Journal of Geodesy (2001) 75: 613619

ITRS, PZ-90 and WGS 84: current realizations

and the related transformation parameters
C. Boucher1, Z. Altamimi2
1

Institut Geographique National, DAEI, 2 Avenue Pasteur, 94160 Saint-Mande, France

e-mail: cboucher@club-internet.fr; Tel.: +33-1-43-98-83-27; +33-1-43-98-84-88
2
Institut Geographique National, ENSG/LAREG, 68 Avenue Blaise Pascal, 77455 Marne-la-Vallee, France
e-mail: altamimi@ensg.ign.fr; Tel.: +33-1-64-15-32-55; Fax: +33-01-64-15-32-53
Received: 9 June 2000 / Accepted: 12 June 2001

Abstract. The rst results of the International GLONASS Experiment 1998 (IGEX-98) campaign have provided signicant material to illustrate the mutual
benets of the GLONASS system and the realization
of the International Terrestrial Reference System
(ITRS). A specic aspect, namely the relationship
between the World Geodetic System 1984 (WGS 84)
and the PZ-90 system using ITRS as a primary standard,
is investigated. A review of current works is carried out.
A transformation strategy is proposed for the three
systems based on recent results from IGEX-98 and an
independent set of transformation parameters derived
by the Jet Propulsion Laboratory from ITRF97 and PZ90 coordinates for 16 global stations.
Key words: Terrestrial reference systems
International terrestrial reference system World
geodetic system 84 PZ-90

1 Introduction
The rapid development and use of global satellite
navigation systems has made the concept of terrestrial
reference systems (TRS) a very important and somewhat
sensitive issue. Comparisons between positions derived
from these navigation systems and existing maps, either
on land or at sea, exhibit noticeable discrepancies caused
by dierent coordinate systems, in particular dierent
TRSs.
The two currently operational global satellite navigation systems, namely the global positioning system
(GPS) developed by the USA and the Global Navigation Satellite System (GLONASS) system developed by
Russia, have made this point clear in their various oCorrespondence to: C. Boucher

cial publications: GPS uses WGS 84 as its reference

TRS, while GLONASS uses PZ-90.
Biases between satellite-derived positions and map
positions currently reach several hundreds of meters and
are therefore visible most of the time, even in the case of
GPS with selective availabilty (SA). But biases between
GPS-derived and GLONASS-derived positions are less
clear. Nevertheless, many users have identied systematic discrepancies by collocating receivers of each type.
Knowledge of the dierence between World Geodetic
System (WGS) 84 and PZ-90 was therefore essential to
better understand such comparisons, and better separate
the reference system eect from other sources of discrepancies. Moreover, the simultaneous availability of
both systems and their technical similarities have encouraged the development of mixed receivers. The increased number of simultaneously tracked satellites is a
key factor in improving the quality of positioning, not
only for accuracy, but also for availability or reliability.
Therefore it is of great interest to use both GPS and
GLONASS measurements in such a process, but it is
required that any reference system discrepancies are
corrected when the broadcast navigation information of
each system is used. This last point gave an impetus to
recent investigations on the WGS 84/PZ-90 transformation performed mainly by the navigation community.
This paper reviews some of these estimates, including
the signicant progress realized thanks to the IGEX-98
campaign, which provided for the rst time a remarkable data set of GLONASS satellites in conjunction with
GPS and laser networks. (Willis et al. 1999a, b). It also
reviews some basic points which better explain the apparent diversity of transformation parameter estimates.
2 The basic concepts and their uses
These concepts were redened extensively in the 1980s
by the astronomical and geodetic communities (see e.g.
Kovalevsky et al. 1989).
An ideal TRS is dened as a three-dimensional (3-D)
reference frame (in the mathematical sense) close to the

