RPO Galore: the complex proximity operations of Kosmos 2581, Kosmos 2582, Kosmos 2583 and 'Object F'

click image to enlarge

Some 590 km above our heads, a complex series of multiple Rendezvous and Proximity Manoevers (RPO) by multiple Russian military spacecraft has been going on the past months, weeks and days.

On 5 February 2025, Russia launched three satellites from Plesetsk, into a 82 degree inclined orbit: Kosmos 2581, Kosmos 2582 and Kosmos 2583 (2025-026A, B and C). Kosmos 2581 was initially placed in a 595 x 578 km orbit; Kosmos 2582 in a 597 x 576 km orbit; and Kosmos 2583 in a 597 x 580 km orbit.

The diagram below sums up what has unfolded over the past few months: a complex series of manoeuvers and approaches to each other.

Click diagram to enlarge

 

About two weeks after the launch, both Kosmos 2581 and Kosmos 2582 were manoeuvered to a slightly higher orbit: Kosmos 2582 did so on 15 February and Kosmos 2581 on 19 February. 

On 24 February, Kosmos 2582 manoeuvered again, this time slightly lowering its orbit to the same orbital altitude as Kosmos 2581, for the initiation of first of a long series of complex proximity manoeuvers. They made a very close approach (1 km or less) on 24 February around 21:44 UTC. They stayed in each others close vicinity for the next three months, manoeuvering to and fro, with Kosmos 2582 doing the manoeuvering and Kosmos 2581 acting as the target..

Meanwhile, on 18 March 2025, the third satellite from the launch, Kosmos 2583, surprised by releasing a fourth object, 'Object F' (2025-026F). This object appears to be passive (i.e. it does not appear to be manoeuvering), but starting in April, Kosmos 2583 has subsequently been doing RPO manoeuvers with it, manoeuvering to and from it repeatedly, untill late May, so perhaps it is not merely debris. 

On 26 May, Kosmos 2583 manoeuvered away from 'Object F' to a higher orbit. It subsequently made a series of manoeuvers that made it closely approach Kosmos 2581 to a few hundred meters or less on June 10 near 9:42 UTC. It next stayed in the close vicinity of Kosmos 2581, approaching it again very closely in the days around June 19.

While Kosmos 2582 meanwhile had stayed in the wider vicinity of the other two, following them at a  distance of several hundreds of kilometers, it manoeuvered on June 23 in order to get close, within kilometers, to the by then close pair of Kosmos 2581 and Kosmos 2583. As a result, all three objects were now orbiting closely together.

 

click image to enlarge

The framestack above is from video footage I shot from Leiden, the Netherlands a few nights later during a near-zenith pass in twilight on 28 June 2025, when the three objects were still in close vicinity, albeit less close than the days before (during the days before, they were so close that I generally could not resolve them). Kosmos 2581, 2582 and 2583 are all visible in the framestack, the latter however being very faint.  

Kosmos 2581 and Kosmos 2582 were at about 1.2 km from each other at the time of observation. Kosmos 2583 was at about 3.6 km from Kosmos 2581.

Below is the actual video footage the framestack was derived from: the two brighter objects are Kosmos 2581 and 2582 (Kosmos 2581 leading), and just in front of them is Kosmos 2583, barely visible (look closely at full screen during the first 4 seconds of visibility). It was shot from my home in Leiden, the Netherlands, with a WATEC 902H2 Supreme and Samyang 1.2/85 mm lens.

'Object F' meanwhile, has drifted to quite a distance from the other three objects. Below is a framestack showing it, from video observations obtained on 17 June 2025: it was faint and might show some brightness variation.

Click image to enlarge

 

This is one of the most complex RPO exercises Russia has been conducting in LEO in decades. The RPO appears to be ongoing, and it will be interesting to follow it: for example, will we see that at one point 'Object F' is becomes the subject of an RPO again or not?

 

Post scriptum: more on another recent event with another Russian satellite, Kosmos 2558, releasing an object in another blogpost from today. 

Kosmos 2558 released an object on orbit on June 26 [UPDATED]

Object C (click image to enlarge)

 

Kosmos 2558 (2022-089A, cat nr. 53323) is a Russian military satellite that was launched from Plesetsk three years ago, on 1 August 2022. It was launched into the orbital plane of an American ADVANCED CRYSTAL spy satellite, USA 326. It is therefore believed to be a 'Nivelir' type 'inspection satellite' (see also this earlier discussion of several of such missions, and Bart Hendrickx' Space Review article on the Nivelir program  here).

