[IGSMAIL-6030]: GLONASS Broadcast Ephemerides (GLSV, KHAR)

[Tim.Springer at esa.int]

******************************************************************************
IGS Electronic Mail      30 Nov 03:04:20 PST 2009      Message Number 6030
******************************************************************************

Author: Tim Springer


Dear Colleagues,

As some of you may have noticed the broadcast orbits in the IGS GLONASS 
combination always look very bad for the Monday of each week. In doing 
some "reprocessing" for the upcoming AGU meeting I have run into a similar 
problem with our software as  well with "crashing" Monday solutions. 
Whilst digging into this problem I found out that in the reprocessing I 
forgot to disable the usage of the combined "brdc" file which we normally 
ignore in our IGS routine operations since it gave some problems in the 
past and still does today. So I have taken some time to look into where 
the problem comes from. It seems that the root cause is that almost every 
Monday, probably due to the week "switch over", the GLONASS broadcast 
ephemerides from the stations GLSV and KHAR are producing erroneous 
epochs. This can easily be seen in the files as they report the epoch as, 
e.g., 44 min and 59.0 sec but all other stations give the same information 
for the 45 min and 00.0 sec epoch. Since the merged "brdc" file just adds 
these different epochs to the merged file also the combined "brdc" file 
contains this error. 

So to resolve this issues I propose the following corrective actions:
Station managers from GLSV and KHAR are to correct their software for the 
generation of broadcast ephemerides
The data centres generating combined GLONASS brdc files should ignore all 
information that is not given for the full minutes
The ACs are advised to do the same as the data centres
The IGS combination software should also follow this recommendation

The Rinex navigation header of GLSV and KHAR indicates that the following 
software(s) were used:
GLSV: teqc 2008Oct2, NovaRS 1.4.2.0 MAO22.05.2009
KHAR: teqc 2008Oct2, NovaRS 1.3.0.0 MAO22.05.2009, and also NovaRS 1.4.2.0

Since teqc 2008Oct2 is used at multiple sites for generating and/or 
checking the broadcast ephemerides it seems very unlikely that teqc is the 
cause.

GLSV and KHAR are the only two sites listing the NovaRS software in their 
broadcast navigation files. So most likely the NovaRS software is causing 
the issue at hand.

To give you an idea of the size of this effect below a Table from the IGS 
GLONASS combination where the broadcast is compared to the IGS combined 
product. The ">>>>>" fields for the RMS and the WRMS "nicely" show the 
size of this effect and this has been happening for a long time now for 
day 1 of the week.

So please take the above corrective actions ASAP!


Table 1.1557.c     GPS week : 1557   MJD: 55143.0-55149.0 

         Daily transformation of broadcast orbits (brd) to the combined 
IGL orbit.
         Mean and standard deviations of the transformation parameters for 
the
         whole gpsweek.

CEN WEEK DAY |   DX    DY    DZ     RX      RY      RZ   SCL   RMS   WRMS 
|    TOFT   TDRFT
             |  [mm]  [mm]  [mm]  [uas]   [uas]   [uas] [ppb]  [mm]  [mm] 
|    [ps]  [ps/d]
-------------+------------------------------------------------------------+----------------
brd 1557 0   |  112   -72    88    1274    2226  -11990  90.1  1610  1610 
|  502456       0
brd 1557 1   |31360 -3751 -7813    -755 -230800  135800  71.4 >>>>> >>>>> 
|  499969       0
brd 1557 2   | -113    22   210    3234    -308  -10710  90.1  2029  2029 
|  497725       0
brd 1557 3   | -154  -136    80     518     459  -16110  90.7  2095  2095 
|  500480       0
brd 1557 4   |  109   -22   100   -2703    1587  -20130  89.1  1831  1831 
|  505017       0
brd 1557 5   |   78     5    22    -371    1875  -21730  86.3  1988  1988 
|  504185       0
brd 1557 6   |  -64    34    57   -1400    1249  -20300  87.0  2004  2004 
|  504162       0
-------------+------------------------------------------------------------+----------------
brd 1557 mean| 4475  -559 -1036     -29  -31958    4975  86.4 >>>>> >>>>> 
|  501999       0
brd 1557 sdev|11855  1408  2988    1928   87684   57844   6.8 >>>>> >>>>> 
|    2691       0

Kind regards,
Tim


---------------------
Tim Springer
Navigation Support Office, OPS-GN (nng.esoc.esa.de)
European Space Operations Centre
Robert Bosch Strasse 5
64293 Darmstadt
Germany
Tel: +49-6151-90-2029  Fax: +49-6151-90-3129