Tracking the Artemis II Lunar mission using TU Delft's MISO telescope

Artemis II imaged with the TU Delft 41.5-cm MISO telescope on 8 April 2026 

'track & stack' of 13 images on the Spacecraft movement. Artemis II is the dot in the yellow circle.

 

I was only two years old when, in December of 1972, Apollo 17 left the Moon, so I don't remember it. But untill last week it was the last time in my lifetime that humans had entered and left the Lunarsphere. So for me, as a nominal "consciously just post-Apollo person", the crewed Lunar flyby of Artemis II 'Integrity' this week was exciting and new. Humans flying to the Moon again!

Of course, as a satellite tracker, it would be nice to capture something of that yourself: not from a TV screen, but as an actual observation. And since shortly, I have access to a nice tool for that: our new MISO telescope ("Multiple Input Single Output'), part of the SCOMlab on the rooftop of our Delft University of Technology Faculty of Aerospace Engineering. It is a 41.5-cm F8 Aluna Ritchey-Chretien telescope on a planewave L500 mount, realised by my colleague Rudolf Saathof for experiments with laser satellite communication. Rudolf involved me in the acquisition process and as a result, the telescope got a configuration to make it also useful for SSA/SOT when not being used for lasersatcom. I specifically want to use this telescope for XGEO/CisLunar observations in the future.

 

TU Delft 41.5-cm MISO telescope

MISO telescope dome on the roof of the TU Delft Aerospace Engineering Faculty

TU Delft MISO 41.5-cm telescope

 

Unfortunately the ARTEMIS II flight trajectory during this mission was not very favourable for the Netherlands, with the Moon (its target) very low in the sky in this season. Over the full mission it barely came over the horizon, and initially weather was not quite cooperative as well.

On the night of April 7/8 2026, a day after the Lunar flyby, with Artemis II homebound but still close to one Lunar distance from Earth, weather prospects were good. Although I had my doubts (unjustified, as it turned out!), Rudolf encouraged me to try to capture ARTEMIS in the early morning of April 8.

Conditions were not very encouraging: the spacecraft would reach a maximum elevation of no more than 8.8 degrees (!) over the horizon, it would be twilight already by that time, and the waning gibbous moon would be a mere 15 degrees distant. Prospects of success looked dim to me. Yet, I decided to go for it, and downloaded the ephemeride predictions for that night through JPL HORIZONS.

I only recently have had my training on the MISO telescope from our technician Martin. Much of the system is still in the commissioning phase (and I still  have to create good flats and darks for example, as this calibration imagery is not yet existent), and it still has its occasional quirks and glitches. This would be only the second time I would use the system on my own, via a remote connection from my home to the dome and instrument: and this deep into the night I'd rather not get Martin out of his bed if things went south. So I was a bit nervous. 

I logged in at about 4 am local time, and started the process of going through the checklist, powering up and activating the dome, telescope and sensors, opening the dustcover and dome doors, followed by some time spent on focussing and finding the best camera settings. Then I moved the telescope to where Artemis II should be, only 8.7 degrees over the horizon (...!). After initially trying 1x1 binning, I moved to 4x4 binning and started to snap 10-second unfiltered images with the QHY600 MM PRO camera that is attached to the prime focus of the telescope.

After transferring the FITS files from the observatory server to my laptop at home, I looked through them. From a previous observing run, I already had created a configuration file for the telescope/camera combination for Astrometrica, my favourite astrometric software. I created a 'blink' (animated sequence, to look for moving objects) of four of the images in Astrometrica, obtained near 3:34 UTC (5:34 am local time), around the start of nautical twilight. 

And next my heart skipped a beat: there was a faint but unmistakable moving dot right where ARTEMIS should be! Hallelujah!  This was very cool!

 

Artemis II imaged with the TU Delft 41.5-cm MISO telescope on 8 April 2026 

The image above shows the 'blink' I created of four 10-second exposures that were taken over a 1.5 minute timespan, with Artemis II visible as a faint moving dot. 

I frankly did not expect the attempt to be successful, so I definitely was very positively surprised by this! The spacecraft turned out to be around magnitude +14, brighter than I had expected and that helped. The sky at low elevation was clear, and our telescope is located on a high-rise at 55 meters above the surrounding landscape, which also helps (a bit). Yet, the influence of atmospheric turbulence and extinction at such a low sky elevation is well visible in the imagery.

The image below is composed of 13 images (each a 10-second exposure) taken some 20 minutes earlier, where I have used the 'track and stack' method, stacking the images with each image shifted in the direction of movement of the spacecraft. As a result, the stars become trails, but the spacecraft is still a dot, and because it combines the light collected over 13 x 10 seconds, Artemis II is much better visible now. It is the dot in the yellow circle:

 

'track & stack' of 13 images on the Spacecraft movement. Artemis II is the dot in the yellow circle.

