From AC Fri Dec 18 03:23:30 1998
From: AC (AC)
Date: Fri, 18 Dec 1998 3:23:30 PST
Subject: [IGSMAIL-2105] ITRF94/96 IGS Change-Update
Message-ID:
******************************************************************************
IGS Electronic Mail Fri Dec 18 3:23:30 PST 1998 Message Number 2105
******************************************************************************
Author: AC Coordinator
Subject: ITRF94/96 IGS Change-Update
Dear IGS Colleagues,
Recently, it was noticed by Dr. Daniel Gambis, Director of the IERS Central
Bureau, that the IGS Final EOP series, when compared to the IERS EOP's,
does not seem to exhibit the estimated
March 1/98 discontinuity of about .2 mas in PM y (Polar Motion - y
component). This small, expected discontinuities are due to the change of
the IGS ITRF realisation, which since March 1, 1998, is based on the
following 47 ITRF96 station positions and velocities:
ftp://igscb.jpl.nasa/gov//igscb/station/coord/ITRF96_IGS_RS47.SNX.Z,
This new IGS ITRF96 realisation, has replaced the previous IGS ITRF94
realisation which was based on, only up to 13 ITRF94 station positions
and velocities and which had been used since June 28, 1996 (see IGSMAIL
#1838).
More specifically, the following March 1/98 changes were expected, as
estimated from the ITRF94 and ITRF96 coordinates of the 13 ITRF stations
that were used for the ITRF realisations of IGS up to Feb 28/98 (GPS Wk
0846; MJD 50872):
(IGS(ITRF94)-IGS(ITRF96); epoch 1998.16 (March 1/98, Wk 0947, MJD 50873);
(the transformation parameters are consistent with the IERS conventions)
IGS Orbits/EOP T1(cm) T2(cm) T3(cm) D(ppb) R1(mas) R2(mas) R3(mas)
PM y PM x (-dUT)
Parameter 0.0 0.1 -.1 0.4 0.21 .01 .22
Sigma .3 .3 .3 .4 .10 .10 .10
Rates per year -.02 .09 -.02 .07 .02 -.01 -.01
sigma .06 .06 .06 .09 .03 .02 .03
As seen above the estimated expected ITRF94/96 changes were indeed quite
small, in most cases they can hardly be considered statistically (and most
likely practically as well) significant, i.e. not significantly different from
zero. (As it should be expected as ITRF94 & ITRF96 are supposed to be
(nominaly) based on the same datum). Furthermore, overlapping solutions done
by some ACs (see e.g. CODE and EMR AC summary reports for Wk 0947) seemed
to support the above estimates. Subsequently, the above estimate for PM y
were applied to the IGS and all AC PM series in the IERS BULL A
combinations. However there appears to be a small apparent discontinuity
when the two IERS combined EOP series are compared, i.e. the Bull A
(which includes the expected PM y shift) and the Bull B (C04) (where no such
PM y corrections were applied). Comparisons with VLBI PM series (which, unlike
Bull A and B(C04), are independent from the IGS/AC PM solutions) are not
conclusive, mainly due to the (7 day) sampling rate. This report is an
attempt to investigate, and perhaps indicate if indeed the IGS PM has a
discontinuity (as expected) or it happened to be, more or less,continuous as
the IERS Bull B (C04) seems to indicate.
As the first step, in the Table 1, the IGS Final (IGS), IGS Rapid (IGR) as
well as the new continuous series by CODE & JPL (CODN, JPLN) were compared to
the IERS Bull A & B (C04) as well as two VLBI EOP series (NEOS (every 7 days)
and one by NASA GSFC (daily by Kalman filter also from the 7 day sampling
data (1102g.98sep29, prepared by John Gipson and Chopo Ma)). Also note that
only the VLBI series are truly independent from IGS,
as both Bull A and C04 contain significant contributions from the daily IGS
as well as some AC series. The CODN, JPLN series, even though they contain the
same data, could also be considered independent from the IGS series, at
least as far as the March 1/98 discontinuity is concerned. As can be seen
from the Table 1, even though far from conclusive, only Bull A shows the
expected jump of IGS of about .2 mas. This should not be any surprise as
the Bull A combination, as already mentioned above, includes the PM y
correction (of .21 mas) based on the above PM y estimates. However, no other
continuous series, including both VLBI's and the CODN, JPLN (i.e.
the recomputed CODE and JPL EOP's in ITRF96) indicate any significant
discontinuities in IGS PM y on March 1, 1998. Though the independent VLBI
comparisons can hardly be considered statistically significant due to the
large sigmas caused by sparse VLBI observations (every 7 days). The largely
independent (in regards to discontinuities)and highly precise series by CODEN
indicates only about half (.09 +-.02) of the expected PM y shift,
while JPLN gives (-.08 +-.02). (This can be seen from Table 1
by subtracting the CODN resp. JPLN PM y values e.g. in the Bull A
or C04 columns from the corresponding IGS values.)
