One month left for Tiangong-1 [UPDATED]

Note: a daily updated post with reentry estimates for Tiangong-1 is here.

image (c) Alain Figer, used with permission

The beautiful image above (used with kind permission) was made by Alain Figer and shows the Chinese Space Station TIANGONG-1 over the French Alps on 27 November 2017.

Tiangong ("Heavenly Palace") 1 was launched on 29 Sept 2011. It was the first Chinese Space Station and was visited by Taikonauts twice, first by the crew of Shenzou 9 in June 2012 and then by the crew of Shenzou 10 in June 2013: six Taikonauts in total.

All eyes are currently on this Chinese Space Station, as it is about to re-enter. Since the station was shutdown in 2016, it has steadily come down, especially so the past year and months. Its orbital altitude has currently descended below 250 km (it currently is ~240 km, with apogee at 251 km and perigee at 229 km on 2018 March 13):

click diagram to enlarge

click diagram to enlarge

Using SatAna and SatEvo, and under the assumption that the re-entry will be completely uncontrolled, I currently estimate it to re-enter one month from now, somewhere between April 7 and April 21  April 1 and April 12.

EDIT:  daily updated re-entry predictions are in a dedicated post here

The station has an orbital inclination of 42.8 degrees, and hence can come down anywhere between 42.8 N and 42.8 S. The map below shows the area that is at risk:

click map to enlarge

Note that newspaper accounts (e.g. this one) that single out a particular area as being at particular risk, are nonsense: At this stage, a month before re-entry, it is impossible to pinpoint a region. That will only be possible during the hours just before actual re-entry (and even then...).

The station has a mass of about 8500 kg and measures 3.35 x 10.4 meter. It is hence a large and heavy object, which is why this re-entry is of concern. It is likely that parts will survive the re-entry and reach Earth surface intact.

Land masses inside the risk zone include southern Eurasia, Australia, Africa, South and Middle America and the United States. It is however most likely that the re-entry will be over an ocean.

As can be seen from the map above, my own country, the Netherlands, is well outside the risk zone.

I will follow the orbital evolution and re-entry predictions for Tiangong-1 on this blog as they evolve.

Tiangong-1 image on 18 July 2017 by Alexandre Amorim from Brazil
this is a stack of 4 separate images
(image (c) Alexandre Amorim, used with permission)

NOTE: new reentry estimates, updated daily, are consolidated in this new blog post.

Analysis: The re-entry of the CZ-4B r/b 2014-049C observed by a Dutch pilot on May 27 [UPDATED]

click to enlarge. Image (c) Christiaan van Heijst, used with permission

click to enlarge. Image (c) Christiaan van Heijst, used with permission

The beautiful, spectacular images of a rocket stage re-entry above were made by the Dutch aviation photographer and pilot Christiaan van Heijst,  the co-pilot of a Cargolux freight aircraft (flight CV760, a Boeing 747-8 with registration LX-VCC) en route to Brazil on May 27, 2017.

While cruising at FL 340, 34 000 feet (10.360 km) over the mid-Atlantic, Christiaan noted a group of 7 to 10 bright yellow, very slow fireballs appearing in the corner of his eye. Here is the story as told by Christiaan on his facebook page:

Suddenly I noticed something in the corner of my eye. I looked to my right and to my own surprise I saw a huge group 7-10 of bright yellow lights move parallel to our track with a much faster speed and very high altitude. This was not an airplane, nor was it a meteorite. Where shooting stars / meteorites often leave a bright trail, they move with very high speed and burn up quickly. This cluster of lights moved far too slow to be a meteorite and its light was far too constant to be an ordinary meteorite. 

Immediately, a lot of excited chatter in Portuguese and other (African) languages I could not identify. was opening up on the frequency we had tuned in. Apparently lots of pilots were seeing the same lights, which is not surprising with such a high and bright appearance. All in all, the lights appeared abeam our aircraft and disappeared on the horizon in roughly two minutes time, keeping their intensity and appearance along the way.

Evidently, what Christiaan and his colleagues were witnessing was a spectacular re-entry of space debris, with the re-entering object breaking up in multiple pieces while it was plunging through the atmosphere. The time of this re-entry event was around 23:18 UT on May 27, 2017, while the aircraft was over the mid-Atlantic near 11^o.93 N, 33^o.28 E (see also later in this post).

In this blog post, I identify the object responsible and provide some model results for this re-entry.

click map to enlarge

Christiaan van Heijst initially thought that this re-entry event was related to a NOTAM issued mid-May, a warning for the splash-down of a Soyuz 2nd stage during the SES-15 launch from Kourou. This launch however had already happened 10 days earlier, on May 18, so evidently was no explanation for this event. Christiaan next posted his story on Facebook, hoping that someone could identify the object responsible.

I was allerted to Christiaan's Facebook post by one of my Twitter followers, Theo Dekkers and could quickly identify the event as the re-entry of 2014-049C, a Chinese Chang Zheng (Long March) 4B (CZ-4B) upper stage from the launch of the Chinese Gaofeng 2 and Polish Heweliusz satellites in August 2014. Time, location, and movement of the witnessed event agree extremely well.

Two days before the sighting, JSpOC had started issuing TIP (Tracking and Impact Prediction) messages for this object via their Space-Track portal. The final TIP message, issued after the actual re-entry, lists the re-entry time as 27 May 23:17 +- 1 min UT, near 15^o.7 N, 34^o W (by the way: we actually believe that such times accurate to 1 minute originate from infrared observations of the re-entry fireball by US SBIRS early warning satellites).

click to enlarge

This time and position closely agrees with the observations of the aircraft crew and the aircraft position. Christiaan van Heijst provided me with a photo of the aircraft flight instruments taken about one minute after the event. It shows the time of that moment, 23:20:43 UT, and the aircraft's GPS coordinates and altitude: 11^o 56.1' N (11.935 N), 33^o 17.3 W (33.288 W) at a flight level of 34 000 ft (10.360 km). [edit: the altimeter in the image above says 33 960 feet but Christiaan informed me that it has a small error and they were flying at FL 340]. The aircraft was heading towards a magnetic bearing of 219 deg, which corresponds to a true bearing of 204 degrees (towards the S-SW).

The time and position are very close to that of the TIP: a difference of about 425 km between the TIP re-entry location and the location of the aircraft, and 1-2 minutes in time.

