Re: some basic Q's

Allen Thomson (
Wed, 15 Sep 1999 08:31:23 -0500

At 08:22 15/09/1999 , wrote:

> How often are elements updated by NORAD?

And Tony Beresford replied:

>sometimes up to twice a day on LEO objects that cross the SPASUR radar
>like the ones you mention Frank.

This question has come up several times over the years and, thanks largely
to Tom Kelso's insights, it's developed that the relationship between
published TLEs and new observations isn't at all straightforward.  As one
would expect, TLEs are generated as the output of a statistical filter that
incorprates orbital models and weights older observations according to
expected error growth.  A point to keep in mind is that the filter doesn't
really need new input to produce new TLE's -- it just keeps running using
its internal models and the latest available observations.  And, of course,
there is the occasional bit of human intervention when the filter gets into
trouble.  So one shouldn't assume that there is a strict correlation between
the appearance of new TLEs and NORAD's receipt of new observations.

Nonetheless, the process generally works pretty well, as the fact that we
see satellites more or less where they should be testifies.  However, it's
clear that  NORAD's routine orbital models don't or haven't -- the models
seem to change as time goes on -- take into account some effects (my pet
object, 17589, is a good example of this). Also, of course, there are
unpredictable effects due to fuel venting, solar activity, etc.

It would be interesting to get someone who works there to provide current
details about how they do business.

(I think this thread is moving toward UseSat territory.)