Re: One more iteration

Bjoern Gimle (b_gimle@algonet.se)
Mon, 1 Sep 1997 07:05:31 +0200

>>Thanks to Pierre Neirinck, Rainer Kracht, Leo Barhorst in providing and 
>>refining the below elements not provided by OIG. I thought I would add the 
>>common names for those subscribers new to this interest. 
>>These and other elements are normally found in the molczan files and Mike 
>>McCants classfd.zip files (see 
>>http://www.satellite.eu.org/tletools.htm for links)
>>
>>USA 86
>>1 22251U 92083  A 97240.19540420  .00000300  00000-0  00000-0 0    10
>>2 22251  97.7578 325.4318 0427657 322.6660  34.2350 14.54941526    10
>>
>>USA 116
>>1 23728U 95066  A 97239.21841770  .00008237  00000-0  00000-0 0    10
>>2 23728  97.9298 349.3098 0541164 289.1420  64.8679 14.76958950    10
>>
>>USA 129
>>1 24680U 96072  A 97239.87378010  .00011115  00000-0  00000-0 0    10
>>2 24680  97.7741 299.7444 0541940  73.9433 291.9555 14.76543125    10
>>
>
Pierre's elsets do not contain the BStar term, so SGP4 model programs
will not work well. 
The error will increase with the square of time, reaching about 2 min.
after two weeks!

Some programs may remedy this by replacing zero values of BStar
by an internally computed approximation from the (non-zero) ndot2
(or v.v.). 

Vec2TLE does this when a value is set to -1:
USA 86                                     1018 * 414 km
1 22251U 92083  A 97240.19659617  .00000300  00000-0  19556-4 0    28
2 22251  97.7578 325.4329 0428678 143.2670 219.8741 14.54801163    20
USA 116                                    1020 * 271 km
1 23728U 95066  A 97239.21996843  .00008237  00000-0  74700-4 0    27
2 23728  97.9298 349.3113 0543315  20.8734 341.3758 14.76614315    28
USA 129                                    1036 * 258 km
1 24680U 96072  A 97239.87364216  .00011115  00000-0  97409-4 0    28
2 24680  97.7741 299.7443 0552440  55.0351 310.1462 14.75222951    21

Of course we are very much indepted to Pierre, Russell, and others, for
making so many accurate observations and orbit determinations of
classified objects, so other more lazy observers can follow them and
detect orbital maneuvres etc.

But Pierre's elset comments raise another issue:

96-72A     Pierre Neirinck's elset #110     mean dev 1.7'

It appears from this, and previous postings, that Pierre more or less
computes an elset after each observation, using only a few days of
earlier obs. This makes the elset fit to almost arcminute accuracy,
but ndot (Bstar) is determined with low precision.

Such suites of elsets would be interesting to orbital analysts, if
it was made available in their entirety, making repeated fit to
observations unnecessary.

But for these kinds of objects, with about four months between each
maneuvre, and quite stable orbits, I think the observing community
is much better served by elsets that fit all observations from the
most recent maneuvre, maybe within 0.5 degrees and 1 minute, and
with possibilities for that accuracy to keep until the next
maneuvre. 

Only short notices of prediction error from new observations would
be sufficient to detect when a maneuvre has taken place, and a new
elset is required.

My recent attempt for USA 129 was computed from a few of Russell
Eberst's and Peter Wakelin's obs July 28-Aug.25
USA 129         15.0  3.0  0.0  5.1 v            1019 x 266
1 24680U 96072  A 97209.91202734 0.00012715  00000-0  13180-3 0    04
2 24680  97.8544 270.5945 0535888 171.6322 188.0471 14.75793218    06

------------------------------------------------------------
-- b_gimle@algonet.se              +46-8-7428086  (home)  --
-- 59.22371 N, 18.22857 E, 44 m     AND member of :       --
-- http://www.algonet.se/~b_gimle   SeeSat-L              --
-- bjorn@tt-tech.se                +46-8-59095783 (office)--
------------------------------------------------------------