As I recently wrote, there have now been four of these missions in a five-year timespan. Apart from Kosmos 2558 discussed here, these are: Kosmos 2542/2543 launched in the orbital plane of USA 245 in 2020;  Kosmos 2576 placed in the orbital plane of USA 314 in 2024 (see this earlier blogpost); and very recently, Kosmos 2588 placed in the orbital plane of USA 338 (see this earlier blogpost). 

With the exception of Kosmos 2542/2543, which have been deorbitted, all three remaining missions are still dedicatedly shadowing their American targets, keeping their orbital altitude and inclination difference such that the rate of RAAN (node) precession matches that of their target. This ensures that the orbital planes do not drift apart.

While mostly seen as 'inspection missions', I have repeatedly voiced a concern that these missions might be the positioning of 'dormant' on-orbit Anti-Satellite (ASAT) weapons. To me, the long, ongoing shadowing of what are some of the most prized US military space assets, their KH-11 Advanced Enhanced Crystal high-resolution optical IMINT satellites, is odd for 'just' an inspection mission. What is there still left to inspect after 3+ years? The satellites make only periodic close approaches (every few days) so monitoring of attitude (pointing) or radio signals is not an explanation.

Understandibly, the 'stalking' behaviour of these Kosmos satellites on their satellites makes the US military nervous. And now they have yet another reason to get even more nervous: on 26 June 2025 near 12:03 UTC (according to my analysis), Kosmos 2558 released  a second object, currently designated as 'Object C' (2022-089C, cat nr. 64627). 

The release is actually visible in the orbital elements for kosmos 2558, as a sudden slight lowering of the orbit:
 

Click diagram to enlarge

 

This is the third time we see this kind of 'Matrushka Doll' behaviour from Russian military satellites in five years time. 

Shortly after reaching orbit in 2020, Kosmos 2542 released a sub-satellite too, Kosmos 2543, which next in turn fired what is interpreted as a projectile, 'object E' (the latter was widely seen as an ASAT technology test at the time). 

In addition, more recently, Kosmos 2583 released an object, 'Object F' (2025-026F), on 18 March this year, as part of a complex series of RPO's (Rendezvous- and Proximity Operations) between Kosmos 2581, Kosmos 2582, Kosmos 2583 and said 'Object F'. I will publish a separate blogpost on these later [edit: separate post now available here].

In 2020 and more recently with Kosmos 2583 and 'Object F', this spawning of other objects happened within weeks after the launch of the parent satellite. But this time with Kosmos 2558 and 'Object C', the release of a second object happened almost three years after launch.

I observed Kosmos 2558 and the newly released 'Object C' last night. 'Object C' reached magnitude +8. The images below are 1-second frame stacks from video observations I made from Leiden, the Netherlands, on the night of 28-29 June 2025, two-and-a-half days after 'Object C' was released. At the time of observation, 'Object C' was some 143 km distant from Kosmos 2558, passing the camera FOV some 16-seconds after it. The first framestack shows 'Object C'; the second shows Kosmos 2558.

 

Object C (click image to enlarge)

 

Kosmos 2558 (click image to enlarge)

Below is the video footage of the pass in question. The first to pass through the FOV is Kosmos 2558 as a bright object: then, some 16 seconds later, 'Object C' passes through the FOV as a fainter object. The footage was shot from Leiden, the Netherlands, with a WATEC 902H2 Supreme camera and Samyang 1.2/85 mm lens.

 

 

It will be interesting to see whether any manoeuvering between the two objects is happening over the coming weeks.

 

UPDATE 4 July 2025:

Since release, Object C has been manouevering with respect to Kosmos 2558. While initially manouevering back to Kosmos 2558, it lowered its orbit significantly (by 15 km) on July 3 near 18:42 UTC. See diagram below.

Click diagram to enlarge

The reentry of the Chinese Zhuque-2E upper stage 2025-103G over Kazachstan on 5 June 2025

still from one of the fireball movies posted on Twitter by @Buggy__Bugler

 

In the evening of 5 June 2025 around 22h local time (17h UTC), a slow profusely fragmenting fireball was seen and filmed from several cities in Kazachstan, including Astana, and Bishkek in neighbouring Kirgistan. It had all the well-known characteristics of a satellite or rocket stage reentry. Indeed, as it turns out, this was a rocket stage reentry: the reentry of the upper stage of a Chinese Zhuque-2E (ZQ-2E) rocket, 2025-103G, from a multiple satellite launch on May 17.