The sky track of the re-entering space debris that can be seen on the photographs also agrees well with the predicted sky track of 2014-049C for the aircraft's location. Below is the predicted track for 2014-049C for the location of the aircraft based on a propagated version of the last available orbital element set for the object. The blue line is the predicted track in the sky, the yellow arrow the approximate trajectory for the brightest fragment visible on Christiaan's photographs:

click to enlarge

There is a discrepancy, in that the observed trajectory is some 11 degrees lower in the sky than the predicted trajectory, with a time lag as well. However, this is what you expect. The track shown is for the pre- re-entry orbital altitude (about 134 km). During the re-entry phase, the altitude of the object however quickly drops, and as a result the observed track will be located significantly lower in the sky. As the object is slowed down by increasing drag of the atmosphere, it starts to lag behind predictions in time as well. At the time of the re-entry, the object was already below 80 km altitude,  40% or more below its orbital altitude.


[UPDATE  6 Oct 2017:]

I have since used the output of a GMAT re-entry model (see below) to reconstruct the expected trajectory in the sky as seen from the aircraft. For this purpose, I used the latitude, longitude and altitude output of the GMAT model, converted these to ECEF coordinates, did the same for the position of the aircraft, and then with the help of relevant equations calculated the azimuth and elevation of the reentering rocket stage as seen from the aircraft from these. The sky positions were plotted on a star map for the location of the aircraft. The result is below (compare to the two photographs in top of this post):

click map to enlarge

As can be seen, the modelled sky trajectory, while not a perfect fit, is nevertheless very close to that visible on the photographs.

Note that the GMAT reentry model, while modelling the influence of the atmosphere, does not take fragmentation and ablation (and from that mass-loss and changes in surface:mass ratio) into account.

[END OF 6 Oct 2017 UPDATE]

To gain insight into the positions and altitude of  the re-entering debris over time relative to the aircraft, I have modelled the re-entry event. I propagated the last five known orbital element sets (TLE) for 2014-049C to its last ascending node passage before re-entry, using SatAna and SatEvo. The resulting, final, pre- re-entry TLE was next used as the starting point for a ballistic simulation in GMAT, using the MSISE90 model atmosphere and actual Space Weather data. With this input, I had GMAT calculate positions and altitudes of the re-entering object over time.

Such modelling always is an approximation only. There are a number of unknowns, one of which is the spatial orientation of the major axis of the re-entering rocket stage with regard to its flight direction. This adds uncertainty to modelling the atmospheric drag experienced by the re-entering rocket stage, which introduces uncertainties in the position and altitude of the stage for a certain time. A CZ-4B 3rd stage is a tube measuring 6.24 x 2.90 meter with a dry mass of about 1 metric ton. The drag experienced depends on whether its longest dimension is facing the flight direction, its narrow end, or whether it tumbles. For the modelling, I choose to use a drag surface that is 50% of the maximum drag surface possible. Breakup of the rocket body, which is evidently happening (see the copious fragmentation in the photographs) adds more uncertainty, as fragmentation drastically alters the drag surface and surface-to-mass ratio. As the images show, the trajectories of individual fragments clearly start to diverge as a result of this. The model, however, treats the re-entering body as one single body with no mass loss.

So, Caveat Lector. But let us look at the results. Mapping the GMAT results along with the position and bearing of the aircraft a minute after the event, yields this positional map and this altitude versus time profile:

click map to enlarge

click diagram to enlarge

For the reasons mentioned above, the altitudes versus time in the diagram are approximations only, with a possible uncertainty of perhaps 25% for a given time instance.

Compared to the JSpOC TIP data, the resulting trajectory I modelled seems to be slightly on the 'early' side, in that it passes the JSpOC location about a minute too early. On the other hand, the time in the TIP is given with an accuracy of no better than 1 minute, and an unspecified inaccuracy in the coordinates of the geographic location as well. What we can conclude from the modelled positions relative to the aircraft, is that the sighting definitely matches the 2014-049C re-entry data closely.

If my modelling is somewhat correct, the re-entering debris was moving from altitudes of ~95 km at the start of the sighting to below 50 km near the end [update 6 Oct 2017: The closest it came to the aircraft was a line-of-sight minimum distance of 157 km near 23:16:50 UT]. It is uncertain whether anything survived to sea level c.q. aircraft flight level. Usually, most materials have burned up before they could reach the surface: it is however not impossible that some pieces nevertheless survived and splashed down in the Atlantic. Notably the pressure spheres of rocket engines tend to survive. If anything, the modelling shows that any surviving debris was well ahead of the aircraft once it reached the flight level of the aircraft.

Ted Molczan has done a similar modelling with similar results. The differences that do exist between Ted's analysis and my results, are due to the choice of slightly different starting parameters for the model.

The final spectacular demise of 2014-049C was the result of a long drop that started short after launch. Below I have mapped the evolution of the orbital altitude of the rocket booster over the past years, starting just after launch:

click diagram to enlarge

The quick decay of (notably) the apogee altitude, but also perigee, can be clearly seen. Early 2017, the drop in altitude starts to increase exponentially. At 23:17 UT on 27 May 2017, after 15772 revolutions around the planet since launch, it was the final end for 2014-049C.

Christiaan asked me why there was no NOTAM issued for this re-entry. NOTAMS or Area Warnings are however generally only issued for controlled de-orbits, and first and second stage splashdowns during launches. Reasonably accurate locations can be predicted in advance for these. For uncontrolled re-entries, such as this event, this is not the case. There are so many uncertainties that anything approaching an accurate prediction can only be issued during the last hour or so before re-entry.

(note 1: for some Frequently Asked Questions about re-entries, see an earlier post here).
(note 2: this post was updated on 6 October 2017 to add some new modelling results)

Acknowledgement: I thank Christiaan van Heijst (www.jpcvanheijst.com) for providing extra information and for his permission to use his photographs. I thank Theo Dekkers for pointing me to Heijst's observations.

Progress M-27M is down!

click map to enlarge

According to US military tracking data from JSpOC, the out of control Progress M-27M cargoship that should have brought supplies to the ISS, re-entered over the southeast Pacific near 51 S, 87 W, moving towards Tierra del Fuego, on May 8 at 02:20 UT, +- 1 minute.

The map above shows its approximate re-entry position. Remember: re-entry in reality is a process that takes several minutes, during which it moves along the shown track (shown track is the last orbital revolution plus a small part of the track in front of the nominal re-entry position).

Several analysts believe that re-entry times given to a plusminus of only 1 minute by JSpOC are based on Space-based detections by the US military's SDP DSP and SBIRS infra-red early warning satellites. These detect and geolocate the infra-red signature of the fireball caused by the re-entry.

