The tumbling behaviour of the mysterious Kosmos 2553 satellite

Framestack (530 frames) showing variable brightness of Kosmos 2553. Click image to enlarge

Over three years ago, on 5 February 2022, Russia launched a mysterious military satellite, Kosmos 2553 (2022-011A), into an unusual orbit at approximately 1995 km altitude, the outermost margin of Low Earth Orbit. Very few satellites orbit there.

Early 2024, US Congressman Michael R. Turner, chairman of the House Intelligence Committee, wrote an unprecedented public letter to House members in which said he had concerns about a "serious national security threath", urging then President Biden to declassify the information. Subsequently, various US news sources quoted various of the proverbial "anonymous sources", with often conflicting information about the nature of the threath, but all indicating some kind of Russian space weapon. And moreover: a nuclear weapon, alledgedly. See my earlier 2024 blogpost here. Based on statements that a kind of prototype of the satellite in question was in Low Earth Orbit 'in a region not used by any other spacecraft', Kosmos 2553 was identified as the likely suspect.

More recently, in April 2025, various news sources (e.g. here and here) reported that as of late 2024, Kosmos 2553 had started to tumble, indicating a possible loss of attitude control.

I imaged Kosmos 2553 on May 20, 2025, and it indeed shows a brightness variation that was not present when I imaged it a year earlier. The image above is a 530-frame (21.2 second) stack, and the brightness variation can be clearly seen in it. Below is a sequence of the actual video footage:

 

We can compare this to video footage from a year earlier (20 May 2024) when the object was steady:

 

I extracted almost 9 minutes of photometric information from the 20 May 2025 video. This shows a prominent flash cycle of (peak-to-peak) 2.22 seconds, with a regular pattern consisting of a brighter flash followed by a fainter flash, ad infinitum. 

Below is a diagram of the full 9-minute photometry series, and a detail of a part of the curve which shows the pattern of the brightness variation: the red line is a fitted multi-sinusoid who's main period is 2.22 seconds. Gaps in the data are moments the camera was repositioned, or the object was closely passing a star.

Click diagram to enlarge

 

 

Click diagram to enlarge

The datapoints in the diagrams are 5-frame running averages. The data in the two diagrams above have been corrected for range and phase angle variation, i.e. to absolute magnitude (normalisation to 1000 km range and 90 degree phase angle). 

The apparent observed magnitude varied between magnitude +5.7 and +9.4. Below are these apparent photometric measurements uncorrected for phase angle and range (note that a calibration of the data to the Visual band has been done to correct for instrument spectral sensitivity):

 

Click diagram to enlarge

The imagery was made from my home in Leiden, the Netherlands, with a WATEC 902H2 Supreme camera and Samyang 1.4/85 mm lens, filming at 25 frames/second. 

The photometry clearly supports reports that Kosmos 2553 has started a tumble or spin. Whether this means it is no longer operational, is another question that is less easily answered. Given the regularity of the flash period, the flashing could be due to spin stabilization. On the other hand: why did this only become apparent some 2 years into the mission?

In orbital data for Kosmos 2553, a sudden subtle change in orbital altitude can be seen starting around 15-16 November 2024 (see diagram below). Perhaps this is when the tumbling or spin started.

Click diagram to enlarge

Multiple analysts, including myself, believe Kosmos 2553 to be a (Radar) imaging satellite (possibly 'Neitron'). It has a ground track that after four days closely repeats itself, which would fit an imaging satellite. It is not clear why some in US Government circles believe that Kosmos 2553 is connected to a 'nuclear space weapon' program (presumably Ekipazh). That suspicion must be based on undisclosed HUMINT.

Russia itself has stated that Kosmos 2553 is a "technological spacecraft […] equipped with newly developed onboard instruments and systems for testing them under the influence of radiation and heavy charged particles". That explanation does not sit entirely well with several analysts: yes, at 2000 km altitude the radiation regime is different and more severe compared to a more typical Low Earth Orbit: but not thát much different and severe, really.