614

Earth and co-rotating with it. In the Newtonian

framework, the geometry of the physical space considered as a Euclidian ane space of dimension 3 provides
a standard and rigourous model of such a system
through the selection of an ane frame. A geocentric
TRS has its origin close to the geocenter and its orientation equatorial (Z axis is the direction of the pole). We
further distinguish two types, depending on the method
used (classical terrestrial or space techniques) to establish and realize such systems.
1. (Really) geocentric, which means that realizations
will be close to the geocenter (a few meters in the early
days of satellite geodesy, nowdays a few centimeters)
and the zero meridians of the equatorial orientations
will be almost identical (currently expressed using the
name of the Greenwich meridian, for historical reasons).
2. Quasi-geocentric, for systems established with
terrestrial geodetic techniques usually using a fundamental station. In such cases, the origin may be oset
from the geocenter by several hundred meters. Notice
also that these systems currently use a zero meridian
linked to some astronomical observatory and therefore
should be realigned to the conventional Greenwich origin by adopting a longitude for this zero meridian with
respect to Greenwich. For instance, in Europe, several
conventional origins are still in use for mapping purposes (e.g. Paris, Rome, Madrid).
Under these restrictions, the general transformation
of the Cartesian coordinates (X) of any point close to the
Earth from any one TRS to any other one will be given
by a tri-dimensional similarity transformation (T is a
translation vector, k a scale factor and R a rotation
matrix) as follows:
X 1 T1;2 k1;2
R1;2
X 2
For practical use, we adopt the convention currently
used by IERS and in Boucher and Altamimi (1996). The
parameters for transforming an X system into an XS
system are denoted T1, T2, T3, D, R1, R2, and R3:
0 1 0 1 0 1 0
10 1
Xs
X
T1
D
R3 R2
X
B C B C B C B
CB C
Ys

T
2

R3
D
R1
@ A @ A @ A @
A@ Y A
Zs

T3

R2

R1

It must be noted that the sign convention for rotation

angles often found in the literature is the opposite of
this.
A conventional TRF is dened as a set of physical
points with precisely determined coordinates in a specic
coordinate system that is the realization of an ideal TRS.
Such frames are currently classied as of dynamical or
kinematical type, depending on whether or not a dynamical model is applied in the determination process of
these coordinates.
The example of a dynamical TRF of specic interest
in this paper is the ephemerides of articial satellites.
The ephemerides derived from the navigation message
either from GPS or GLONASS, or those computed by
analysis centers processing space geodetic data, are of

this type. We shall categorize these as `E-frames' in

this paper. It is important to note that what we label
E-frames are satellite ephemerides expressed in an Earthxed geocentric system, not ephemerides in a quasiinertial frame. This implies in particular that they are
sensitive to the Earth orientation parameter (EOP) values adopted in computations generating the satellite
ephemerides.
However, kinematical TRFs are the most user oriented type. We can again distinguish between two types
in this study.
1. `T-frame' will designate a set of coordinates (and
possibly velocities) of a tracking network of stations
used to produce ephemerides by dynamical processing.
2. `N-frame' will designate a network of points with
coordinates (and possibly velocities) determined by
processing tracking data of the satellite systems
considered here (GPS or GLONASS).
Some fundamental remarks must be made at this point
to illustrate the relative complexity of the consistency of
the various realizations of a TRS.
To some extent, transformation parameters have to
be understood as representing the transfer of any systematic changes which t with some part of this similarity transformation (shift, scale, or rotation). The
values of these parameters are determined by an estimation process [usually least squares (LS)], based on
two frames with corresponding points. This can therefore be done for T-, E- or N-frames. A more theoretical
presentation of this can be found, for example, in Sillard
and Boucher (submitted).
In the case of operational orbit computations, as
done for generating broadcast messages for each satellite system, we can easily understand that a transformation of the T-frame used in this computation will
map, to some extent, especially in orientation into the
resulting E-frame, a dynamical law such as Kepler's
third law introducing constraints on origin and scale,
depending on orbital arc length. It is important to note
that there is not a simple law presently available to
describe this transfer function between T-frames and
E-frames. It certainly depends upon tracking network
geometry, orbital arc characteristics, type of orbit, the
arc length, and the number of simultaneously observed
satellites). For instance, recent investigations on the
DORIS (Doppler Orbitography and Radiopositioning
Integrated by Satellite) system showed results on this
specic topic using numerical simulations in the case of
the Topex/Poseidon mission (Morel 2000). In the case
of the GPS type conguration, the current ndings of
analysis done by the International GPS Service (IGS)
indicate a T- to E-frame transfer only signicant in rotations (Kouba personal communication). In addition,
we must also understand that other error sources can
infer systematic biases in the E-frame which are not in
the T-frame (this was the case for uncorrected antenna
osets of satellites).
In the case of N-frames generated by point positioning of ground networks, the E-frame will generally map