In this particular case, this is the more likely because the last three published orbital element sets from (we assume) regular tracking facilities are clearly not too accurate. So the very short uncertainty interval in the re-entry time given, must be based on some other unpublished data source.

The clear problems with the last few element sets issued is one reason why I did not make any further forecasts after the one I issued at 18:30 UT yesterday, 8 hours before the actual re-entry.

My prediction at that time (2:03 UT +- 2 hrs, which I rounded to 2:00 UT +- 2 hrs given the uncertainty interval) actually is not that far off from JSpOC's final re-entry time.

Meanwhile, official Russian ROSKOSMOS sources give a re-entry time that is 15 minutes earlier (and very similar to my prediction hours before) than that suggested by US military tracking sources. The Russians state re-entry at 2:04 UT (5:04 Moscow time) over the central Pacific south of Hawaii (red dot on the map above). This is likely to be a 'forecast' (or 'aftercast' rather) based on tracking data and/or telemetry obtained during the hours before. No uncertainty interval is given.

[UPDATED] Progress M-27M re-entry predictions

[predictionlast updated 7 May 18:30 UT]
[update 8 may 2015: NEW post  about the re-entry HERE]

I am providing occasional estimates for the uncontrolled re-entry of Progress M-27M on my Twitter feed today. Likewise, several other people are providing estimates, of which I would want to recommend those by Ted Molczan on the Seesat mailing list. You will note that the re-entry predictions will vary from source to source!

My current estimate is for re-entry to occur:

between 17 and 22 UT on May 7
between 21:10 and 01:30 UT  May 7-8

between 00:00 and 04:00 UT May 8

(estimate update issued 7 May, 18:30 UT based on elsets up to epoch 15127.69084936 processed with Alan Pickup's SatAna and SatEvo software).

Within this window, the slightly more likely moments are around and just after perigee passage, i.e. the tracks over the Pacific in this case.


The estimate may still change as new orbits are released. Previous released elsets initially strongly pushed the estimated re-entry time back from early May 8 to late May 7, as the result of solar activity yesterday and the resulting effects on the outer Earth atmosphere. As a result, my earlier estimate of this morning (May 7) overestimated the decay rate. With the new orbital update from epoch 15127.50860911 and onwards things are settling towards more realistic values, with a trend of slightly moving the decay window to a later time.

Above is a rough map of where the re-entry might occur, based on current uncertainties.

For updates, keep an eye on my Twitter feed.

[update: final re-entry results have been posted in a new post here]

Tweets door @Marco_Langbroek

Analysis of the 2014-074B Soyuz r/b re-entry on 26 Nov 2014

In the early morning of 26 November 2014 between 03:35 and 03:40 UT, a very slow, long duration fireball was observed from the Netherlands, Germany and Hungary (see earlier post).

The fireball was quickly suspected to be caused by the fiery demise of a Soyuz third stage, used to launch ISS expedition crew 42, including ESA astronaut Samantha Cristoforetti, to the International Space Station on November 23.

Video still image from Erlangen, Germany (courtesy Stefan Schick)

Analysis

In this blog post, which is a follow-up on an earlier post, I will present some results from my analysis of the re-entry images, including a trajectory map, speed reconstructions and an altitude profile. The purpose of the analysis was:

1) to document that this indeed was the re-entry of 2014-074B;
2) to reconstruct the approximate re-entry trajectory;
3) to reconstruct the approximate altitude profile during the re-entry.


Data used

Three datasets were available to me for this analysis:

1) imagery from three photographic all-sky meteor cameras in the Netherlands, situated at Oostkapelle, Bussloo and Ermelo (courtesy of Klaas Jobse, Jaap van 't Leven and Koen Miskotte);

2) data from two meteor video camera stations (HUBAJ and HUBEC) situated in Hungary (courtesy of Zsolt Perkó and Szilárd Csizmadia);

3) imagery from a wide angle fireball video camera situated at Erlangen, Germany (courtesy Stefan Schick).

Some example imagery is below:

Detail of one of the Bussloo Public Observatory (Netherlands) all-sky images, courtesy Jaap van 't Leven

Detail of the Cyclops Oostkapelle (Netherlands) all-sky image, courtesy Klaas Jobse

Detail of the Ermelo (Netherlands) all-sky image, courtesy Koen Miskotte

Stack of video frames from Erlangen (Germany), courtesy Stefan Schick

Stack of video frames from HUBEC station (Hungary), courtesy Szilárd Csizmadia and Szolt Perkó

Astrometry

The Hungarian data had already been astrometrically processed with METREC by Szilárd Csizmadia and came as a set of RA/Declination data with time stamps. The Dutch and German images were astrometrically processed by myself from the original imagery.

The German Erlangen imagery was measured with AstroRecord (the same astrometric package I use for my satellite imagery). An integrated stack of the video frames resulted in just enough reference stars to measure points on the western half of the image. As it concerns an extreme wide field image with low pixel resolution and limited reference stars, the astrometric accuracy will be low.

AstroRecord could not be used on the Dutch All-Sky images because of the extreme distortion inherent to imagery with fish-eye lenses. They were therefore measured by creating a Cartesian X-Y grid over the image, centered on the image center (the zenith). Some 25 reference stars per image were measured in this X, Y system, as well as points on the fireball trail. From the known azimuth and elevation of the reference stars, the azimuth and elevation of points on the fireball trail were reduced. While obtaining the azimuth with this method is a straight forward function of the X, Y angle on the images, obtaining the elevation is more ambiguous. Based on the known positions of the reference stars and their radius (in image pixels) with respect to the image center, a polynomial fit was made to the data yielding a scaling equation that was used to convert the radius with respect to the image center of the measured points on the fireball trail to sky elevation values.

Unlike meteoric fireballs, rocket stage re-entries are long-duration phenomena. The German and Hungarian data, being video data, had a good time control. The Dutch all-sky camera data, being long duration photographic exposures, had less good time control, even though the start- and end-times of the images are known. The trails for Oostkapelle and Ermelo had no meaningful start and end to the trails. Bussloo does provide some time control as the camera ended one image and started a new one halfway the event: the end point of the trail on the first image corresponds to the end time of that image, and similarly the start on the next image corresponds to the start time of that image. There was 7 seconds in between the two images. Time control is important for the speed reconstructions, but also for the astrometry (notably the determination of Right Ascension).


Data reduction and problems

The Azimuth/Elevation data resulting from the astrometry on the Dutch data were converted to RA/DEC using formulae from Meeus (1991). For these Dutch data, the lack of time control is slightly problematic as the RA is time-dependent (the declination is not). There is hence an uncertainty in the Dutch data.