Table 1: Comparisons of the 100 day mean differences before March 1/98
(ITRF94) and after March 1/98 (ITRF96) in .01 mas units. The sense is
ITRF94-ITRF96 so that the (discontinuous) IGS should show PMy shift of 21
(mas/100) according to the above estimates. (CODN, JPLN are new, recomputed,
continuous ITRF96 EOP series from CODE and JPL)
EOP - BullA EOP - C04 EOP - VLBI(neos) EOP-VLBI(gsfc)
PMx PMy PMx PMy PMx PMy PMx PMy
IGS ITRf94-96 -0.8 17.4 5.7 -1.4 5.3 -3.5 8.5 2.9
sig 1.0 0.9 1.5 1.4 8.5 6.6 6.5 5.3
IGR ITRf94-96 9.1 45.0 15.5 26.1 12.8 16.1 18.3 30.4
sig 3.3 3.5 3.5 3.4 8.7 8.0 7.0 6.4
CODN ITRf94-96 -1.9 8.4 4.5-10.5 4.1-14.7 7.4 -6.2
sig 1.0 1.3 1.6 1.6 8.9 6.4 6.1 5.0
JPLN ITRf94-96 -12.5 25.4 -14.8 4.4 -8.4 2.5 -3.6 9.8
sig 1.8 1.7 2.0 1.8 8.9 6.1 6.9 5.3
Bull A ITRf94-96 6.4 -18.9 4.8 -22.2 9.2 -14.6
sig 1.1 1.2 8.1 6.1 6.1 5.0
C04 ITRf94-96 -1.2 -5.7 2.8 4.3
sig 7.8 6.6 6.3 5.3
Since the above results can hardly be considered to be conclusive and
significant, it was decided to try to investigate possible discontinuities
in IGS orbit rotations, as seen by the precise navigation solutions for
stations WILL, BRUS and USUD, that are performed every week for evaluations
of the IGS and IGR orbits/clocks. Though, the IGS summary reports show only
the rms of 15 min epoch navigation solutions w.r.t. to the daily means.
Unfortunately, the GIPSY precise navigation solutions has been introduced
only after March 8, 98, but fortunately, for this purpose, Yves Mireault of
NRCan kindly recomputed 4 weeks before and 5 weeks after the March 1 ITRF
change . Here, only daily means, which in fact approximate cumulative (over
24 h) precise point positioning solutions were used to check the possible
(orientation) discontinuities of IGS and IGR orbits. The differences of the
average of 28 daily mean positions, before and after March 1/98, for each
of the three stations, (corrected only for the small change due to
station velocities over the 28 days), were used for the solutions of rotation
and scale changes. The Table 2 summarises the results for the IGS and IGR
orbits.
Table 2. Orientation (ITRF94-ITRF96) discontinuities of IGS and IGR orbits
as derived from the daily means of GIPSY 15 min navigation solutions used
for IGS orbit/clock evaluations. (Four weeks before and after March
1/98)(units: 0.01 mas)
IGS Orbits IGR Orbits
PMx(R2) PMy(R1) R3 PMx(R2) PMy(R1) R3
ITRF94-96 4.3 10.2 30.0 2.6 14.2 100.0
Sigma 4.4 5.3 16.5 4.5 5.5 55.0
As can be seen from the Table 2, the IGS orbit orientations, as seen by the
precise point positioning, tend to indicate about half of the expected jump
in PM y. This is also supported by the independent and continuous VLBI
series, and the CODN series in particular, once the respective sigmas are
taken into account. On the other hand the IGS R3 rotation of 30 in Table 2
agrees quite well (statistically speaking) with the expected value of 22
(mas/100) which is based on the ITRF94/96 direct comparisons. The scale
correction estimated for the IGS Final orbit (not shown here) came out also
.4 ppm. The R3 and scale changes (not shown here) for the IGR orbits
are in relatively poor agreement with the expected values.
Summary
Comparisons of the independent PM series (CODN in pacular) and the IGS Orbit
orientations (i.e. shifts of the daily mean positions at WILL, BRUS and
USUD) seem to indicate that the IGS Final PM series shows a PMy shift of about
half (i.e. 0.10 mas) of the expected value of .21 mas. The remaining rotations
(R2(PM x) and R3) and the scale changes were found to be in good agreement
with the expected values listed above (i.e. as estimated from the 13 ITRF94 and
ITRF96 station positions). So, for the highest accuracy applications, and in
particular when it is desired to remove apparent discontinuities (on March 1/98)
in series involving the IGS final products (orbits, EOP),the following values
PM y/R1 (itr94-itrf96) = .10 +-.05 mas should likely be used.
[Mailed From: Jan Kouba ]