615

directly into the resulting N-frame if the orbits are not

readjusted. This may not be true if strong constraints are
placed on a priori values, if the adopted satellite clock
data aect the origin (Kouba personal communication),
or if there are constraints on some station positions, as is
often the case (see example below).
3 Description of the three systems and history
of their realizations
3.1 The International Terrestrial Reference
System (ITRS)
The International Earth Rotation Service (IERS) is
presently responsible for the realization, use and promotion of the International Terrestrial Reference System (ITRS) as dened by the International Union of
Geodesy and Geophysics (IUGG) resolution No. 2
adopted in Vienna in 1991 (Geodesist's Handbook 1992,
Bulletin Geodesique 66).
The ITRS denition fullls the following conditions.
1. It is geocentric, the center of mass being dened for
the whole Earth, including oceans and atmosphere.
2. The unit of length is the meter (SI). This scale is
consistent with the TCG (Geocentric Coordinate
Time) time coordinate for a geocentric local frame, in
agreement with the International Astronomical Union (IAU) and IUGG (1991) resolutions. This is obtained by appropriate relativistic modeling.
3. Its orientation was initially given by the Bureau International de l'Heure (BIH) orientation at 1984.0.
The IERS Reference Pole (IRP) and Reference Meridian (IRM) are consistent with the corresponding
BIH reference directions to within +/) 0.005. The
BIH reference pole was adjusted to the Conventional
International Origin (CIO) in 1967; it was then kept
stable independently until 1987. The uncertainty of
the tie of the IRP with the CIO is +/) 0.03.
4. The time evolution of the orientation is ensured by
using a no-net-rotation condition with regard to
horizontal tectonic motions over the whole Earth.
[See IERS Conventions (McCarthy 1996) for a detailed
description of the ITRS.]
Realizations of the ITRS are produced by the Terrestrial Frame Section of the IERS Central Bureau under the name International Terrestrial Reference Frames
(ITRF). These consist of lists of coordinates (and velocities) for a selection of IERS sites (tracking stations
or related ground markers) that are obtained by combining the results of several space techniques, currently:
very long baseline interferometry (VLBI), lunar laser
ranging (LLR), GPS, satellite laser ranging (SLR), and
DORIS. General documentation on terrestrial reference
systems and frames is available at http://lareg.ensg.
ign.fr/ITRF/.
Almost every year since the beginning of the Service,
a new global solution has been performed and published, replacing the previous solutions. The most recent
solutions are ITRF94, ITRF96, and ITRF97, which is