The data where then reduced by a method originally devised for meteor images: fitting planes through the camera's location and the observed sky directions, and then determining the (average) intersection line of that plane with planes fitted from the other stations, weighted according to plane intersection angle. This is the method described by Ceplecha (1987). The plane construction was done in a geocentered Cartesian X-Y-Z grid and hence includes a spherical earth surface. The whole procedure was done using a still experimental Excel spreadsheet ("TRAJECT 2 beta") written by the author of this blog, coded serendipitously to reduce meteoric fireballs a few weeks before the re-entry.

I should warn that this method is actually not too well suited to reduce a satellite re-entry. The method is devised for meteoric fireballs, who's luminous atmospheric trajectory is not notably different from a straight line (fitting planes is well suited to reconstruct this line). A rocket stage re-entering from Low Earth Orbit however has a notably curved trajectory: as it is in orbit around the geocenter, it moves in an arc, not a straight line. This creates some problems, notably with the reconstructed altitudes, and increasingly so when the observed arc is longer. Altitudes reconstructed from the fitted intersection line of the planes come out too low, notably towards the middle of the used trajectory arc. The resulting altitude profile hence is distorted and produces a U-shape. The method is also problematic when stations used for the plane fitting procedure are geographically far removed from each other. In addition, the method is not very fit for long duration events.

The data were reduced as three sets:

1) data from the Dutch stations (independent from the other two datasets);
2) data from the German station combined with the two eastern-most Dutch stations;
3) data from the Hungarian stations (independent from the previous two datasets).

Dataset (2) combines data from stations geographically quite far apart. This is probably one of the reasons why this dataset produces a slightly skewed trajectory direction compared to the other two datasets.

The Dutch images have the event occurring very low in the sky (below 35 degrees elevation for Oostkapelle and below 25 degrees elevation for Bussloo and Ermelo). The convergence angles between the observed planes from the three stations is low (14 degrees or less). This combination of low convergence angles and low sky elevations, means that small measuring errors can have a notable scatter in distance as a result.


Results (1): trajectory

reconstructed trajectory (red dashed line and yellow dots)

The map above (in conic equal-area projection) shows the reconstructed trajectory as the yellow dots and the red dashed line. White dots are the observing stations.

The thin grey line just north of the reconstructed trajectory is the theoretical ground track resulting from a SatAna and SatEvo processed TLE orbital efemerid set for the rocket stage. This expected ground track need not perfectly coincide with the real trajectory, as the orbit changes rapidly during the final re-entry phase.

The reconstructed trajectory converges towards the theoretical (expected) ground track near the final re-entry location, above Hungary, but is slightly south of it earlier in time. The horizontal difference is about 30 km over southern England, 25-20 km over northern France due south of the Netherlands, 17-16 km over southern Germany and less than 3 km over Hungary.

This difference is most likely analytical error, introduced by the low sky elevations and convergence angles as seen from the Netherlands. On the other hand, the Hungarian observations (with stations on the other, southern side of the trajectory compared to the Dutch and German stations and reduced completely independent from the other data) place it slightly south too. So perhaps the deviation is real and due to orbital inclination changes during the final re-entry phase. Indeed, a SatEvo evolution of the last known orbit suggest a slight decrease in orbital inclination over time, although not of the observed magnitude.

The results from Erlangen come out slightly skewed in direction, likely for reasons already discussed above. The Hungarian results are probably the best quality results.


Results (2): altitudes and speed

altitudes (in km) versus geographic longitude

As mentioned earlier, the altitudes resulting from the fitted linear planes intersection line come out spurious due to the curvature of the trajectory. Altitudes were therefore calculated from the observed sky elevations and known horizontal distance to the trajectory. The horizontal distance "d" between the observing station and each resulting point on the trajectory were calculated using the geodetic software PCTrans (software by the Hydrographic Service of the Royal Dutch Navy). Next, for each point the (uncorrected) altitude "z" was calculated from the formula:

         z = d * tan(h),

where "h" is the observed sky elevation in degrees.

This is the result for a "flat" earth. It has hence to be corrected for earth surface curvature, by adding a correction via the geodetic equation:

        Zc =  r - sqrt (r ² + d ²)     [all values in meters],

where "r" is the Earth radius for this latitude, "d" the horizontal distance between the observing station and the point on the trajectory, and "Zc" the resulting correction on the altitude calculated earlier.

The results are shown in the diagram above, where the elevation has been plotted as a function of geographic longitude. It suggests an initial rapid decline in altitude from ~125 km to ~100 km between southern England and northern France, an altitude of ~100 km over southern Germany, and a very rapid decline near the end, with altitudes of 60-50 km over Lake Balaton in Hungary. Whether the curvature in the early part of the diagram is true or analytical error is difficult to say, although it is probably wise to assume it is analytical error.

Apart from the match in trajectory location, speed is another measure to determine whether this was the decay of 2014-074B or not. Meteors always have an initial speed larger than 11.8 km/s (but: for extremely long duration  slow meteors deceleration can decrease the terminal speed considerably below 11.8 km/s later on in the trajectory). Objects re-entering from geocentric (Earth) orbit have speeds well below 11.8 km/s, usually between 7-8 km/s depending on the orbit apogee. When speed determinations come out well below 11.8 km/s, a re-entry is a likely although not 100% certain interpretation. When speed determinations come out at 11.8 km/s or faster, it is 100% certain a meteor and no re-entry.

By taking the distance between two points on the trajectory with a known time difference, I get the following approximate speeds:

- from the Hungarian data: 7.0 km/s;
- from the Dutch data: 9.0 km/s;
- from the German data: 9.4 km/s.

These are values that are obviously not too accurate, but nevertheless reasonably in line with what you expect for a re-entry of artificial material from geocentric (Earth) orbit.

It should be noted that if the southern deviation (see trajectory results above) of the trajectory data is analytical error, the speed of the Dutch and German observations is a slight overestimation, while the Hungarian results will be a slight underestimate. This would bring the speeds more in line with each other, and even closer to what you expect from a rocket stage re-entry from Low Earth Orbit.


Discussion and Conclusions

The trajectory and speed reconstructions resulting from this analysis strongly indicate that the fireball seen over northwest and central Europe on 26 November 2014, 03:35-03:40 UT indeed was the re-entry of the Soyuz  third stage 2014-074B from the Soyuz TMA-15M launch. Although there are some slight deviations from the expected trajectory, the results are close enough to warrant this positive identification.

