Re: TLEs for ASTRA 1F
Bruno Tilgner (Bruno_Tilgner@compuserve.com)
Fri, 21 Nov 1997 10:24:40 -0500
Hi Jay,
>I've seen many sync sats, but there are also many I have not seen when I=
=
>looked for them. I thought it was because they were too faint. Now you=
=
>indicate the elements may be wrong.
In at least one case, ASTRA 1F, I am absolutely certain they ARE wrong,
because the position is 6 degrees to far east. This is a very clear-cut
case. For the others 5 members of the ASTRA family the error, if any,
is much smaller. I have mustered two teams, one in southern Switzerland
with a Celestron 8 and CCD camera, the other in Germany with a rather
fast and powerful Schmidt camera to take more photos.
Both techniques should allow to determine positions to better than 1 arc
minute. This is the same order of magnitude that NASA GSFC achieve on a
routine basis with their multi-million $ equipment.
The photos will be taken as soon as possible. But at this time of the
year we have mostly clouds (you remember driving to EUROSOM last year).
>What prediction program are you using?
I am using my own programs, i.e. I am going right into the NORAD SDP4
and SGP4 models, so I know what I am doing. Any errors would be due to
the NORAD models themselves; not necessarily the algorithms but the
way they have been coded. At least one error, which occurs under
certain conditions only, has been detected. I have been in touch with
Rob Matson and Dave Cappellucci. Their programs (Skymap and SatSpy
respectively) have been corrected, but only a far as this known bug is
concerned. Although I have the source code, it is absolutely impossible
to check it. The SGP4/SDP4 subroutine alone is 15 1/2 pages long and
written in a very awkward programming style with lots of GOTOs which
are difficult to follow. So there may be more hidden bugs.
In fact the exercise about the ASTRAs started in order to discover such
bugs. I found large discrepancies between the positions photographed
and those calculated. I got practically no response from SEESAT when I
posted a first message about 2 to 3 weeks ago, but I also posted the
message to the CompuServe Astro Forum. Someone from NASA GSFC replied
and sent me state vectors (position and velocity) for the TDRS 4 satellit=
e
which have been derived completely independently from USSPACECOM.
At least for this satellite the results I got with the NORAD model were
not too bad; the maximum position error was 15 km compared with the
GSCF position which is accurate within 100 meters. This would lead to
an angular error of just over 1 arc minute which is something one could
live with.
At this point I began to suspect the TLEs from USSPACECOM. It could well
be they don't actully measure the objects but calculate their positions,
and in this way systematic error might go by unnoticed for a long time.
I wonder anyway how USSPACECOM measure positions of satellites which are
not visible from the US or any US possession. There used to be a radar
station in Great Britain and one in Saudia Arabia. Maybe there are still
in use, maybe not.
As soon as the weather improves, I hope to get images from which the ASTR=
A
positions can be determined and then we'll see how they compare with
computed positions using TLEs. I am also trying to make contact with the
flight dynamics experts of the company in Luxemburg which owns and
operates the ASTRA spacecraft.
Regards
Bruno