the current one (a new solution is in progress and will be

published as ITRF2000).
Each such solution is an N-frame realizing ITRS.
In addition, the IGS has adopted the ITRS as its
reference system. Consequently the IGS products are all
referred to this system. The GPS precise ephemerides
produced by the IGS (either precise, rapid, or predicted)
are therefore E-frames in the ITRS for all GPS satellites.
3.2 The World Geodetic System 1984 (WGS 84)
WGS 84 is a complete standard for georeferencing and
was established by the US Department of Defense
(National Imagery and Mapping Agency, formerly
Defense Mapping Agency). A TRS was of course
adopted for WGS 84. However, it must be understood
that WGS 84 as a TRS has had several successive
realizations, improving its geometric quality from the
meter level to the centimeter level (Malys and Slater
1994; Slater and Malys 1998). The initial realizations
used the Transit satellite navigation system (Satellite
Doppler), where various residual errors were contributing to limit its nal accuracy to meter level. For instance,
ionospheric residual error caused signicant height
biases (Sillard and Boucher 1996).
The GPS broadcast orbits represent the basic realization of WGS 84 for modern users. According to the
present terminology, this is an E-frame. As already
stated, any improvement in processing strategy may
produce systematic changes that map into the similarity
transformation parameters. Numerically the corresponding frame will change accordingly. This is particularly true for the eect of the T-frame on the
generation of GPS broadcast orbits. The operational
GPS tracking station network (ve OCS stations) is
small and any change in adopted positions for these
stations will map to some extent into the broadcast orbits. As reported by Malys and Slater (1994), changes at
the meter level have occurred. Since January 1997, the
T-frame labeled WGS 84 (G873), which has been
adopted for OCS and broadcast orbits, is consistent with
ITRS to better than a decimeter. This T-frame was derived from a global GPS solution for the combined
network of OCS and NIMA GPS tracking stations. [See
also the recent paper by Malys and Slater (1998)].
3.3 The PZ-90 system
The PZ-90 system is a global system developed by
Russia, similar to WGS 84, and used as the nominal
system for GLONASS navigation. It was realized by
positioning 26 ground stations established from observations of the Geo-IK geodetic satellite, including
photographing it against a star background, Doppler
measurements, laser ranging, and satellite altimetry. It
also included electronic and laser range measurement of
GLONASS and Etalon satellites. A subset of these
stations is used to generate the broadcast PZ-90 GLONASS orbits (Boykov et al. 1993; Mitrikas et al. 1998).

616

4 Review of previous determinations

of transformation parameters between PZ-90
and other systems (WGS 84 or ITRS)
Several investigations have been published about the
transformation between PZ-90 and WGS 84. We
mention briey some of them for which we have
information, giving for each a short summary and a
critical comment.
4.1 Misra et al. (1996)
Summary:
Purpose: link between PZ-90/WGS 84 E-frames.
Method: comparison of E-frames for GLONASS satellites 63 and 67.
Period: 19951996.
E-frame in PZ-90: from broadcast message, estimated by
authors at 20-m accuracy level.
E-frame in WGS 84: estimated by laser tracking using a
global network. Accuracy estimated at 10-m level.

Comments:
This transformation is globally valid but shows instabilities and time variations of the PZ-90 E-frame for
GLONASS with regard to the computed ITRS
E-frame derived by laser data. Its quality is at best at
the 1-m level.
4.4 Bazlov et al. (1999)
Summary:
Purpose: link between PZ-90/ITRS N-frames.
Method: comparison of two N-frames located in Russian territory.
Period: 1997.
N-frame in PZ-90: original PZ-90 positions derived from
Geo-IK data.
N-frame in ITRS: GPS positioning at the stations,
computed in ITRS.
Comments:

Comment:

This formula is valid for the Russian territory and based

on an N-frame not related to GLONASS.

It is dicult to guarantee this estimate to better than a

few meters.

5 Contribution of the IGEX-98 campaign

4.2 Rossbach et al. (1996)

Summary:
Purpose: link between PZ-90/ITRS N-frames.
Method: comparison of two N-frames located in Europe. The network included six stations (Herstmonceux, Madrid, Wettzell, Metsahovi, Maspalomas, and
Zvenigorod).
Period: 1996.
N-frame in PZ-90: determined with the Bernese software
GLONASS data with broadcast orbits.
N-frame in ITRS: IGS values of the stations in ITRS.
Comments:
The resulting transformation is valid at better than
meter level for Europe and the epoch 1996. Extrapolation in time or in other geographical areas cannot
guarantee this level of accuracy.

