Re: Deviations of satellites from their orbits

On 18 Apr, 19:28, T Driver <drive...@xxxxxxxxx> wrote:
On Apr 17, 4:51 pm, claudegps <claude...@xxxxxxxxx> wrote:



On 18 Apr, 00:31, Junoexpress <MTBrenne...@xxxxxxxxx> wrote:
[cut]

Thank-you for all of the replies. Yes, I came across the JPL paper in
which they claimed they had reduced the error to something within the
cm range, if I recall correctly. I appreciate the comment about the
complexity of their analysis, as that is something I will have to
consider.

I have compared JPL prediction with post-processed precise ephemeris:
the predictions are impressive!
Very high precision. They are able to predict orbits for some days
mantaining an error of few meters.

For the time being, however let me try to summarize what I think has
been said and get some final input on this:
1) It is safe to say that an error on the order of a meter or two is
an average estimate for the error from the predicted error in the
orbit.
2) The source of this error is generally unspecified

I think that the source of error is generally the approximation
required to use orbital data that does not describe correctly the
satellite movements.
It's like tryng to describe a polinomial function with too few
factors.

and I have one final question: the satellites themselves are tracked
and the orbits corrected when they "stray" too far from their orbits.
Does anyone know the error tolerance used to decide a satellite is
unhealthy and needs its orbit corrected (or is that classified)?

Delta-V manuevers (I think) are scheduled about once per year, mainly
because of gravity.
But I don't know if there is a particular tolerance to be respected.

Actually, Delta-Vs are used to keep the GPS satellites in a specific
orbital slot. These slots are defined to provide the optimal
constellation geometry for DOP coverage. New orbit predictions are
uploaded about once per day. This corresponds to an orbital
prediction error of roughly 4 meters, according to the graphs I
pointed to earlier.

I think that the upload is much more frequent than once per day,
because the ephemeris data are valid only for 2 hours after they are
broacasted.
With once-per-day upload you can't have data valid for few hours...
Otherwise the system does not work.
The almanac can be updated once per day without problem, as it least
more than a week (but can't be used in the navigation solution)

The graphs are interesting, but unclear(to me :) maybe I'm missing
something).
What does the Y represents? Error in ECEF, LOS, or radial? (I think
LOS, but I don't find a reference)
How old is the ephemeris data used?


.