Unfortunately, some Russian and Ukranian language twitter accounts started to disseminate the footage with the wildly wrong suggestion that this was a failing Russian Oreshnik IRBM breaking up. This misinformation next proliferated very quickly via various social media, and later also traditional media (e.g. Newsweek). This while it is not a Russian Oreshnik missile at all, as already mentioned.

I was alerted to the event somewhat later that evening when several of my social media followers tagged me and asked me for my opinion. It didn't take me long to identify the event as a space-launch related reentry rather than a Russian missile. Indeed, the Kazachstan MoD had meanwhile also deemed it a space debris reentry, per various news outlets.

The object in question was 2025-103G (cat. nr. 64054), the Zhuque-2E (ZQ-2E) upper stage from a Chinese satellite launch on 17 May 2025 from Dongfeng (Jiuquan). The upper stage from this launch was left in an in initially 600 x 175 km, 96.1 degree inclined orbit.

The CsPOC TIP for this object's reentry available at that time (it was updated later), had forecast reentry at 5 June 2025, 15:40 UTC with an uncertainty of  ± 3 hours. The Kazachstan event (~17h UTC) hence was within the reentry window. When I checked the trajectory over this full window, it showed that the object would make a south-to-north pass over eastern Kazachstan around 17:10 UTC, very close in time to the Kazachstan event (which was reportedly around 10 pm local Astana time = around 17 hr UTC). Direction of movement in the various videos of the event matched well.

Later, CSpOC published a final TIP placing the reentry at 5 June 2025 17:08 UTC  ± 1 min, near 36.6 N, 73.5 E. Given the 1 minute accuracy, this is likely based on a DoD satellite observation of the reentry fireball. The TIP position is just south of Kazachstan, but a reentry is a process of several minutes duration. Moving south-to-north, the fireball created by the fragmenting space debris would move northwards, over eastern Kazachstan, in the minutes directly after this time mark. In other words, this final TIP matches the event quite well.

Below is the trajectory over the final revolution of the ZQ-2E upper stage, based on the last available TLE (epoch 25156.48638369) which dates from ~5 hours before the reentry:

 

Click map to enlarge

I next ran a reentry model with our Delft Technical University open source Astrodynamics package Tudat. The ZQ-2E upper stage has a dry mass of approximately 5000 kg and measures 12 x 3.4 meter. Using 60% of the maximum drag area for that dimension, a value usually representing the average drag surface of a tumbling (and hence showing a variable drag surface) elongated rocket stage well, the Tudat model predicts reentry near 17:03 UTC ± 1.1 hr (remember, this is based on a 5 hour old orbit), the nominal value being well in line with the Kazachstan event and nominally within 5 minutes of the CSpOC final TIP.

Next I ran the model again adjusting the drag area slightly, via trial-and-error, such that the model would conform to the estimated start of visibility of the reentry fireball, at just below  ~100 km altitude, at the time/location of the CSpOC TIP (17:08 UTC). 

I get a reasonable match when I reduce the drag area to about 58% of the maximum drag area. Below is a map showing the resulting estimated reentry trajectory (movement is from south to north):

click map to enlarge

When I use this Tudat-estimated reentry trajectory to generate a sky track for Astana, Kazachstan, I get this approximate sky trajectory (movement is from left to right, i.e. south to north, low through the west):

Click map to enlarge

This conforms quite well to some of the video footage of the reentry (several of which show the reentry fireball pass the waxing moon), e.g.:

 

still from one of the movies posted on Twitter by @Buggy__Bugler

The predicted sky track for Astana of course depends on how accurate our Tudat-modelled atmospheric altitudes of the reentering rocket stage are. If they in reality are a bit lower than we modelled, the trajectory will be located slightly lower in the sky (and vice versa, slightly higher if the altitudes are in reality a bit higher). It is very clear however that the general direction and location of the trajectory in the sky matches well with what was seen and filmed.

As usual, I feel some frustration about the general absence of information on camera locations and time of the footage regarding the imagery of this event distributed via various social networks. Those data are important but almost never included. These matches of observations with reentry data would get so much easier if people posting footage would include the geographic location and the time. So please people: next time you post something like this, include these important data.

This event once again also showed the failure of AI as a reliable source of information for answering questions regarding events like this (we earlier saw that with the Kosmos 482 Descent Craft reentry too). Twitter's AI "Grok" generated some profound nonsense (i.e., misinformation) when people asked it to identify the character of the event, likely partly as a result of the large amount of disinformation already doing the rounds on social media about this event.