4.3 Mitrikas et al. (1998)

Summary:
Purpose: link between PZ-90/ITRS E-frames.
Method: comparison of E-frames for GLONASS satellites 63 and 67.
Period: 19951997.
E-frame in PZ-90: from broadcast message.
E-frame in ITRS: orbit computation using laser data
expressed in ITRF94.

5.1 Analysis done at the IERS Terrestrial

Frame Section
The IERS Terrestrial Frame Section issued a call for
results among the analysis centers participating in the
IGEX-98 campaign. Two centers provided results in
response to this call.
5.1.1 GeoForschungsZentrum, Potsdam (GFZ)
solutions
GFZ provided 11 weekly SINEX les corresponding to
GPS weeks 991 to 1001. These solutions were derived
with the following properties.
1. GLONASS as well as GPS data for mixed dual-frequency receivers were used.
2. GPS orbits from the IGS nal solution were introduced and xed (except eclipsing GPS satellites).
3. Earth rotation parameters were xed to the estimated
values of the IGS.
4. Coordinates of only one station were xed to their
initial values in the determination of orbits.
5.1.2 Jet Propulsion Laboratory, Pasadena
(JPL) solutions
The JPL provided two types of daily solutions using
over 100 days of data.
1. Station precise point positioning (PPP) solutions
computed from GPS dual-frequency tracking data by
xing the GPS satellite orbit and clock to the IGS/
FLINN solution. The reference frame for this set of
solutions is the one dened by IGS/FLINN orbit
solution, i.e. ITRF96. In this solution, station position,

617

receiver clock and tropospheric delay for each individual

site are solved for, station by station.
2. Station PPP solutions computed from GLONASS
dual-frequency tracking data by xing the GLONASS
satellite orbit and clock to the broadcast ephemeris and
clock values; the reference frame for this set of solutions
is the one dened by the GLONASS broadcast orbit,
which is PZ-90. In this solution, the broadcast orbit is
rst smoothed by a dynamic t (trajectory t) to remove
outliers and gross errors. The 3-D RMS error of the t is
around 5 m. Station position, receiver clock and tropospheric delay are then solved for individual sites by
xing the orbit to the smoothed orbit le and xing the
satellite clock to the broadcast clock values.
The results of these data were presented at the IGEX
Workshop held in Nashville in September 1999 (Altamimi and Boucher 1999).
5.1.3 Use of GFZ data
The GFZ data are N-frames with some constraints on
station positions. The way they were processed indicates
they are realizations of ITRS. We checked their N-frame
with regards to ITRF97. We found an agreement to
better than 3 cm for each weekly solution. Consequently, the corresponding E-frame for GLONASS satellites
is in ITRS and could be compared with the broadcast
message. Unfortunately these data were not available to
us for this study.
5.1.4 Use of JPL data
The JPL provided data which could be used to compute
transformation parameters on N-frames.
1. One hundred daily N-frames in PZ-90 using point
positioning of mixed receivers with GLONASS phase
data and GLONASS broadcast orbits. The comparison
between each daily solution in PZ-90 and ITRF97 was
presented at the Nashville (Altamimi and Boucher 1999)
Workshop. The mean values of the 100 sets of transformation parameters are given in Table 1, item (A).
2. The 100 daily solutions were combined together to
provide an additional N-frame in PZ-90. The transformation between the cumulated N-frame in PZ-90 and
ITRF97 is given in Table 1, item (B). The RMS residual
over 16 stations is 46 cm. The parameters refer to the
central epoch of the data span, namely 1999.5. Figure 1
shows the distribution of the 16 common stations.
We can conclude from the consistency of both
methods (A) and (B) that the precision of the resulting
transformation is better than 0.1 m, resulting in the
following ndings.
1. The agreement between (A) and (B) is of the order of
10 cm.
2. The standard deviation of the parameters of the regression lines computed in (A) are at the 10 cm level
(largest one 16 cm). This includes the network geometric eect (Fig. 1), which is not negligible. We estimated it by comparing the JPL daily ITRF96
solution for the 16 stations to ITRF97 and performed
a similar regression. The resulting mean transformation exhibits an agreement better than 5 cm.