(this image was created without flat- and dark substraction, which shows, as it has vignetting and some obnoxious "dust donuts". Not the prettiest picture, but happy with the result).

Artemis was at a whopping range of 359 000 km (pretty much still one Lunar distance) from the observatory at that time!

At the moment of writing, Artemis II is still on its way back to Earth. The capsule with crew should reenter the earth atmosphere near 00:00 UTC on the night of 10-11 April 2026 (April 10 local time on the US west coast), having separated from the Command module around 23:44 23:33 UTC (April 10). It then parachutes down for a splashdown in the ocean near San Diego. 

Below, I have created a map showing the ground-projected return trajectory. It is based on vector and event timeline data from NASA and a ballistic continuation of the last predicted vector with Tudat towards the splashdown area. The red boxes are the exclusion zones from Navigational Warning NAVAREA XII 232/26 and NAVAREA XII 242/26: they represent the areas where the Command Module reenters into the atmosphere, and the Crew Capsule splashdown area some 35 km out of the California coast.

(updated map, as a new exlusion area in front of the California coast was added in Navigational Warning NAVAREA XII 242/26, and the map is now based on NASA provided trajectory and event data rather than JPL HORIZONS).

UPDATED MAP. Click to enlarge

Note that this is a ground-projected trajectory, and in the early part of the approach (before 23:00 UTC) the spacecraft is still at a large distance. The seemingly accute "turn" around 22:00 UTC is not a real turn, but a perspective effect created by the earth surface rotating underneath the approach path, specifically parts of the earth surface that initially were "behind" the earth limb turning to the front as seen from the approaching spacecraft: and initially the effects of earth rotation nunder the trajectory being the major factor, the as Artemis comes closer, relative speed with respect to the Earth surface taking over.. 

 

click image to enlarge

Note: Astrometry from the observations is here, with a typo correction here.

Guest in the Euronews podcast "Tech talks"

 

I recently was a guest in the Euronews podcast series Tech Talks, in an episode called "How to track a spy satellite". 

In it, host Alice Carnevali and me chat about how I came to work in the field of SSA (despite originally having a completely different scientific background); how we track spy satellites and what we can learn from that; what is currently happening in Space (notably in terms of military activities and threats) and what this could mean to us (civilians), and what I foresee will happen in the near future. The episode is 25 minutes long, and in English. The intro is quite James Bond-esque. Listen to it via the embedded youtube link above.

The observed reentry of the Chinese CZ-2C upper stage 2018-054D over Australia on 28 February 2026

still from reentry movie. Click to enlarge. image (c) Robert H. McNaught, used with permission

On 28 February 2026 near 18:40 UTC, a Chinese CZ-2C upper stage (catalogue nr 43521, International Des. 2018-054D) reentered over southeast Australia.

Astronomer Robert H. McNaught filmed the early stage of the reentry from Coonabarabran (near Siding Spring Observatory), New South Wales. The event occurred low - about 10 degrees - above his northern horizon. His footage captures just over a minute of the early stages of the reentry, showing a single object sporting a clear plasma tail. After about a minute, the object moves behind cloud cover, only briefly to be seen again in a gap very low on the horizon. Above is a still from his movie, showing the rockert stage as a bright dot, sporting a bright plasma tail in its wake. Below is a snippet of footage from the movie:

 

 

Enough stars can be seen in the footage to do decent astrometry. I therefore measured several astrometric positions from the reentry movie, using AstroRecord.

Next, I used the last available orbit (dating from about 6h 45m before the reentry) to create a reentry model, using our open source TU Delft Astrodynamics Toolbox (Tudat). The goal was to see if I could recreate the trajectory as observed from Coonabarabran with our model.

Setting the mass at 3800 kg, I varied the drag area in the model untill I got a trajectory that indeed closely matches that filmed from Coonabarabran. 

Below is the result: red crosses are astrometric positions from the video footage, the blue line is the Tudat-modelled trajectory, and the blue numbers along the trajectory are the modelled atmospheric altitudes of the reentering object, in km. 

 

click diagram to enlarge

 

There is a discrepancy of about 20 seconds in time between the modelled trajectory and the observations, but otherwise it fits well in terms of the observed sky trajectory. 

The exercise provides information on the approximate altitude of the object during the reentry observations, and the geographic extend of the reentry trajectory. 

