RE: FW: Forced reentry of Cosmos 2495

From: Ted Molczan via Seesat-l <>
Date: Fri, 12 Sep 2014 23:02:56 -0400
Yesterday, Marco Langbroek brought word of the vital clue that the fireball moved at orbital velocity, enabling the
meteor hypothesis to be discarded:

Based on that news and the information I will present in this message, I am now confident that the object that
re-entered was uncatalogued debris of Cosmos 2495 (14025A / 39732).

1. Ballistic Properties

In my last post, I reported that a hypothetical piece of debris jettisoned at the approximate time of the de-orbit burn
(2014 Sep 02 17:40 UTC), made to decay near the time the fireball occurred, would have passed at about the correct time,
and in reasonably good agreement with the available trajectory data. Below is the TLE that I created:

1 70001U          14245.73611113  .13400000  00000-0  78682-2 0    01
2 70001  81.3770 299.0950 0034344 130.9033 134.4989 16.17169265    09

Subsequent analysis enables me to report that the corresponding ballistic properties would have been similar to those of
several other pieces of catalogued debris from previous launches of the same model of spacecraft.

1.1 Area to Mass Ratio

The area to mass ratio (A/m) that yields reasonable agreement with the time and trajectory of the fireball observed over
the U.S. on 2014 Sep 03 UTC near 04:33 UTC has been estimated using GMAT R2013a (General Mission Analysis Tool),
"developed by a team of NASA, private industry, public, and private contributors".

The analysis was performed using GMAT's Dormand-Prince 78 numerical integrator, with a four degree, four order gravity
field, and the MSISE90 atmosphere model, with actual space weather data entered manually. The above TLE was converted
for GMAT propagation using TLE Analyzer 2.12. Assuming co-efficient of drag (Cd) of 2.2, the analysis yielded A/m =
0.0773 m^2/kg.

USSTRATCOM (U.S. Strategic Command) has catalogued 12 pieces of debris from five of the previous eight Kobal't-M
missions. TLEs are available for eight of those objects. The TLEs of four objects were sufficient in number and time
span to reasonably estimate A/m using GMAT.

GMAT was used to propagate the orbit of one TLE to the epoch of a later TLE. Typical spans were several hours. The A/m
was adjusted by trial and error to obtain the closest match between the propagated and actual perigee and apogee of the
later TLE, near its epoch. Cd was assumed to be 2.2. The resulting A/m in m^2/kg are summarized below:

Cosmos   SSN    COSPAR      A/m
 2410   28502  2004-038D  0.0655
 2420   29258  2006-017D  0.0686
 2445   33969  2008-058C  0.0790
 2445   33970  2008-058D  0.0685

A larger sample could be wished for, but it seems reasonable to conclude that the hypothetical debris of Cosmos 2495 had
similar ballistic properties to the debris shed prior to landing by earlier spacecraft in the series. These values are
much greater than those of large spacecraft or rocket bodies, but typical of small, low mass debris.

2. Time of Debris Jettison

A significant unresolved question was the time of jettison of the debris. Since most such debris from past missions was
first detected on or about the day that the spacecraft landed in Russia, it was clear that jettison must have been prior
to the de-orbit burn, but the proximity to the burn was unclear.

To determine the approximate time that the debris was shed relative the de-orbit burn, I attempted to perform
conjunction analyses between debris and parent spacecraft. In the case of the first mission, 04038A / 28396, no special
analysis was required to see that both pieces of debris were shed nearly one day prior to landing. This seems to have
been the exception. On later missions, separation occurred closer to the de-orbit burn, but the relative time could not
be determined with precision.

Due to the combination of low orbit and high A/m, it was impossible to obtain close, reasonable-looking conjunctions. In
a few cases, SGP4 yielded a not too terrible-looking result, that turned out to be physically impossible because it
would have occurred later than the time of the de-orbit burn. 06017D / 29258 yielded the best result. Conjunction to
within 0.95 km was obtained, but ~42 min. after de-orbit; however, even at de-orbit, the along-track separation was less
than 15 km (a couple of seconds of time at orbital velocity). At de-orbit there was several kilometres of altitude
mismatch, but ~15 min., before de-orbit, the altitudes nearly matched, and the along-track difference was less than 17

GMAT was tried with a couple of the cases for which it had been used to determine A/m. The propagation was simply
performed in reverse, and positions manually compared with those of the parent object. In the case of 06017D / 29258,
the along-track difference reached a minimum of ~15 s near the time of the de-orbit burn. For 08058C / 33969, this
difference was ~34 s, and was still trending to a minimum hours after the landing!

