 2009 February 10 at 1502 UTC, SOCRATES predicted a close approach of 584 m between Iridium 33 and Cosmos 2251. This was not the top predicted close approach for that report or even the top predicted close approach for any of the Iridium satellites for the coming week. But, at the time of predicted close approach (1656 UTC), Iridium 33 suddenly went silent. The US Space Surveillance Network (SSN) subsequently reported that they were tracking debris clouds in both the Iridium 33 and Cosmos 2251 orbits, confirming a collision.

This is the first time two satellites are known to have collided on orbit. While Cosmos 2251, a Russian communications satellite, is thought to have ceased operations about two years after it was launched in 1993, Iridium 33 was part of the operational Iridium constellation of 66 satellites at the time of the collision.

As of today, the SSN has cataloged 382 pieces of debris (14 pieces of which have already decayed from orbit) associated with Iridium 33 and 893 pieces of debris (27 pieces of which have decayed) associated with Cosmos 2251. The materials below are provided to give a sense of the current relationship between the Iridium constellation and the resulting debris clouds.

Figure 1 shows the current Iridium constellation with the orbits for the operational satellites shown in green, the spares shown in blue, and the inactive satellites shown in red. The Iridium 33 debris is shown in yellow and the Cosmos 2251 debris is shown in orange. An AGI Viewer 9 file (see bottom of this page for more information on AGI Viewer 9) of the event is also available.

Click for larger image

Figure 1. Screen shot from AGI Viewer 9 file of current Iridium constellation and collision debris clouds

Download Current Iridium Constellation Scenario

Gabbard plots of the two debris clouds show the relative distribution of the debris in altitude and suggest that more of the Iridium 33 debris went to slightly higher orbits while much of the Cosmos 2251 debris went to lower orbits, with some of it already reentering the Earth's atmosphere.

Click for PDF file

Figure 2. Gabbard Plot of Iridium 33 Debris Cloud

Click for PDF file

Figure 3. Gabbard Plot of Cosmos 2251 Debris Cloud

Together with a 3D interactive plot of the relative velocities at the time of the collision (coming soon), these Gabbard plots should help develop a better understanding of the actual collision geometry.

Tracking a Collision

There has been much discussion about why this collision wasn't reported in SOCRATES or why Iridium didn't act on the information provided in SOCRATES. In reality, SOCRATES did predict a close approach between Iridium 33 and Cosmos 2251 at the time of the actual collision in each of the 14 reports in the week leading up to the event. None of these, however, made the Top Ten list. It is instructive to see just what SOCRATES reported to help understand the limitations of the data used by that system and what needs to be done to improve these types of screenings.

Starting with the report on February 4 at 0401 UTC and continuing right up to the report issued on February 10 at 1502 UTC, SOCRATES did predict a close approach between Iridium 33 and Cosmos 2251 at between 16:55:59.670 and 16:55:59.990 UTC. As can be seen in Figure 4, the predicted close approaches ranged from 117 m to 1.812 km over that interval.

The graph shows the predicted minimum range for the closest conjunction in each report, the closest Iridium conjunction, the closest Iridium 33 conjunction, and the conjunction with Cosmos 2251. The variability of the predictions over the week before the collision speaks to the inherent uncertainty in the TLE data used to make the predictions. While Report Number 5 (February 6 at 0402 UTC) did predict a close approach of 117 m, the predicted close approach grew to 1.243 km the very next day.

Click for PDF file

Figure 4. SOCRATES Min Range

In fact, as can be seen in Figure 4, the Cosmos 2251 conjunction is not even the closest predicted conjunction for Iridium 33 in most of the reports, much less for the entire Iridium constellation. Looking at the relative rankings of these conjunctions can help understand the limitations in using conjunction reports based on data as inaccurate as the TLEs.

Figures 5 and 6 show the rankings in each SOCRATES report for the Iridium 33/Cosmos 2251 conjunction in the total report, against all Iridium conjunctions, and for all Iridium 33 conjunctions. Over the 14 reports, the overall ranking ranges from a high of 1,611 in Report Number 3 to a low of 11—again just the very next day. At the time of the collision, it was ranked 152 overall.

Looking at the rankings against all Iridium conjunctions gives a sense of the magnitude of the problem for Iridium—or any satellite operator—in trying to screen with TLEs. The Iridium 33/Cosmos 2251 ranges from a high ranking of 149 in Report Number 3 to a low ranking of 2 in Report Number 5, with an average ranking of 64. Over this interval, Iridium 33 saw between 10 to 15 times each week when something was predicted to come within 5 km of it—out of the 1,007 to 1,095 such predicted conjunctions for the entire Iridium constellation. Determining that the predicted conjunction for Iridium 33 and Cosmos 2251 was more significant than the many dozens of other Iridium conjunctions for that week is simply not possible using the TLE data.

