Gerhard HOLTKAMP wrote: > They seem to boost their orbit in small increaments on a daily basis > as long as the Shuttle has enough fuel to spare. This proves > frustrating for people like me who try to figure out how the chances > are to see a sun transit on the weekend when the ISS and Discovery > are about to separate NASA's current set of predicted elements now appear to take into account these small boosts, as evidenced by the seemingly erratic evolution of mean motion of successive TLEs: some days increasing, other days decreasing. Each TLE yields an accurate position near its epoch, but we need a way to account for manoeuvres between successive TLEs. As Jim Oberg reported (http://www.msnbc.msn.com/id/8814642), the unusual boosts are due to thruster firings, thus the small increments that you noted, Gerhard. This suggests a way to improve the prediction accuracy of NASA's TLEs: model the small boosts as a pseudo-drag term. Unfortunately, NASA's predicted elsets contain only approximate drag terms, which are seldom revised. Lately, for ISS they are: ndot/2 = 0.00016071 rev/d^2 B* = 0.12697e-3 er^-1 I have computed more realistic values for the appended versions of NASA's TLEs. I computed the ndot/2 of each TLE as the mean value to the subsequent TLE. For example, for the epoch 05214.82045700 elset: ndot/2 = 0.5 * (15.71559505 - 15.71587498) / (05215.83786716 - 05214.82045700) = -0.00013757 rev/d^2 It is negative because the boosts will raise the orbit between those two epochs. Of course, ndot/2 is of no use to the vast majority of us who use the SGP4 orbital model, which uses the B* drag term. Some years ago, I derived a formula which accurately relates B* and ndot/2, which I used to compute the B* in the appended elsets: http://www.satobs.org/columbia/SGP4decayterms.rtf As a test, I used my version of the epoch 05214.82045700 elset to compute the position of ISS at the epoch of the subsequent elset, and found agreement to within a fraction of one second (0.6 s early). NASA's B* would have resulted in a 2.4 s early prediction. Same test of my epoch 05216.79167367 version agreed to within 0.02 s of the subsequent elset's epoch. NASA's version would have been 3.2 s early. Of course, the thrusters' small boosts probably occur sporadically, and are not of equal magnitude, so modelling their effect as linear is imperfect, but hopefully most of the time it will be better than nothing. One important note of caution: since my decay terms look forward to the next TLE's epoch, the appended TLEs should not be used for predictions earlier than their epoch, except when high precision is not required. Ted Molczan The following TLEs are as downloaded from NASA.gov on 2005 Aug 03 UTC, except for heir decay terms, which have been modified as described above. http://spaceflight1.nasa.gov/realdata/sightings/SSapplications/Post/JavaSSOP/orb it/ISS/SVPOST.html 1 25544U 98067A 05214.82045700 -.00013757 00000-0 -10457-3 0 9005 2 25544 51.6445 22.1174 0002239 210.6479 149.4552 15.71587498 1151 1 25544U 98067A 05215.83786716 .00014800 00000-0 11263-3 0 9011 2 25544 51.6439 16.9117 0001998 211.5799 148.5243 15.71559505 1311 1 25544U 98067A 05216.79167367 -.00038477 00000-0 -29248-3 0 9023 2 25544 51.6437 12.0311 0001901 219.3452 140.7572 15.71587737 1461 1 25544U 98067A 05217.80909304 .00016540 00000-0 12615-3 0 9037 2 25544 51.6446 6.8253 0001776 221.7200 138.3827 15.71509442 1626 1 25544U 98067A 05218.82653588 -.00024030 00000-0 -18301-3 0 9044 2 25544 51.6440 1.6203 0001787 230.1338 129.9667 15.71543100 1780 1 25544U 98067A 05219.84397518 .00013715 00000-0 10467-3 0 9057 2 25544 51.6440 356.4145 0001465 233.5800 126.5227 15.71494202 1941 ------------------------------------------------------------------------- Subscribe/Unsubscribe info, Frequently Asked Questions, SeeSat-L archive: http://www.satobs.org/seesat/seesatindex.html
This archive was generated by hypermail 2b29 : Wed Aug 03 2005 - 23:32:33 EDT