The tumble periodicity of the Chang'e 3 upper stage (2013-070B) revisited

click image to enlarge

Brightness variation due to tumbling of the Chang'e 3 upper stage (2013-070B)
stack of 15 images taken with the 0.51-m telescope of MPC Q65 Warrumbungle
11 September 2015

I have written before on this blog about tracking very distant space debris: the CZ-3C upper stages of the Chinese Lunar missions Chang'e 2 and Chang'e 3, which move in chaotic trans-Lunar orbits. I have embarked on a long-term project to follow these objects.

Apart from positions to keep their orbits up to date, these observations also provide information about the tumbling behaviour of these objects. Both objects have a periodic variation in brightness: a very rapid one for 2010-050B, the Chang'e 2 upper stage, and a slow one for 2013-070B, the Chang'e 3 upper stage.

Earlier, in July 2015, I had established a tumbling periodicity of  ~7 minutes for 2013-070B. I have now been able to refine that value much better, to only a few hundreds of a second.

With the help of Peter Starr from Warrumbungle Observatory (MPC Q65) in Australia and Krisztián Sárneczky from Szeged University's Piszkéstető Observatory (MPC 461) in Hungary, I could obtain two nice series of data the past week. The data were gathered on September 11 (Warrumbungle 0.51-m telescope) and September 14 (Piszkéstető 0.60-m Schmidt telescope).

The first set, taken by Peter from Warrumbungle, is a set of 15 exposures of 30 seconds each, taken in ~1 minute intervals. The image at the top of this post is a stack of these images. The brightness maxima can be clearly seen.

The second set, obtained in twilight by Krisztián from Piszkéstető at the end of a run of the Szeged Asteroid Survey, is a set of 18 exposures of 3 seconds (!) each, in ~20 second intervals, with a brief pause halfway the series.

Single sinusoid fit to data from Sep 11 (lef) and Sep 14 (right)
click diagram to enlarge

The data allow to fit a sinusoid to both sets simultaniously, and from that get a very accurate periodicity. The double diagram above shows this sinus-fit to the data. It allows to establish a peak-to-peak periodicity of 423.01 ± 0.03 seconds for the tumbling of 2013-070B.

The flashing behaviour of North Korea's tumbling Kwangmyongsong 3-2 satellite

North Korea's first satellite Kwangmyongsong 3-2 (KMS 3-2) cannot be seen from the northern hemisphere at the moment (and hence cannot be observed by me currently). On the southern hemisphere, Greg Roberts (CoSatTrak) in South Africa is however successfully tracking the satellite.

He had a particular good pass on December 20th and obtained a very nice video record, tracking on the satellite with a motorized mount (note: movie has a period of black screen between opening title and start of the video record):

Greg Robert's video from S-Africa

(posted with permission)

The satellite is the object near the center of the screen, flashing about each 8.5 seconds with periods of invisibility inbetween. The moving streaks are stars (the mount is tracking the satellite as it moved along the sky): the other stationary dots in the image are hot pixels on the sensor of the video camera.

The video allows for an analysis of the flashing behaviour of the satellite. I used LiMovie to measure the satellites' brightness on the frames, resulting in the following lightcurve:

click diagram to enlarge

Visible is a clear ~8.45s periodicity with flashes of a specular character (suggesting a flat reflective surface). I have marked this with red triangles 8.45 seconds apart. In between the main flashes, a pattern of smaller secondary flashes can be discerned in a semi 8.45 second peridicity too (green triangles). They are not exactly positioned halfway between major flashes.

Assuming that each major flash is a flash caused by one of the sides of the KMS 3-2 cube-shaped body, then it completes a tumble once every ~33.8 seconds. Assuming that the less clear secondary flashes are due to a side of the cube as well, the tumbling periodicity would be half of that, i.e. 16.9 seconds.

