Date: Fri, 02 Jan 2009 08:39:17 -0500 From: "Jim.Ray" Subject: [IGSMAIL-5874]: status of IGS Ultra-rapid products To: IGSMail Reply-to: jim.ray@noaa.gov Message-id: <495E1905.5060202@noaa.gov> Content-type: text/plain; charset=ISO-8859-1; format=flowed Content-transfer-encoding: 7bit User-Agent: Thunderbird 2.0.0.18 (Windows/20081105) Sender: owner-igsmail Precedence: bulk ****************************************************************************** IGS Electronic Mail 02 Jan 05:39:18 PST 2009 Message Number 5874 ****************************************************************************** Author: Jim Ray & Jake Griffiths Since November 2000 the International GNSS Service (IGS) has produced Ultra-rapid (IGU) products for near real-time (RT) and true real-time applications. They include GPS orbits, satellite clocks, and Earth rotation parameters for a sliding 48-hr period. The first day of each update is based on the most recent GPS observational data from the IGS hourly tracking network. At the time of release, these observed products have an initial latency of 3 hr. The second day of each update consists of predictions. So the predictions between about 3 and 9 hr into the second half are relevant for true real-time uses. Originally updated twice daily, the IGU products since April 2004 have been issued four times per day, at 3, 9, 15, and 21 UTC. Up to seven Analysis Centers (ACs) contribute to the IGU combinations: * Astronomical Institute of the University of Berne (AIUB/COU) * European Space Operations Center (ESOC/ESU) * Geodetic Observatory Pecny (GOP/GOU) * GeoForschungsZentrum (GFZ/GFU) Potsdam * Natural Resources Canada (NRCan/EMU) * Scripps Institution of Oceanography (SIO/SIU) * U.S. Naval Observatory (USNO/USU). Thanks to the ACs, this redundancy affords a high measure of reliability and enhanced orbit accuracy. IGU orbit precision has improved markedly since late 2007. This is due to a combination of factors: decommissioning of the old, poorly behaved PRN29 in October 2007; upgraded procedures implemented by GOP around the same time, by ESOC and SIO in spring 2008, by USNO in June 2008, and by ESOC in December 2008; better handling of maneuvered satellites at the combination level starting July 2008; and stricter AC rejection criteria since July 2008. As a consequence, the weighted 1-D RMS residual of the IGU orbit predictions over their first 6 hr is currently about 24 +- 6 mm (after a seven-parameter Helmert transformation), compared to the IGS Rapid orbits and averaged over the constellation. The median residual is about 18 +- 3 mm. This small, stable median residual should permit RT users to easily detect and reject the rare IGU orbit prediction outliers that do occasionally occur. When considered over the full 24 hr prediction period, the IGU orbit errors double. Systematic rotational offsets are probably more important overall than random errors due to limitations in EOP predictions, especially UT1, reaching up to about +- 35 mm around the Z axis (equatorial at GPS altitude). See the table below for complete statistics. The near real-time observed orbits in the first half of each IGU update have WRMS residuals of about 10 to 15 mm recently; the median residuals are very similar. Note that while the precision of the Rapid orbits used for reference comparison is <10 mm compared to the IGS Finals, discontinuities between successive daily orbits imply an inaccuracy in the IGS Rapid and Final orbits of at least 21 mm WRMS. So it is likely that the observed IGU orbits are only slightly less accurate than the Rapids and that the current IGU real-time orbit predictions are normally worse by no more than a factor of two or so. Only four ACs (ESOC, GFZ, NRCan, USNO) contribute estimates of the satellite clocks, which limits the robustness and quality of the IGU clock products. Because the stochastic component of clock variations is not predictable, errors for the second-half IGU clock predictions grow quickly to the same level as the broadcast navigation values. The ultimate solution to this limitation is to provide true real-time clock estimates to use with the IGU predicted orbits, as is being developed in the IGS Real-Time Pilot Project. But the IGU observed clocks have typical errors just about double that of the Rapids, most of the time. The scatters of precise point positions using the IGU observed products are thus only slightly greater than for the Rapids. IGU Orbit Differences wrt IGS Rapids ====================================================================== (for first 6 hr of predictions from each IGU update) (647 points from 2008-07-23 06:00/1489_3_06 to 2008-12-31 18:00/1512_3_18) ====================================================================== DX DY DZ RX RY RZ SCL RMS* WRMS MEDI mm mm mm uas uas uas ppb mm mm mm ---------------------------------------------------------------------- mean 3.1 0.7 0.8 -62.2 -4.2 3.5 -0.10 31.4 23.9 17.9 SDev 4.7 4.0 3.3 161.2 188.4 268.6 0.12 14.0 6.2 3.0 mm mm mm (equatorial @ GPS hgt) mm -------------------------- mean -8.0 -0.5 0.5 -0.7 SDev 20.8 24.3 34.7 0.7 ====================================================================== * unweighted RMS is included for completeness only but users should always apply reported accuracy codes as differential satellite weights ====================================================================== See also: http://acc.igs.org/igsacc_ultra.html http://acc.igs.org/orbits/igu-products_aguf08.pdf