The footage picks the object up at an altitude of ~86 km, just as it begins to ablate, forms a plasma tail, and becomes incandescent, and before it starts to break up. The object is lost behind cloud cover at around 80 km altitude. Breakup likely occurred later, after the object went behind the cloud cover, once it reached 70-60 km altitude. By the time of breakup, from the reconstructed trajectory, it was likely already over sea: any remaining fragments fell into the ocean in front of the east coast of Australia.

In the map below, the blue line is the final trajectory: the yellow line is the part of the reentry trajectory that was above the horizon as seen from Coonabarabran (the part covered by the footage is shorter).

click map to enlarge

 

No final TIP was issued for this reentry by the US Space Force. The footage by Rob McNaught and this analysis is the only confirmation on where the object reentered. 

The  pre-reentry forecast by our Tudat model (see https://reentry.langbroek.org), based on a drag area that was slightly smaller than the result of the current analysis, had it reenter 20 minutes later, which given the almost 7 hours gap between the last available orbit and the moment of reentry, is actually not that bad an estimate.

With many thanks to Rob McNaught for communicating his observations, and for allowing me to use his imagery in this blogpost. Reentry footage (c) Robert H. McNaught.

Another failed deorbit of a Falcon 9 upper stage

Falcon 9 upper stage. Image: SpaceX

On 2 February 2026, SpaceX launched Starlink 17-32, with 25 new Starlink satellites, from Vandenberg SFB in California. The satellites successfully reached Low Earth Orbit, but something went wrong with the upper stage, and not for the first time. SpaceX stated on their website:

During today’s launch, the second stage experienced an off-nominal condition during preparation for the deorbit burn. The vehicle then performed as designed to successfully passivate the stage. The first two MVac burns were nominal and safely deployed all 25 Starlink satellites to their intended orbit. Teams are reviewing data to determine root cause and corrective actions before returning to flight.

This is not the first time a controlled Falcon 9 upper stage deorbit failed: this in fact was the fourth time in barely 1.5 years time. 

In July 2024, during the launch of Starlink 9-3, the upper stage failed to re-light, the engine destructing instead, leaving the payloads in a too low orbit. The upper stage later made an uncontrolled deorbit.

In September 2024, the upper staged used for the Crew Dragon 9 launch failed to do a proper deorbit burn, with the stage missing the designated deorbit area. 

In February 2025, after the launch of Starlink 11-4, the upper stage failed to do a deorbit burn. The resulting uncontrolled reentry 18 days later caused a spectacle in the NW European sky, with parts surviving and impacting in Poland (see this earlier blogpost).

And now something similar happened again with the latest, 2 Feb 2026 launch. The upper stage (2026-022AB, catnr. 67675) failed its controlled deorbit. Initially left in a 110 x 241 km orbit, it had an uncontrolled reentry twelve hours after the launch. 

Where it deorbitted is not exactly known. The US Space Force/CSpOC published a TIP for 3 February 2:29 ± 1h UTC. Our Tudat model suggests reentry one orbital revolution later, at 3:47 ± 1.5 h UTC. The maps below give the trajectory over the uncertainty window of respectively the CSpOC TIP, and our Tudat model. The uncertainties overlap each other.

trajectory over CsPOC TIP uncertainty window. Click to enlarge

 

trajectory over the Tudat uncertainty window. Click to enlarge

 

The controlled deorbit aimed for should have happened around 17:35 UTC, just after completion of the first orbital revolution (see map below of what the original plan was). Instead, the upper stage continued to orbit for some 7 more revolutions (or 10 extra hours).

original deorbit plan. Click to enlarge

[UPDATED] You Only Die Twice (redux): the unusual and confusing double reentry of a ZQ-3 upper stage (2025-282A)

Click map to enlarge

(The title of this blogpost is inspired by a similar titled post from several years ago) 

The uncontrolled reentry of a large (assumed 7-8 ton weight) Chinese ZQ-3 (Zhuque-3) upper stage, 2025-282A (catnr. 66877), on 30 January 2026 for some reason created quite some public attention, especially in Europe. But  the event became decidedly unusual when, nine hours after the US Space Force published the Final TIP, a second Final TIP appeared. We have not seen that happening before. 

So what do we have here: an object that reentered TWICE?

The first published "final" TIP, published about an hour after the listed reentry time, was for 30 Jan 2026, 12:39 UTC ± 1m, near 54.3 S, 170.4 W. 

This time and location incidentally where a very close match to the final result of the experimental Tudat reentry model we were running for this object at Delft University of Technology (nominal 12:39 UTC, 56.0 S, 179.3 W, see here).