Click for PDF file

Figure 5. SOCRATES Rank (All)

Click for PDF file

Figure 6. SOCRATES Rank (Top 200)

So does that mean it isn't possible to provide meaningful conjunction screening for satellites operators? No. SOCRATES was actually set up to demonstrate several key points:

  • That it is possible, using off-the-shelf software, standard orbital data, and basic computer hardware to automatically screen large numbers of satellite (over this interval, we screened 2,946 payloads against between 11,761 and 11,848 total objects) and provide access to the reports via a standard web interface.
  • To help people understand that close approaches with satellites in Earth orbit are not uncommon events. Over the period examined here, between 11,169 and 11,493 conjunctions were reported each week—over 1,600 times a day. Even though each conjunction has a low probability of occurrence, over time, a collision with an operational satellite was bound to happen.
  • To illustrate the need for better data to reduce the uncertainty in conjunction predictions, thereby enabling satellite operators to work to protect their assets. Better data is available—directly from satellite operators and from the US and other governments. Our SOCRATES-GEO effort has shown the advantages of operators working together and sharing their best orbital data to reduce risks of collisions, but a more open data sharing policy and wider participation is needed to make this process even better.

Anatomy of a Collision

The following figures illustrate the geometry of the collision of Iridium 33 and Cosmos 2251.

Figure 7 shows the orbital geometry just prior to the collision at 16:55:59.806 UTC. In this figure, Iridium 33 is moving from the lower left to the upper right while Cosmos 2251 is moving from the upper left to the lower right. It can be seen that the objects collided at nearly right angles over northern Russia.

Click for larger image

Figure 7. View of Iridium 33 and Cosmos 2251 orbits just prior to collision

Figure 8 shows the initial evolution of the debris cloud just 10 minutes later. This figure uses the first TLEs published by AFSPC when each debris piece was added to the public catalog. Note that not all of the debris lines up well at the time of the collision due to limitations of the data. Some of the pieces were just cataloged and are being propagated backward several months.

Click for larger image

Figure 8. View of Iridium 33 and Cosmos 2251 orbits and debris 10 minutes post-collision

Figure 9 shows the evolution of the debris clouds 180 minutes post-collision, almost two revolutions later. The spread of each debris cloud around its respective orbit is already becoming apparent.

Click for larger image

Figure 9. View of Iridium 33 and Cosmos 2251 orbits and debris 180 minutes post-collision

Of course, the best way to see the evolution of these debris clouds is to view the interactive AGI Viewer file of the event, which is provided below.

Download Iridium 33/Cosmos 2251 Collision Scenario

Video captured by Kevin Fetter on 2009 March 12 shows the main piece still associated with Iridium 33 in the SATCAT (NORAD Catalog Number 24946) followed by the inactive Iridium 28 satellite. The double flashes from Iridium 33 suggest that at least two of the MMAs (Main Mission Antennas, seen at the bottom of the satellite in Figure 10) on that object survived the collision relatively intact.

Click for larger image

Figure 10. View of Iridium satellite showing MMAs (bottom)

The video was taken from Brockville, Ontario, Canada (44.6062 N, 75.6910 W) looking at 8h 45m RA, +04° 56' Dec between 00:54:03 and 00:54:26 UTC. The screen shot seen in Figure 11 verifies the basic conditions of this observation, including that the necessary visibility conditions were satisfied.

Click for larger image

Figure 11. Snapshot of Kevin Fetter's observation conditions as seen in video

Note: Larger versions of all the images provided on this page are available by clicking the images. The interactive AGI Viewer files of these scenes are also provided to give you a far better sense of the overall environment by allowing you to zoom in and out and move around the Earth while watching all the satellites moving in their orbits.

In the coming days, I will be providing additional analysis showing:

  • The relative velocity spread of the initial debris

Note: AGI Viewer is a free product which allows anyone with a Windows computer to view an STK (Satellite Tool Kit) scenario. With it, you can animate a scenario forward or backward, pause the animation, and zoom or pan the view for a more complete understanding of the event. Just like with Adobe Acrobat, where the authoring software requires a license but the Adobe Reader is free, STK can produce AGI Viewer files—also known as VDFs—which can then be viewed by anyone with the AGI Viewer software. You can find the free AGI Viewer 9 on the AGI web site at (download here if you are experiencing problems with the AGI site).

Note: My ongoing support to Iridium and the Joint Space Operations Center on this event since February 10 has necessarily taken priority over some of my additional duties operating CelesTrak and has subsequently delayed the completion of this report. As the situation continues to evolve and more data becomes available to the public, I will continue to update the information provided here, along with providing additional analysis. —T.S.

TLE Data Space Data
Current GPS
Archives EOP
Documentation Space Weather
SATCAT Columns
Boxscore Software
Dr. T.S. Kelso []
Follow CelesTrak on Twitter @TSKelso
Last updated: 2013 September 25 23:41:30 UTC
Accessed 30,728 times since 2000 December 16
Current system time: 2018 July 16 14:12:15 UTC