Re: Question on Elset Corrections

Björn Gimle (
Mon, 15 Mar 1999 08:34:36 +0100

>I see frequent elset updates using observations that apparently
>use Elcor.  Given a reasonable elset and one good obs, how does
>one determine which parameter should be changed?  Is it an
>iterative process..changing each one until the residual is the
ELCOR tells you how many elements you can compute based on the number of
observations you have. For each element you choose, you must give a
reasonable delta (uncertainty/computational step), and you will see how
much effect each delta has on each position. If you picked reasonable
deltas, you should normally compute those elements that have the largest
effect on the observation(s).

If the (latest) elset is the end of a normal OIG sequence, you may check
that they are stable (e,i) and generate similar predictions. As you say,
the ones most likely in error are MM and drag (Bstar, the ndot2 and
ndotdot6 are computed from it). The effect of a delta in MM is delta*age
(in days), in ndot2 it is delta*age*age. Unless the elset is very recent,
or the drag very low, you would usually need an unreasonably large deltaMM
to achieve the effect of a (20%?) delta in ndot2. The Bstar delta should
be proportional to the ndot2 delta.

If the elset is very old, not verified against other elsets near that time,
generated after a maneuvre, subjected to a maneuvre between the elset and
the observation, and/or drag has changed considerably in the interval, all
elements will be affected. The SGP4 model assumes a regular variation. If
this is not true, the elset accurate at the time of observation would
extrapolate backwards to the time of the known elset into something quite
different. Thus, the average MM would be different, and the precession of
the perigee and nodes.

If the elset variation may be smooth, but the balance between ndot2 and MM
deltas is unknown, you can generate a few predictions around the time of
the epochs of the good elsets, in various directions and/or for different
locations, and use these as additional observations to determine more or
all of the elements.

If a maneuvre is suspected, I use VEC2TLE to generate an elset at the time
of the maneuvre, or adjust ndot2/B* to give a reasonable average MM for the
time until the observation and generate an elset for that Epoch.
Then I correct MM and ndot2/B* of the computed elset to expected values at
that epoch, and use it as a base for ELCOR.

-- (office) (home)      --
--                   --
-- COSPAR 5918, 59.298 N, 18.104 E, 55 m                     --
-- SeeSat-L / Visual Satellite Observer Home Page found at   --
--       --

for what