The deviations are easily explained by analytical error, given the used reduction method and the not always favourable configuration of the photographic and video stations with regard to the fireball trajectory. Notably, the large distance of the Dutch stations to the trajectory resulting in very low observed sky elevations and low plane fitting convergence angles for these stations is a factor to consider. Nevertheless, and on a positive note, the final result fits the expectations surprisingly well.

The data suggest that the object was at an altitude approaching 125 km (close to the expected final orbital altitude on the last completed orbit) while over southern England and the Channel, had come down to critical altitudes near 100 km while over southern Germany, and was coming down increasingly fast at altitudes of 60 km and below while over Hungary.

The Hungarian observations show that the rocket stage re-entry continued beyond longitude 19.3^o E and below 46.45^o N, and happened some time after 03:39:20 UT. It likely not survived much beyond longitude 21^o E.

The nominal re-entry position and time given in the final JSpOC TIP message for 2014-074B are 03:39 +/- 1 min UT and latitude 47^o N longitude 17^o E, with the +/- of 1m in time corresponding to a +/- of several degrees in longitude. This is in reasonable agreement with the observations.

Acknowledgements

I thank Zsolt Perkó, Szilárd Csizmadia, Stefan Schick, Jaap van 't Leven, Klaas Jobse and Koen Miskotte for making their images and data available for analysis. Carl Johannink contributed some mathematical solutions to the construction of the spreadsheet used for this analysis.

Note: another Soyuz rocket stage re-entry from an earlier Soyuz launch towards the ISS was observed from the Netherlands and Germany in December 2011, see earlier post here. As to why it takes such a rocket stage three days to come down, read FAQ here.



Literature:
- Ceplecha Z., 1987: Geometric, dynamic, orbital and photometric data on meteoroids from photographic fireball networks. Bull. Astron. Inst. Czech. 38, p. 222-234.
- Meeus J. (1991): Astronomical Algorithms. Willmann-Bell Inc., USA. 

[UPDATED] Re-entry of Soyuz third stage 2014-074B from Soyuz-TMA 15M launch observed from the Netherlands and Hungaria.

Update 23 Dec 2014: further analysis of imagery in new post
(28 Nov 2014, 10:45 UT: updated with more imagery)

click image to enlarge

Today Carlos Bella alerted the seesat list that Hungarian amateur astronomers had captured imagery of a re-entry in the early morning of November 26.

It concerns the re-entry of 2014-074B, the Soyuz third stage from the launch of Soyuz-TMA 15M which launched expedition crew 42 to the ISS on 23 November 2014.

Below is one of several casual phone-camera video's also shot from Hungaria Serbia, showing the fragmenting fireball:

(video by Aleksandar F, Belgrade)

According to the TIP message of  JSpOC, the re-entry happened near 3:39 UT on the early morning of 26 November, 2014, near 47 N,  17 E. This perfectly fits the Hungarian observations. See also the map above, which shows the predicted trajectory of 2014-074B resulting from processing the last known orbital elements with SatAna and SatEvo.

Moreover, the speed determination by the Hungarian meteor camera network, 7.4 km/s, confirms this is not a meteor but a re-entry. The speed is too low for a meteor (which are always faster than 11.8 km/s, the earth escape velocity) but matches the speed of an object re-entering from Low Earth Orbit.

Realizing that the rocket stage made a pass over the Netherlands/Belgium only minutes earlier,  I asked the operators of the DMS All-Sky meteor cameras to check their imagery of that morning. As it turns out, three Dutch All-sky stations did capture the re-entry: Bussloo (Jaap van 't Leven), Oostkapelle (Klaas Jobse) and Ermelo (Koen Miskotte).
 

Detail of the Bussloo Public Observatory all-sky image (courtesy Jaap van 't Leven)

Detail of the Cyclops Oostkapelle all-sky image (courtesy Klaas Jobse)

Detail of the Ermelo image (courtesy Koen Miskotte)

Parts of the three Dutch images (courtesy Jaap van 't Leven, Klaas Jobse and Koen Miskotte) are shown above. All stations have it very low above the horizon at elevations of 20 degrees or lower.

The Oostkapelle image shows that the incandescent phase of the re-entry already started over the UK, as the image shows the trail well to the west (and Oostkapelle is on the Dutch West coast).

As soon as I can find some time, I will analyze the imagery to see whether I can get altitude data from them. It would be nice to document the last minutes of this rocket stage in this way!

So stay tuned for an update....

UPDATE 23 Dec 2014: new post with a further analysis with trajectory and altitude reconstructions based on observations from the Netherlands, Hungaria and Germany now available. 

(I thank Jaap van 't Leven, Klaas Jobse and Koen Miskotte for permission to use their imagery)

[Updated] You Only Die Twice: the confusing end of the Russian Kosmos 2495 Kobalt-M spy satellite mission

Update 15:00 UT, Sep 11: a very brief update confirming the object was artificial is provided at the end of this post

Introduction: a spectacular fireball over the USA on September 2-3

In the evening of September 2 (in local time: early September 3 in UT), 2014, a spectacular event was seen and filmed in the skies over the southwestern States of the USA. A very slow fireball crossed the skies, seen by many casual eyewitnesses in several US States who reported their observations to the American Meteor Society (AMS). It was also captured by a number of all-sky video stations. A very nice compilation of images and what is known and what is still debated, has been made by Spaceflight101 on their website. Below is imagery of the event by Thomas Ashcraft from near Lamy, New Mexico:

video footage  by Thomas Ashcraft, New Mexico, USA

The event happened on September 3, 2014, between 4:31-4:33 UT (the evening of September 2 in local time) and was seen from Colorado, Wyoming and New Mexico. A very slow fireball, with a duration of at least 40 seconds and variable in brightness in what looks like a semi-regular pattern, moved across almost 180 degrees of sky. It penetrated deeply into the atmosphere, leaving a debris cloud at low altitude lingering for 30 minutes, detected by Doppler weather radar.

Lingering debris cloud on Doppler radar after the event (image: Rob Matson)

Initially seen as a meteor event, it was somewhat ignored by the amateur satellite community until brought to their attention a few days later.

Suspicion of a satellite re-entry

The suspicion arose that this was in fact a satellite re-entry, with the prime candidate being Kosmos 2495 (2014-025A), a Russian Kobalt-M/Yantar 4K2M photoreturn spy satellite. This is a satellite that uses analogue film rather than electronic image sensors. The exposed film is returned to earth in three recoverable return capsules, the last of which also returns the camera (for re-use).