Table 1. Results using IGEX-98 data

X

XS

T1
T2
T3
D
R1
R2
R3
(cm) (cm) (cm) (ppb) (mas) (mas) (mas)

ITRS PZ-90 )30

)24
3
0

10
15
2
0

90
77
45
110

0
31
)13
)9

3
3
37
16

13
19
)10
4

)355
)353
)350
)357

(A)
(B)
(C)
(D)

3. In case B the residuals of 40 cm at

each of the 16
p
stations are consistent with 40= 16 10 cm for
transformation parameters.

5.2 Other analysis using IGEX-98 data

Two other recent results beneting from the IGEX
campaign are available.
1. The group from the University of Bern (Ineichen
et al. 1999) computed daily transformation parameters
between E-frames in PZ-90 from GLONASS broadcast
and the ITRS using their processing of GLONASS phase
data with the Bernese software and T-frame in ITRF96.
The resulting average transformation is labeled (C) in
Table 1. A recent report by Habrich (1999) that uses a
similar analysis of IGEX data conrms these numbers.
2. A Russian group from the GEO-ZUP Company
and the Russian Mission Control Center (Mitrikas et al.
1999) has computed orbits using the laser tracking data
from eight GLONASS satellites collected during IGEX98, based on a T-frame expressed in ITRF94. The
resulting E-frame in the ITRS was compared to the
E-frame from the GLONASS broadcast ephemerides
and to the E-frame based on an averaged post-processed
orbit determined from ranging measurements. The corresponding formula is labeled (D) in Table 1. Mitrikas
et al. (1999) also noted that transformation rotations
about the x and y axes (R1, R2) are correlated with
polar motion because of the way that the GLONASS
Satellite Control Segment computes the broadcast
ephemeris predictions.
Table 1 shows that all results agree at the few decimeter to 1-m level.

Fig. 1. Distribution of the 16 common stations between ITR97 and

the PZ-90 realization obtained by cumulating 100 JPL solutions

618

6 Critical results
In order to summarize in a compact way the evolution of
the estimates, we have collected transformation parameters standardized as PZ-90 to WGS 84. None of the
studies [except (1)] used a direct realization of WGS 84,
but a realization of the ITRS. We have seen that, since
1997, the WGS 84 GPS broadcast ephemerides T-frame
is consistent with the ITRS at better than the 5-cm level.
This may not be completely true for earlier epochs.
Furthermore, as mentioned previously, the WGS 84
broadcast E-frame will have larger discrepancies. The
IGS is monitoring daily the WGS 84 E-frame for GPS
satellites, by comparing it with IGS E-frames (expressed
in the ITRS). Their RMS variations are currently 23 m
with biases of several decimeters (see IGS Analysis
Coordinator Reports, e.g. Kouba and Mireault 1999).
Table 2 presents the values.
At this level, we must now undertake a critical review.
1. First, the last three determinations show an agreement
at the decimeter level (maximum disagreement of
45 cm). They all use IGEX data and therefore the
same epoch (1999).
2. (1) shows disagreement at the 2-m level, which is acceptable considering its intrinsic value.
3. Same remark for (2) at 1-m level.
4. (5) is an extension of (3) using more data, especially
IGEX laser tracking. The disagreement is at the 1-m
level. Their paper has, moreover, clearly shown the
time variation of the E-frame transformations computed using laser data. But (5) shows a fairly good
agreement with (6)(8), better than 50 cm.
We therefore select as useful estimates (5), (6), and (8).
They are relevant to the present epoch through IGEX
data, the rst two are E-frames using laser and phase data
respectively, the last is a N-frame of the global IGEX
network. The other determinations are of less interest, all
being compatible considering their intrinsic qualities.
7 Conclusion: recommendations on transformation
strategy
Before concluding on the choice of a transformation
formula, we can make some general recommendations,
as follows.
Table 2. Transformation parameters from PZ-90 to WGS 84

