[IGSMAIL-5252]: BRD orbit accuracy results

[Jim Ray (NGS 301-713-2850 x112)]

******************************************************************************
IGS Electronic Mail      10 Nov 08:18:01 PST 2005      Message Number 5252
******************************************************************************

Author: Jim Ray

With respect to IGS Mail #5209 by Jim Slater, and for other reasons, it
is perhaps interesting to visualize the recent performance of the GPS
broadcast orbits.  The following plots (also in ps format) are available
at ftp://www.ngs.noaa.gov/dist/jimr/brd/ :

 brd-wrms.pdf -- RMS orbit residuals (1-D)
 brd-tz.pdf   -- Tz translation of orbit constellation
 brd-rz.pdf   -- Rz rotation of orbit constellation
 brd-clk.pdf  -- RMS clock residuals

all with respect to the IGS Rapid products (IGR) and shown since 1 Jan 2002.
Plotted values are taken from the daily combination reports generated by the
IGS Analysis Coordinator.  Note that the IGS normally includes all
broadcasting satellites, regardless of official health status.  However, if
any AC fails to submit a solution for a particular satellite, then that
satellite is not included in the statistics.  So, some of the poorer orbital
results are due to satellites that were set unhealthy but tracked by the IGS
anyway.

Orbit residuals -- These are the RMS differences after removing a daily
7-parameter Helmert transformation, compared to the IGR orbits.  There has
been a steady (nearly linear) improvement over the past 4 years, from about
2.5 m RMS to 1.3 m RMS now.

Axial translation -- There has been a longstanding annual oscillation of the
GPS constellation with respect to the equatorial plane, with an average
overall bias.  That annual trend continues, but with a decreasing amplitude
and the north-south bias has gradually faded.

Axial rotation -- Rotations about the Z axis are unavoidably connected with
the quality of predictions of UT1 variations.  The range of GPS Z-rotations
has decreased significantly over the past 4 years, esp in the last year.
However, recent rotations, while smaller, are temporally more coherent
(except possibly in the very latest results).

Clock residuals -- Broadcast clock predictions have also improved over the
past 4 years, from about 4.5 ns RMS to around 3 ns RMS now.  Most of the
improvement seemed to occur around late 2003 and has relatively stable and
unchanged since then.