In terms of duration, the September 2-3 event is a borderline case: with a duration of at least 40 seconds but possibly a minute or more, both a very slow 11.8 km/s meteoric fireball of asteroidal origin, or the decay of an artificial satellite are possibilities. [but see update at the end of this post: NASA camera data show it was not a meteor but an object entering from Low Earth Orbit, i.e. a satellite]

Timing and path over the sky however closely match predictions for Kosmos 2495. The observed object passed only ~3 minutes earlier than the predicted pass of the satellite, in a very similar trajectory. This actually fits with a decay, as in a lower orbit the object starts to slightly speed ahead of an object in a similar but higher orbit. The slight eastern displacement of the sky track also fits with this: in a few minutes time, the earth rotates under the orbital plane slightly, displacing the sky track westwards.

Predicted Kosmos 2495 sky trajectory for Thomas Ashcraft's site in Lamy, New Mexico. Note remarks in text about slightly eastward displacement of trajectory for a slightly earlier passing object in the same orbital plane, relative to the sky trajectory shown here

(click image to enlarge)

As this satellite should have been in earth shadow at that time of the event and hence not illuminated by the sun, it is immediately clear that if this was Kosmos 2495, it was in the act of re-entering and already producing a plasma envelope (a fireball).

[paragraph slightly rewritten 12:10 UT, Sep 11]
But why? The last known orbital element set for the satellite with epoch 2 Sep 17:12 UT show it at an orbital altitude too high for an imminent natural decay.

JSpOC however issued an "administrative decay" for the satellite early on September 3, an indication that it has been deliberately de-orbited.

Yet it was unlikely that the Russian military intended this satellite to re-enter over the USA  instead of over Russia itself, or over the Pacific.

So, if this was Kosmos 2495, did something go wrong? It initially looked like it.

Then came the confusion

Then came the confusion. On the Seesat-list, Ted Molczan reported having received reports of sightings of a re-entry earlier that same day, near 18:14 UT on September 2, seen over southwest Kazachstan. A number of video's exist of this event and show a glowing object followed at some distance by a cloud of glowing fragments.

footage from Kazachstan

The location of these observations, timing and general direction fits well with an object on a trajectory to Orenburg in Russia, the designated touchdown locality of the Kobalt-M re-entry capsules. Indeed, the timing of the observations (~18:14 UT) matches a pass of Kosmos 2495 over the area, and the trajectory of the latter indeed brings it over Orenburg near that same time.

So if this was the Kosmos 2495 re-entry over southwest Kazachstan and the Kaspian sea, then what was it that re-entered over the USA 10 hours later?

In denial

Next, the Russian military weighed in and flatly denied that anything went wrong with Kosmos 2495, implicitly suggesting that the object decaying over the USA was not their satellite (spoiler: it nevertheless likely were parts of the satellite, see below).

Multiple parts

For a solution of this confusion, we have to look at the construction of a Kobalt-M satellite, and previous Kobalt-M missions. An excellent and detailed description of the Yantar/Kobalt satellites translated from a Russian publication can be found here on Sven Grahns website.

We have to realize that the Kobalt-M satellites are made up of multiple modules:

1) The Equipment Module (AO) that contains the main power and propulsion systems;
2) The Instrument Module (PO) that contains electronic equipment necessary for the control and functioning of the satellite;
3) The camera re-entry vehicle (OSA), containing the camera and the last batch of film. This is a true re-entry vehicle, designed to survive re-entry through the atmosphere for recovery of the camera and film. The target area for these re-entry vehicles is near the Russian town of Orenburg;
4) a 2.5 meter sun shade with additional antennae and sensors on the tip of the OSA, that is presumably jettisoned at re-entry.

The satellite also has two additional small re-entry and landing capsules for the recovery of film mounted on the side of the OSA: these are jettisoned for re-entry at 1/3rd and 2/3rd into the mission, so should no longer have been present on Kosmos 2495 on September 2.

Of importance is that the OSA re-entry module eventually separates from the satellite for re-entry. This potentially leaves satellite parts in orbit after the OSA re-entry, even though it is generally believed that the AO and PO go down with the OSA, with the AO providing the retrofire burn for the de-orbit of the OSA.

Re-entry of the Kosmos 2495 OSA return vehicle observed over Kazachstan towards Orenburg at 18:14 UT, Sep 2

The event seen from Kazachstan was, given the location and timing, most likely the OSA return vehicle with the camera and film re-entering the atmosphere for recovery at Orenburg. The single object in front visible in the videos is likely the returning OSA itself. The cloud of fragments at some distance behind it, might be the jettisoned sun shade disintegrating in the atmosphere. It could also be the AO (propulsion) module, the PO module, or both (it is believed by analysts that the AO (propulsion) module is providing the retrofire boost necessary for the de-orbit of the OSA re-entry vehicle. It is believed that the OSA does not have its own retrofire rocket).

Additional Kosmos 2495 parts surviving until re-entry over the USA at 4:30 UT, Sep 3

How does this fit in with the observations over the USA 10 hours later?

A clue is provided by previous Kobalt-M missions. At the end of five of these (Kosmos 2410, Kosmos 2420, Kosmos 2427, Kosmos 2445 and Kosmos 2462) pieces of debris were detected and catalogued by US tracking facilities that survived for several hours after the OSA re-entry vehicle touched down at Orenburg. In four of the five cases, it concerns two debris pieces (the fifth case, Kosmos 2462, produced three pieces). These debris pieces had the following SSC catalogue numbers and usually Cospar sub-designations C and D, or D and E:

For Kosmos 2410: 28501 and 28502
For Kosmos 2420: 29258 and 29259
For Kosmos 2427: 32048 and 32049
For Kosmos 2445: 33969 and 33970
For Kosmos 2462: 36821, 36822 and 36823

 
Of interest is that these debris pieces are only detected at the very end of the Kobalt-M mission, around the time of the OSA return vehicle re-entry at Orenburg. They hence seem to have to do with alterations to the satellite in preparation for the OSA separation and re-entry. As it happened on at least five of the missions, it seems a normal element of these missions. In fact it might have happened on all missions, but not all might have been detected: most of the objects above have only one or two element sets released indicating short detection spans. Their lifetimes typically are no more than a few hours to a day, so they can be missed.

From the catalogued orbits of these debris pieces, there are suggestions that the separation of these objects from the original satellite body actually happens a few hours before the OSA re-entry. For Kosmos 2410, this is very clear as the debris pieces were first detected some 16 hours before the OSA re-entry, and while the A-object (presumably containing the OSA) was still being tracked.