Jonathan McDowell and I believe that the final TIP's with a quoted 1-minute uncertainty are in fact based on space-based (satellite) observations of the reentry fireball, so they are accurate (and refer to the object starting to ablate at roughly 90-80 km altitude).

So far so good: observed and modelled reentry moment well in agreement. Nice!

But then it got confusing. Several hours later that day, the US Space Force published a second "final" TIP, also with a quoted 1-minute accuracy: 13:43 UTC ± 1m, near 3.9 S, 60.7 E. This is half-a-revolution (1h 4m) later than the first TIP.

 

Screenshot of the two relevant TIP's as published on Space-Track

 

Both locations are indicated by the yellow circles in the map in top of this post (the blue cross is the nominal result of our Tudat reentry model, the solid blue line the one-sigma uncertainty in that estimate).

So what happened here? How did this object appear to reenter TWICE?

While it could all be a clerical error or a mix-up/false detection, I suspect that this unusual "double reentry" is genuine. This particular reentry was from a somewhat eccentric orbit, more so that your average reentry. The last available orbit from ~2 revolutions before the 12:39 UTC TIP, was 211 x 102 km, with apogee decidely higher than perigee. Under such circumstances, parts might surve a low perigee (low enough to initiate ablation and partial reentry).

My suspicion therefore is that when the rocket stage initially reentered in perigee at 12:39 UTC and started to ablate and break up, a single massive/solid part survived this perigee and continued for half a revolution, before finally reentering at 13:43 UTC.

(alternatively, you could think of this as one reentry with a very, very long stretched debris strewnfield)

The longer surviving part might well be the dummy payload of this experimental launch, which remained attached to the ZQ-3 upper stage but might have separated from the upper stage during the 12:39 UTC perigee/reentry (edit: or the previous perigee pass, see post update below). If this dummy payload was a solid weight, meaning it had a much larger mass-relative-to-area (a lower area-to-mass ratio) than the rest of the rocket stage, it might have survived and come out of perigee again, while the actual upper stage meanwhile did not survive this perigee and reentered in the first spot at 12:39 UTC. The dummy payload then finally came down in the second spot at 13:43 UTC.

Although a different situation, it reminds me a bit of a confusing case from 2014, the reentry of a Russian Kobalt-M spy satellite (on which I also wrote under the title "You Only Die Twice" at the time, a blogpost which you can read here). The latter consisted of the uncontrolled reentry of shed parts over the USA, preceded by a controlled reentry of the film return capsule over Russia a few hours earlier. So a different situation, but equally confusing.

 

UPDATE  2 Feb 2026

I further investigated the hypothesis of a solid piece coming off the upper stage and surviving the initial 12:39 UTC reentry, by means of running trial-and-error models in Tudat, integrating the R/B to a certain time/altitude and taking the State Vector from that integration with a new mass and area to see if I could get a piece to survive to 13:43 UTC. I modelled for solid steel spheres (from the initial idea of a solid mass representing the dummy payload).

I have trouble to get anything surviving assuming it came off during or just before the 12:39 UTC first TIP. In order to create something that survives untill 13:43 UTC, I needed to go to a mass of 2500 kg coming off as much as half an hour before the 12:39 first TIP, at 129 km altitude: but then the remaining mass for the R/B without dummy payload does not have the R/B itself reenter at 12:39 UTC.

(added note: within error margins of the model, it however still might be possible) 

I do get a result that neatly matches both TIP's however, if I detach a small mass (only ~7.9 kg!) during the previous perigee pass, at about 11:25 UTC at an altitude of 109.5 km. This mass (with a corresponding diameter of about 12.4 cm for a solid steel object) in our Tudat model survives this perigee and the next and reenters at nominally 13:43 UTC at about 5 S, 62 E, in close agreement with the time and location of the second TIP. The R/B minus shed mass reenters earlier, around 12:39 UTC, the time of the first TIP. 

So, in summary of this scenario (see also map below):

Event I:  a ~7.9 kg. ~12.4 cm solid object detaches from the ZQ-3 R/B in perigee at 11:25 UTC;

Event II: the remaining ZQ-3 R/B reenters around 12:39 UTC at next perigee at location of TIP a;

Event III: the detached solid 7.9 kg object survives and reenters at 13:43 UTC at location of TIP b.

The object in question would however be too small to be a dummy payload (but could be a part of it) and I also wonder whether it is big enough to create a clear reentry fireball (clear enough to be seen from Space by SBIRS). So I am not entirely convinced this simulation solves the matter.

Click map to enlarge

(facetious added note: I am suddenly having hilarious visions of the dummy payload being a solid steel bobble-head statue of the Chinese LandSpace CEO, with the head coming off....)