1. The best simultaneous use of GPS and GLONASS

data should be based on IGS products, where GPS and
GLONASS precise ephemerides are available in the
ITRS. Moreover, accurate clock information is also
provided, consistent with these E-frames.
2. Although the IGS is providing progressively real
time products, broadcast ephemerides for both GPS and
GLONASS will still be used for many applications. The
most useful information is therefore to obtain the instantaneous transformation between the two E-frames.
Again, the IGS is a reference source for these data.
Comparisons between IGS orbits and broadcast messages are being disseminated for both GPS and
GLONASS. The IGS now makes this information
widely available, through the IGS Rapid combination
summaries for GPS and through IGLOS (International
GLONASS Pilot Project) information for GLONASS.
3. It may nevertheless be useful to adopt a reference
transformation when previous data are not available to
a particular user.
This is why we suggest adopting the following values
based on an average of the three determinations (5), (6),
and (8), which agree to better than 50 cm with each
other. At this level, we consider this formula a viable
reference, considering the previous remarks.
0

B C
@Y A
Z WGS84
0 1
0
1
X
0:07 m
B C
B
C
@Y A
@ 0:0 m A
Z
0:77 m
1
0 PZ-90
10
3 ppb
353 mas
4 mas
X90
C
B
CB
@ 353 mas
3 ppb 19 mas A@ Y90 A
Z90
4 mas
19 mas
3 ppb
Ultimately, such a reference formula should be approved by the relevant organizations, in particular the
scientic services of the IERS and IGS.
As a nal recommendation, we again draw attention
to the fact that the various realizations of WGS 84 and
PZ-90 transformations remain inconsistent at the meter

T1
(cm)

T2
(cm)

T3
(cm)

D
(ppb)

R1
(mas)

R2
(mas)

R3
(mas)

Frame
type

Epoch

0
0
)47
)110
0
)3
30
24

250
0
)51
)30
0
)2
)10
)15

0
0
)200
)90
)110
)45
)90
)77

0
0
22
)120
9
13
0
)31

0
0
2
0
)16
)37
)3
)3

0
0
1
0
)4
10
)13
)19

392
330
356
169
357
350
355
353

E
N
E
N
E
E
N
N

9596
95
9597

Europe
Russia
IGEX
IGEX

9699
99
99
99

(1)a
(2)
(3)
(4)
(5)
(6)
(7)
(8)

(1) Misra et al. (1996); (2) Rossbach et al. (1996); (3) Mitrikas et al. (1998); (4) Bazlov et al. (1999); (5)
Mitrikas et al. (1999) [(D) values of Table 1], assuming identity between ITRS and WGS 84; (6) Ineichen
(1999) [(C) values of Table 1]; (7) Altamimi and Boucher (1999) [(A) values of Table 1]; (8) This paper
[(B) values of Table 1, using JPL point positioning solution]

619

level. Thus, such transformations should be considered

as `conventional'. In particular, as mentioned before, the
best way to remove systematic biases between the WGS
84 E-frame (GPS brodcast ephemerides) and PZ-90
E-frame (GLONASS broadcast ephemerides) is presently to combine the daily transformation parameters
between each of these E-frames and the ITRS E-frames
provided by IGS ephemerides (for GPS and GLONASS
satellites respectively).
Acknowledgements. We want to thank all groups who have
provided data, in particularly GFZ and JPL. We also thank Jan
Kouba and Jim Slater for their valuable comments. Finally, we are
thankful to Pascal Willis for his stimulating role, as IGEX coordinator and editor of this special issue of the Journal of Geodesy.

