ASTRA 1F TLE Problem

Bruno Tilgner (Bruno_Tilgner@compuserve.com)
Sat, 15 Nov 1997 07:19:44 -0500

Hello everybody,

The following 6 satellites make up the cluster of the ASTRA direct TV
broadcast satellites:

Name       COSPAR   CATALOG
---------------------------
ASTRA 1A   88109B   19688
ASTRA 1B   91015A   21139
ASTRA 1C   93031A   22653
ASTRA 1D   94070A   23331
ASTRA 1E   95055A   23686
ASTRA 1F   96021A   23842

These satellites are co-located at a nominal longitude of 19.2
degrees east, so they can all be captured with a single dish.

During an ongoing photographic project of recording the short-term
movements of these satellites and their light curves, persistent
problems were encountered to determine the position of ASTRA 1F.

In fact, recent orbital elements show it several degrees further
east so that it would be impossible to appear on the same image
with the other 5 satellites. During the last few weeks a number
of photos were taken on film and with CCD cameras by three different
observers on different dates and at different locations.

All pictures (several hundred) show 6 objects, whereas the TLEs used
would allow only 5, given the field of view of about 0.5 degree
by 0.5 degree.


I have checked older TLE sets and found the following evolution
of the position of ASTRA 1F:

  Date     Longitude
--------------------
17 Dec 96   19.24 E
28 May 97   19.18 E
07 Jun 97   19.29 E
03 Jul 97   21.19 E
10 Jul 97   21.81 E
18 Jul 97   22.61 E
02 Oct 97   21.78 E
16 Oct 97   23.14 E
30 Oct 97   24.95 E
10 Nov 97   26.58 E
--------------------

Whilst all other members of the ASTRA family have stayed near their
nominal position at 19.2 E, ASTRA 1F seems to have drifted by more
than 7 degrees during the summer, starting some time in the second
half of June.

This is a totally unrealistic result which can only be attributed
to faulty TLEs. Can anybody offer an explanation why the TLEs for
just one of these 6 satellites should get out of hand by so much ?

On a related subject: The SDP4 code in two FORTRAN and one PASCAL
implementations give the position of geosynchronous satellites
with enough precision so they can be found in a telescope. However,
the code is not capable of correctly computing short-term movements,
in the order of one to several hours. To this end, an accuracy of
about 1 arcminute would be necessary. The ASTRA satellites offer
a unique opportunity (at least from the eastern hemisphere) to
observe these movements in real-time and to compare them with
computed positions. The SDP4 model tends to result, at least for
these satellites, in a larger eastward drift in longitude than is
actually observed.


Bruno Tilgner
Saint-Cloud, France
Bruno_Tilgner@compuserve.com

 =