Greg recorded the UNHA-3 r/b from the launch too. That one too is tumbling:

Greg Robert's video from S-Africa

(posted with permission)

Again, I used LiMovie to extract brightness information from each video frame. That was less successful with this video, because Greg's mount had difficulty keeping up with the fast-moving r/b for much of the record. A considerable part of the video could not be used for analysis, and I had to chop up the analysis in little non-continuous chunks:

click diagram to enlarge

What can be seen, is a flashing behaviour that starts slow and gentle and is increasing in rapidity near the end of the analysis, this being an effect of changing viewing angle.

Contrary to what some alarmist (sometimes almost hysterical) media reports have suggested, the tumbling of KMS 3-2 is by no means dangerous. David Wright over at All Things Nuclear has a very good debunking story about this all, pointing out the many misconceptions rampant in the reporting.

Brightness variability of the NOSS 3-3 Centaur Upper Stage (with video)

NOSS 3-3 is a pair of US Navy NOSS surveillance satellites launched early 2005. The Centaur upper stage of this NOSS 3-3 launch, NOSS 3-3r (2005-004B) is still orbiting earth as well. And as it does so, it is tumbling.

This tumbling is visible to an observer as a regular variation in brightness. Currently, the rocket stage brightens up every 11.4 seconds.

Below video shows the regular variation in brightness: watch it go from faint to bright to faint etcetera with an 11.4 second period. It is footage from a pass over Leiden which I filmed in the evening of March 22, using the WATEC 902H and a Canon EF 2.0/35 mm lens:

click images to enlarge

Using LiMovie, I extracted the brightness variation from the movie on a frame by frame basis, resulting in the depicted brightness profile above. Note that the tops of the curve are sharp, not rounded. It is a nice saw-tooth pattern. The integrated video frame picture shows the brightness variability nicely too.

Documenting this kind of tumbling behaviour (and notably how it changes over the years) can actually provide some valuable scientific data. A number of amateur observers specialize in these "flash observations", notably my fellow members of the Belgian Working Group Satellites (BWGS).

Altering tumbling period of the USA 144/Misty-2 decoy (1999-028C)

In August I used a series of photographs to determine the tumbling period of 99-028C, the enigmatic USA 144/Misty-2 "Decoy" (see here).

As the tumbling period of this object is known to alter, I am repeating the experiment. I still need some additional nights to construct a full curve: but the partial curve obtained from the November 19 observations (6 images) already shows a clear change compared to August:

click diagram to enlarge

The sinusoid is for a period of 62 seconds, which compares well to a very similar period visually determined by Ted the same night. It is nice to see the two results coming out so similar.

Back in August the period was 71 seconds. A change of 9 seconds in 3 months time.

In the diagram above, the greyed data points are data from when the trail was very close to both edges of the FOV. Their absolute levels have suffered from lens vignetting, so I scaled them to show that the trend of these points at least is similar to the trend of the period determined from the other four images. The black data points are raw, unaltered data from the latter images.

The tumbling period of the USA 144/Misty-2 Decoy (99-028C)

On August 25 and August 27, I obtained a series of photographs of the USA 144/Misty-2 decoy (see here and especially here). On the request of Pierre Neirinck, I did some simple photometry on the image series, to see whether I could retrieve information out of this on the current tumbling period of the object.

It concerns the following two image series. In both cases, the object is moving from right to left, and the image series has to be "read" from right to left (i.e., the most right image is the earliest in time, the most left the latest). Some clear brightness variation from image to image is already apparent.

Click images to enlarge

From these images, the following two partial photometric curves were obtained, both suggesting a period near ~70 seconds:

Click images to enlarge

Next I combined these two partial curves (from two separate nights) into one amalgamated curve (this included of course some data normalization/scaling), shown below:

Click image to enlarge

The result is a curve that can be approximated by a sinusoid with a period of 71 seconds. This suggests the object's current tumbling period is 2 * 71 = 142 seconds.

(in the curve above, the dark dots are (normalized) data points, the grey line is a 20-point running average, and the black line a sinusoid with a period of 71 seconds)