******************************************************************************
BSW Electronic Mail 13-Feb-2015 08:43:23 CET Message Number 0335
******************************************************************************
Author: Rolf Dach
Subject: Consistency with CODE orbit modeling for BSW user with version 5.2
Dear colleagues,
since the version 5.2 of the Bernese GNSS Software has been released in
December 2012, the CODE analysis center has carried out two major
changes in the GNSS orbit modeling for its IGS products:
Phase 1:
before 14-Jul-2013 (GPS week 1749):
- a priori solar radiation pressure model as provided for version 5.2
users in SATELLITE.I08
- estimated orbit parameters:
- six initial conditions
- constant acceleration in D,Y,X components
- once-per-revolution in X
- stochastic pulses at noon
Phase 2:
after 14-Jul-2013 (GPS week 1749) but
before 04-Jan-2015 (GPS week 1826)
(as for the recent IGS reprocessing):
- no a priori solar radiation pressure model applied anymore
- Albedo model according to Rodriguez et al. 2012
- Antenna thrust (GPS: block-specific according to
http://acc.igs.org/orbits/thrust-power.txt;
GLONASS: assumed averaged value of 100W)
- same estimated orbit parameters as above:
- six initial conditions
- constant acceleration in D,Y,X components
- once-per-revolution in X
- stochastic pulses at noon
Phase 3:
after 04-Jan-2015 (GPS week 1826):
- no change in the orbit model but for the estimated orbit parameters:
- six initial conditions
- constant acceleration in D,Y,X components
- once-per-revolution in X
- twice- and four-times per revolution in D
- stochastic pulses at noon
- a new angular argument for the periodic term was used
The orbit model in ORBGEN of version 5.2 is only consistent with the
orbit modeling in phase 1 (residuals of a fit RMS <= 1mm because of the
number of digits in the precise orbit file). The differences in the
orbit modeling during phase 2 have mostly been absorbed by the
once-per-revolution terms that are estimated in all three components in
ORBGEN (RMS of the fit 1..5mm). The most recent change in the orbit
model at CODE introduces an RMS of up to 30mm (and even above) for
GLONASS satellites if the Sun is close to the orbital plane.
To compensate for this discrepancy we have published the improvement
B_049 (see http://www.bernese.unibe.ch/support/Bugs2013-07-18.php;
announced by BSW mail ftp://ftp.unibe.ch/aiub/bswmail/bswmail.0334).
With this software update also the higher order periods in the D
component may be setup in ORBGEN to fit existing orbits from CODE. After
applying this software improvement the orbits can be represented with a
RMS in the order of 5..15mm for the CODE product generated in Phase 2
and 3. The reason is that the albedo effect cannot be absorbed by the
once-per-revolution terms in the three components (because they are only
present in the X component with the D2X orbit parameter selection).
The albedo and antenna thrust models added in phase 2 will not be
provided as a patch to version 5.2 because it would affect too many
routine and even the satellite information file format.
As proposed in the README-files to the processing examples, orbit model
differences may be compensated by adding more stochastic pulses. If you
increase for instance the spacing of the stochastic pulses in ORBGEN
from 12 to 2 hours all CODE orbits from the phase 2 and 3 can also be
represented with an RMS of 1..2mm. The older orbits from phase 1 are not
affected (the dispensable parameters in ORBGEN are simply estimated to
zero).
I hope this helps to clarify the situation.
Best regards
Rolf Dach
[Mailed from: Rolf Dach ]