The likely re-entry seen from Wyoming, Colorado and New Mexico 10 hours after the OSA re-entry vehicle return over Orenburg, could very well concern similar debris pieces generated by Kosmos 2495. Analogues from another Kobalt-M mission suggests this is a realistic option.

The Kosmos 2445 analogue

Kosmos 2445 (2008-058A), another Kobalt-M mission from 2009, provides a very nice analogue. On its last day of existence it produced two debris pieces with catalogue numbers 33969 and 33970, that survived for several hours after the OSA re-entry. The OSA return occured on 23 Feb 2009 at 16:15 UT. We know this because this OSA re-entry was observed, as reported by Lissov. The last available tracking data for the two Kosmos 2445 debris pieces have an epoch near midnight of Feb 23-24, 2009, indicating survival for at least 8 hours after the Kosmos 2445 OSA return at Orenburg.

I have used Alan Pickup's SatEvo software to further analyse the likely decay time for these debris pieces: the analysis suggests decay near 1:30-1:40 UT on 24 Feb, 2009. This is 9.5 hours after the OSA return.

This 9.5 hours survival time of the Kosmos 2445 debris pieces is similar to the time difference between the Sep 2, 18:14 UT Kosmos 2495 OSA return observed from Kazakhstan, and the possible decay event observed over the USA at Sep 3, 4:30 UT. The time difference between these is about 10 hours, which is not much different from the ~9.5 hours for the Kosmos 2445 debris in 2009.

During their last few orbits in February 2009, the Kosmos 2445 debris pieces C and D moved somewhat in front of where the A-object (the part including the OSA re-entry module) would have been had it not been de-orbitted. The difference in pass time was a few minutes.

Relative position of Kosmos 2445 C and D debris pieces a few minutes in front of where the A-body would have been, just before decay early Feb 24, 2009 (movement is top to bottom)

(click image to enlarge)

This again provides a nice analogue to the September 2-3 event over the USA: the decaying object observed from the USA moved along the Kosmos 2495 A-object trajectory, but passing 2-3 minutes earlier than the predicted A-object passage (i.e., it was moving slightly in front of where the A-object would have been had it not been de-orbitted over Orenburg). Also note the slight westward displacement of the A object (red) trajectory.

So: likely Kosmos 2495 debris re-entering over the USA after all!

I feel that this all justifies to conclude that what was seen from the USA on the evening of September 2-3, indeed were parts of Kosmos 2495 re-entering. The close agreement of the observed fireball track with the predicted trajectory and predicted pass times for Kosmos 2495 is too good to be likely coincidence. The whole event moreover fits patterns of previous Kobalt-M missions, notably that of Kosmos 2445 in 2009: debris pieces surviving for a few hours after the OSA return vehicle re-entry, decaying ~ half a day later.

So while it was not the return capsule with the camera and film that re-entered over the USA, it were nevertheless almost certainly parts of Kosmos 2495.

Remember that denial (see another version here) by the Russian military? Read it carefully. What they actually deny is that Kosmos 2495 exploded, and they say "that nothing out of the ordinary happened".

That is true. The return capsule separated successfully and presumably landed safely at Orenburg near 18:14 UT, as observed from Kazachstan. And Kosmos 2496 did not explode over the USA: debris parts left after the OSA separation decayed over the USA. Generation of such debris pieces seems to be normal for a Kobalt-M mission. So yes, "nothing out of the ordinary happened". It is all a clever word game.

On the nature of those debris pieces

What the nature of those debris pieces generated at the end of most (if not all) Kobalt-M missions and probably seen decaying over the US exactly is remains unclear. Behind the scenes, several independent analysts including me have had e-mail discussions about this the past 24 hours. Separation of the Kosmos into three modules (AO, PO and OSA), one of which (the OSA) makes a controlled re-entry over Orenburg for recovery, would make you think the remaining two debris pieces are these two other modules, the OA and PO. However, it is generally believed that the AO/PO combination provides the retrofire necessary for the OSA de-orbit and hence goes down with the OSA.  It is believed that the OSA module itself has no retrofire capacity (if it would have, it would separate from the other modules and then fire its own retrorocket, leaving the other two modules in orbit).

So analysts have proposed that the debris pieces instead are satellite parts like solar panels (which are 6 meters in lenght each)  and antennae shed somewhat before the OSA re-entry. That idea is more likely yet in itself not entirely unproblematic either. In the case of Kosmos 2410 in 2005, the debris pieces were generated at least 16 hours (if not more) before the OSA reentry. It seems somewhat unlikely that you shed power sources (solar panels) and communication equipment (antennae) so many hours before the OSA re-entry.

The observations from the USA on September 2-3 suggest a seizable object. This is not small debris, but definitely a large object.

So that part of the story remains a bit of a mystery.


UPDATE 1, 11 Sep 2014, 15:00 UT:

Dr Bill Cooke of the Meteoroid Environments Office at NASA's Marshall Space Flight Center informed me (and this information is posted here with his kind permission) that their camera systems catched the event from New Mexico. From the data they determined that the object entered with a speed of  7.69 +/- 0.07 km/s.

That is too slow for an object in heliocentric orbit (a meteor), but the typical speed of an object entering from Low Earth Orbit. Basically, this confirms that the event over the USA was the decay from orbit of (a part of) an artificial satellite.

I thank Dr Cooke for communicating this vital piece of information.

UPDATE 2, 15 Sep 2014, 15:30 UT: 

Ted Molczan has published an excellent analysis into the area-to-mass ratio's of past Kobalt-M debris, which compares favourable to the area-to-mass ratio needed for Kosmos 2465 debris shed at OSA separation to decay over the US at 4:33 UT.

Acknowledgement: I thank Ted Molczan, Jon Mikkel and Jonathan McDowell for the exchange of ideas. Igor Lissov provided valuable data on the Kazakhstan sightings and earlier sightings of Kobalt-M OSA re-entries from that region on Seesat.

[updated] The bright fireball over Germany of 31 March 2014, 22:34 CEST: an earthgrazing meteor, not a satellite re-entry

[updated 20:55 UT (1/4/2014) to reflect revised fireball duration]

Yesterday evening German astronomical internet fora and my Twitter timeline erupted in a frenzy about a very bright, magnitude -10, west to east moving, very long duration fireball seen over southern Germany near 20:34 UT (22:34 CEST, March 31).