References
Altamimi Z, Boucher C (1999) GLONASS and the International
Terrestrial Reference System. IGEX-98 Workshop, Nashville,
1314 September, pp 3746
Bazlov YA, Galazin VF, Kaplan BL, Maksimov VG, Rogozin VP
(1999) GLONASS to GPS, a new coordinate transformation.
GPS world January
Boucher C, Altamimi Z (1996) International Terrestrial Reference
Frame. GPS World 7(9): 7174
Boucher C, Altamimi Z, Sillard P (1999) The 1997 International
Terrestrial Reference Frame (ITRF97). IERS tech note 27.
Observatoire de Paris
Boykov VV, Galazin VF, Kaplan BL, Maksimov VG, Bazlov YA
(1993) Experience in developing the PZ-90 Geocentric Coordinate System. Geod Kartogr 11: 1721
Geodesist's Handbook (1992) J Geod 66
Habrich H (1999) Geodetic applications of the Global Navigation
Satellite System (GLONASS) and GLONASS/GPS Combinations. PhD dissertation, Astronomical Institute, University of
Bern
Ineichen D, Rothacher M, Springer T, Beutler G (1999) Computation of precise GLONASS orbits for IGEX-98. In: Schwarz
(ed) Geodesy beyond 2000. International Association of

Geodesy Symposia, vol 121. Springer, Berlin, Heidelberg New

York, pp 2631
Kouba J, Mireault Y (1999) 1998 Analysis coordinator report.
1998 IGS tech rep. IGS Central Bureau, Pasadena
Kovalevsky J, Mueller II, Kolacek B (1989) Reference frames in
astronomy and geophysics. Kluwer Academic, Dordrecht
Malys S, Slater J (1994) Maintenance and enhancement of the
World Geodetic System 1984. Proc ION-94 Salt Lake City,
September
McCarthy D (ed) (1996) IERS Conventions. IERS tech note 21.
Observatoire de Paris
Misra P, Abbot R, Gaposhkin E (1996) Integrated use of GPS and
GLONASS: transformation between WGS 84 and PZ-90. Proc
ION GPS96, pp 307314
Mitrikas VV, Revnivykh SG, Bykhanov EV (1998) WGS 84/PZ-90
transformation parameters determination based on laser and
ephemeris long-term GLONASS orbital data processing. Proc
ION GPS98, pp 16251635
Mitrikas VV, Revnivykh SG, Glotov VD, Zinkovski MV (1999)
PZ-90 GLONASS to ITRF transformation as a result of IGEX98 laser tracking campaign. IGEX-98 Workshop, Nashville, 13
14 September
Morel L (2000) References Geodesiques pour les futures missions
altimetriques: application a la mission JASON. PhD dissertation, Paris Observatory
Rossbach U, Habrich H, Zarraoa N (1996) Transformation parameters between PZ-90 and WGS 84. Proc ION GPS96,
pp 279285
Slater JA, Malys S (1998) WGS 84 past, present and future. In:
Brunner FK (ed) advances in positioning and reference frames.
International Association of Geodesy Symposia, vol 118.
Springer, Berlin Heidelberg New York
Sillard P, Boucher C (1996) Improvement of the transformation
between ITRF and Doppler-realized WGS 84. J Geod 70: 768
780
Sillard P, Boucher C (submitted) A review of algebraic constraints
in terrestrial reference frame datum denition. J Geod
Willis P, Beutler G, Gurtner W, Hein G, Neilan RE, Noll C, Slater
JA (1999a) IGEX, International GLONASS Experiment scientic objective and preparation. Adv Space Res 23(4): 659664
Willis P, Slater JA, Beutler G, Gurtner W, Noll C, Weber R, Neilan RE, Hein G (1999b) The IGEX-98 campaign: highlights
and perspectives. IAG General Assembly, Birmingham,
pp 2225