Although neither approach could reveal the time of separation with complete confidence, allowing for the unavoidable
along-track errors due to unknowable and/or un-modelled variations in rate of decay, there is no reason not to believe
that separation occurs at or shortly before the de-orbit burn. However, in view of the uncertainty and the exceptional
case of 04038A / 28396, I analyzed an additional hypothetical Cosmos 2495 case.

If the debris had separated one revolution prior to the de-orbit burn, how would this have affected the correlation of
the propagated trajectory with the fireball in question? I created a new hypothetical debris TLE based on separation on
2014 Sep 02 at 15:18 UTC. With GMAT, I found that for Cd of 2.2, an A/m of 0.063 m^2/kg yielded the best match to the
fireball trajectory, but just about 1 min. too early. This seems to suggest that the separation could not have been much
earlier than the time of de-orbit, certainly not ~90 minutes earlier; however, a firm conclusion requires more complete
trajectory data than currently available, which I believe will be forthcoming eventually. This would be a good point to
discuss the observational data and trajectory analysis results in greater detail.

3. Estimated Re-entry Trajectory Compared With Observations

The analysis in my previous post was based on data from a meteor camera located in Denver, operated by the private
Cloudbait Observatory. Before proceeding with the present analysis, I e-mailed the observatory's owner, Mr. Chris L.
Peterson, seeking his advice regarding the reliability of the data, specifically from the camera named DMNS, which I had
been using. Chris responded promptly, and informed me that when he began to look into this case, he discovered that the
internal calibration database for that camera had become corrupted some time ago. Also, the time is consistently fast by
over a minute. He advised that the azimuth accuracy is close, but could be off a couple of degrees. The altitude data is
unusable. It is unclear to me whether or how much of the data may be salvageable. Fortunately, a second camera (named
Cloudbait), located at Guffey, CO., was working properly. That data has yet to be fully reduced, but Chris informed me
that from Guffey, "the event was recorded from 043213 to 043241 [UTC], and those times are accurate to the second. The
maximum altitude observed at Guffey was 57 (az 100) at 043225."

For the GMAT analysis discussed in this report, I attempted to match the maximum altitude data for Guffey when I
propagated the hypothetical debris orbit to final decay. The trial and error process that yielded the aforementioned A/m
of 0.0773 m^2/kg, crossed azimuth 100 deg at 04:32:23 UTC, at 57.8 deg elevation.

The observer who filed AMS Fireball Report #1973p from Alamogordo, NM, was observing Comet Jacques in 15X50 binoculars
when the fireball entered the field of view. The estimated re-entry trajectory passed within 2.6 deg of the comet at
04:31:00 UTC, probably just outside the FOV, assuming the comet was centred. The estimated altitude above the geoid was
~102 km - close to the common rule of thumb of 100 km for initial self-illumination (I normally assume 96 km).

As shown on the following plot, the estimated re-entry trajectory is a good match to the locations of known sightings,
which span about 1,600 km:

The fireball began over New Mexico, at about 100 km altitude. By the time it reached the northernmost observers, in
Wyoming, South Dakota and Montana, it had descended to about 70 km. In the unlikely event that any debris reached the
ground, the estimated toe of the footprint would have been near 45.40 N, 103.38 W, at the northernmost end of the
plotted ground track.

The blue streak that parallels the ground track over northern Colorado and parts of Wyoming and Nebraska depicts the
suspected Nexrad radar detection that first appeared about 45 min. after the re-entry. The following plot provides a
closer view:

I have almost zero knowledge of the subject, so am unable to comment on the relevance of the Nexrad data at this time.

The following Excel spreadsheet contains the estimated re-entry trajectory, and can be used to generate an ephemeris for
any location at 5 s intervals:

The first sheet contains the ephemeris generator. Simply enter the observer's coordinates in the input cells at upper
right, and the ephemeris is automatically computed. The second sheet performs no calculations, but displays ephemerides
for a selection of sites that reported observations, mainly ones in the AMS fireball reporting database. The main
selection criterion was to provide a wide geographic sample. The final sheet summarizes all 38 of the AMS sightings
available at this time, including site coordinates that could be used to compute an ephemeris. Full details are
available at the AMS site:

I look forward to analyzing additional meteor camera and other data when it becomes available. I would also like to try
to determine whether the object could have been a solar array and estimate its approximate mass.

Ted Molczan

Seesat-l mailing list
Received on Fri Sep 12 2014 - 22:04:24 UTC

This archive was generated by hypermail 2.2.0 : Sat Sep 13 2014 - 03:04:24 UTC