The fireball was widely seen by eye witnesses and captured by a video all-sky station near Ulm. The very spectacular image, by Thomas Tuchan, can be seen here (scroll down in the message list) on the AKM message forum.

As usual, it was science writer Daniel Fischer who was the first to knock on my digital door for an opinion. The question that had popped up, as it does with every long duration slow fireball, was whether this was a meteoric fireball or perhaps a satellite re-entry? In most cases, it is not, although there are exceptions.

My first check in such cases always is with JSpOC to see whether there was a suitable re-entry candidate in the TIP-messages. There was not. This while a re-entering artificial object of this brightness must be a very big object, for which you expect a TIP message.

Next more information came available on the fireball length and duration, notably through Thomas Tuchan's all-sky video image. It shows an almost horizon to horizon event, with a duration of 16 33 seconds. It starts at approximately 15 degrees elevation in Perseus, culminates at 60 degrees North, and ends low on the opposite horizon, at an elevation of about 12 degrees. A span of some 150 degrees!

[Update 20:55 UT: the duration was later revised to 33 seconds]

The very long 150 degree trajectory with a duration of 16 seconds rules out the re-entry of an artificial object. It shows that this was a meteoric fireball, and one that entered the atmosphere at a very shallow angle: a so called Earthgrazing meteor. There are even some examples (most famous one the Grand Tetons fireball of 1972) where such Earthgrazing fireballs left the Earth's atmosphere again!

Satellite re-entries take place between 150 and 50 km altitude. At such altitudes, an earth-orbiting object has a speed of 7.5-7.8 km/s and the resulting apparent angular velocity is about 3 degrees/second for 100 km, and about 5 degrees/seconds for 50 km altitude: but only in the zenith. Lower above the horizon, the angular speed is much less.

I constructed an artificial set of orbital elements for an orbital altitude of 90 km (re-entry in progress) as a test: it takes such an object 1 minute 15 seconds to move from 15 degrees elevation above the western horizon to 15 degrees elevation on the opposite horizon. By contrast, it took the German fireball only 16 33 seconds: i.e. almost a factor two-and-a-half faster.

[Update: the duration was first reported as 16 seconds, later revised to 33 seconds]

All this makes very clear that the German fireball of March 31 was not the re-entry of an artificial object, but a meteoric fireball, most likely an Earth-grazing object of asteroidal origin..

Guest post at ESA's Rocket Science blog

On request of one of the editors I have written a long guest post for ESA's Rocket Science blog titled:

  "Predicting GOCE re-entry: a citizen- scientist’s view"

The post details how I tried to forecast GOCE's re-entry time and position, using Alan Pickup's SatAna and SatEvo software. It provides some information about what factors are involved, and what problems you bump into. Basically, it is a consolidation and extension of posts that earlier appeared on this SatTrackcam blog.

Read the post on ESA's blog here.

GOCE re-entry photographed from the Falklands?

This has just appeared on Twitter:

@BBCAmos We saw GOCE satellite burn up over East Falkland at about 9.20pm last night. pic.twitter.com/KpFGAYALkY
— Bill Chater (@Cheds23) November 11, 2013

Reported time and geographic location seem a match (21:20 Falkland time is 00:20 UT)!

Alas, poor GOCE, I knew him well...

click map to enlarge

Last night just after 0h UT, GOCE, ESA's Gravity Field and Steady-State Ocean Circulation Explorer, died an heroic death, plunging into the atmosphere while passing over the ice cold wastes of Antarctica, within minutes of passing over the Falkland islands.

ESA reported the decay time as "close to 01:00 CET on Monday 11 November" (= close to 00:00 UT, Nov 10-11).

USSTRATCOM gives a final TIP placing decay at 11 Nov 00:16 UTC +/- 1 m near 56 S 60 W.

My initial last pre-decay forecast, made in a haste late last evening after returning from a full day surveying in the field (later more on that...), was too early.

This was before the final few orbits for GOCE were published, and before I learned from Alan Pickup of a secret setting in SatAna and SatEvo that makes it possible to tweak details that are important in the last few orbits at very low altitude. My tweet at that time:

Reentry forecast #GOCE: Nov 10, 22:10 UT +/- 25 min #GOCEreentry
— Dr Marco Langbroek (@Marco_Langbroek) November 10, 2013

As this window was including a pass over Australia, I also tweeted:

Observers in mid-Australia: watch if #GOCE is perhaps reentering above you NOW @drspacejunk
— Dr Marco Langbroek (@Marco_Langbroek) November 10, 2013

ESA however next reported having received telemetry from a GOCE pass at 22:42 UT from Troll station on Antarctica, making clear GOCE was still alive and functioning while only just above 110 km altitude!

So my 22:10 UT forecast was wrong. We now know it was wrong by an hour two hours, actually.

Alan Pickup mailed me around that time about some 'hidden' experimental options in SatAna and SatEvo that take into account spacecraft dimensions and some dimension-related effects that are significant at very low altitudes only.

Together with the addition of two more orbital updates that have since appeared, I have therefore re-done the exercise, as an "aftercast".

With solar flux at 154, a 0.3 day tle arc (the last 5 available orbit updates) processed in SatAna and the result then fed into Satevo, and setting the length of GOCE at 5.0, I get re-entry at:

11 Nov 00:13 UT +/- 14 m
69 S, 52 W

This is only 3 minutes from the time given by USSTRATCOM.

In the map on the top of this post, the blue dot gives the USSTRATCOM position, the red dot and red line give the SatAna + SatEvo nominal  prediction and window.

Below is the SatEvo result in 3D, looking towards the south polar region:

I am rather surprised about how well (after tweaking some internal settings) the final SatEvo result compares to USSTRATCOM's final TIP. Kudo's to Alan who wrote the software! (of course, and Alan agrees, the near-perfect match can be a lucky coincidence).

The diagram above shows how quickly GOCE dropped in the end. The last available orbital elements from an epoch about an hour before reentry, are for a perigee altitude of only 110 km! A day earlier the perigee was still at 150 km altitude.

One of the most amazing things about the re-entry of GOCE is that the spacecraft retained its drag-reducing attitude right up to the end. The designers of the spacecraft deserve some serious kudo's for that.

Of all the ways a spacecraft can go, GOCE died gracefully and heroically!  GOCE, clutching on to life to the bitter end, victim of the same forces that it helped map in so much detail. Now let us mourn our brave little spacecraft...

R.I.P.
GOCE

(17 Mar 2009 - 11 Nov 2013)

(here imaged 1.5 months before it's re-entry)