After the recent discussion on this site about the posted updated version of the SGP4/SDP4 code I decided to run a few tests on my own. As ESOC Flight Dynamics posts state vectors and TLEs for their own satellites on msss.esoc.esa.de and assuming that they know best where their satellites are this could serve as a reference to check SDP4 calculations of Spacetrack TLEs against. A test I ran with XMM (1999-066A; 19500km x 101600km, 53.5 incl) showed a marked improvement of the updated SDP4 code vs. the old one I had used (which was essentially based on the original Report #3 with only some modifications). The error stayed within 15 km over two orbits (4 days) which for this kind of orbit I would find acceptable. With my old version it was 10 times as much. Moving on to INTEGRAL (2002-048A; 12800km x 149800km, 82.5 incl) produced a surprise. For epoch 13-NOV-06, 16:00 UTC there were TLEs available from Spacetrack as well as a state vector and TLEs from ESOC. Avaluating right at that epoch I found a whopping difference of 7190 km between the state vector and the respective Spacetrack TLEs! The ESOC TLEs agreed with the state vector if I used the SGP4 rather than the SDP4 algorithm (with SDP4 I would get a difference of 260 km - this simply means that ESOC generates TLEs via the standard SGP rather than the SDP4 so SGP is what you have to use to evaluate their TLEs). The reason for this dismal performance does not seem be the SDP4 algorithm but rather the underlying TLEs. Spacetrack posted TLEs for XMM, INTEGRAL, CXO and a few other satellites on these high eccentricity orbits always show epochs at the exact hour. My guess is that US Spacecom tracks these satellites only sporadically and in the meantime generates TLEs via numeric propagation of their last tracking. I was told that ESOC recently had done some orbit maneuvers with INTEGRAL (but none during the last two weeks) so it is possible that the Spacetrack TLEs of three days ago were actually based on tracking which was at least three weeks old. Of course all those sats are not really on the most observed list so the relevance of all this for SeeSat may be limited. Gerhard HOLTKAMP Darmstadt, Germany ------------------------------------------------------------------------- Subscribe/Unsubscribe info, Frequently Asked Questions, SeeSat-L archive: http://www.satobs.org/seesat/seesatindex.html
This archive was generated by hypermail 2b29 : Thu Nov 16 2006 - 16:23:56 EST