[IGSMAIL-6355] Upcoming switch to IGS08/igs08.atx - Details on igs08.atx

Ralf Schmid schmid at bv.tu-muenchen.de
Mon Mar 7 04:00:04 PST 2011

Dear colleagues!

As announced in [IGSMAIL-6354], the IGS will adopt a new reference frame 
called IGS08 that should be used together with igs08.atx, an updated set 
of receiver and satellite antenna phase center corrections, in the near 
future. Those will replace the absolute phase center correction model 
igs05.atx adopted in November 2006 (GPS week 1400) together with the IGS 
reference frame IGS05.

Since then, it was only possible to add correction values for new 
satellite or receiver antennas. In general, updates of existing values 
were not possible in order not to jeopardize the consistency of IGS 
products. So, for a lot of IGS stations converted field calibrations 
were still applied, although azimuthal phase center corrections down to 
the horizon from robot calibrations would have been available.

Due to a strong correlation between the satellite antenna phase center 
offsets in the Earth radial direction (z-PCOs) and the scale of the 
terrestrial reference frame, the scale difference between ITRF2005 and 
ITRF2008 is also relevant for the antenna model. If satellite z-PCOs 
were not adapted before the switch to IGS08, the scales of the 
terrestrial frame solutions of the IGS analysis centers (ACs) would no 
longer be close to the ITRF scale.

Due to these reasons a general update of the IGS antenna phase center 
correction model became necessary. As coordinate jumps are unavoidable 
when adopting a new reference frame, receiver antenna calibrations are 
ideally updated at the same time. So, besides replacing converted field 
calibrations, more or less all type-specific correction values were 
updated with results from recent calibrations of further individual 

All these changes will have an impact on the users. However, this impact 
will be much smaller than was the case with the switch from relative to 
absolute phase center corrections in 2006 (see [IGSMAIL-5189]). Details 
on the various model improvements will be given in the following. If you 
have any further comments or questions, please send those to 
schmid at bv.tum.de.

A preliminary version of the antenna model without GLONASS satellite 
antenna corrections is currently available at
As soon as the GLONASS satellite antenna corrections are finalized, the 
final model will be posted at ftp://igs.org/igscb/station/general/ and 
announced via IGSMAIL.

Many thanks to
- all ACs providing sat. PCO estimates within their reprocessed SINEX files
- Xavier Collilieux (IGN) for analyzing those SINEX files
- CODE and ESA for providing GLONASS satellite antenna corrections
- Martin Schmitz (Geo++ GmbH) for providing updated robot calibrations
- Paul Rebischung (IGN) for analyzing the impact on station coordinates

With best regards,

Ralf Schmid
on behalf of the IGS Antenna Working Group

Major changes of igs08.atx w.r.t. igs05.atx

* Satellite antenna phase center corrections (GLONASS still tbd.!)

+ The satellite z-PCOs from igs08.atx are consistent with the new IGS 
reference frame IGS08, whereas those from igs05.atx were approximately 
consistent with IGS05.

+ The z-PCO estimates are based on the results of more ACs. The GPS 
values in igs08.atx were derived from the weekly SINEX files of five ACs 
(CODE, ESA, GFZ, MIT, NRCan), whereas igs05.atx was based on two ACs 
only. The GLONASS corrections (still tbd.) were estimated by CODE and 
ESA, whereas igs05.atx contained a CODE-only solution.

+ All preliminary z-PCOs for satellites launched in recent years could 
be replaced by actual estimates. It is the intention for igs08.atx that 
preliminary values should be replaced within few months after the launch.

+ Due to an improved quality of the ITRF vertical rates (ITRF2008 
compared to ITRF2000), the z-PCOs are no longer trend-corrected.

+ igs08.atx also contains information on historical satellites. All GPS 
Block I satellites were added with block mean values. The same is true 
for all GLONASS satellites back to the IGEX-98 campaign (still tbd.).

* Receiver antenna calibrations

+ Robot calibrations for 15 additional antenna types could be added. For 
those types, converted field calibrations or copied values were applied 
so far or they were not contained in the file:
   ASH701941.B     SCIS
   ASH701945C_M    NONE
   ASH701945C_M    OLGA
   ASH701945E_M    NONE
   ASH701945E_M    SNOW
   ASH701945G_M    SNOW
   LEIAX1202GG     NONE
   NOV503+CR       SPKE
   NOV600          NONE
   TRM29659.00     SCIS
   TRM29659.00     SCIT
   TRM39105.00     NONE
   TRM41249.00     TZGD

+ For 9 additional antenna types, robot calibration results were copied 
from antenna types that should be identical in construction. For those 
types, converted field calibrations were applied so far or they were not 
contained in the file:
   ASH700228B      NONE
   ASH700228E      NONE
   ASH700718A      NONE
   ASH700936A_M    SNOW
   ASH700936E_C    SNOW
   ASH701945B_M    SNOW
   ASH701945D_M    SNOW
   LEIAS05         NONE
   LEIAT202-GP     NONE

+ Existing type-specific phase center corrections from robot 
calibrations were updated with results from recent individual antenna 
calibrations, if available. Therefore, 46 robot calibrations changed:
   AOAD/M_T        NONE
   ASH700228A      NONE
   ASH700936A_M    NONE
   ASH700936B_M    NONE
   ASH700936C_M    NONE
   ASH700936D_M    NONE
   ASH700936D_M    SNOW
   ASH700936E      NONE
   ASH700936E      SNOW
   ASH700936E_C    NONE
   ASH700936F_C    NONE
   ASH701073.1     NONE
   ASH701073.3     NONE
   ASH701945B_M    NONE
   ASH701945D_M    NONE
   ASH701945G_M    NONE
   ASH701946.2     NONE
   ASH701946.3     NONE
   LEIAS10         NONE
   LEIAT504        NONE
   LEIAT504        LEIS
   LEIAT504GG      NONE
   LEIAT504GG      LEIS
   LEIAX1202       NONE
   LEIGS09         NONE
   TPSCR.G3        NONE
   TPSCR.G3        TPSH
   TPSCR4          NONE
   TRM14177.00     NONE
   TRM14532.00     NONE
   TRM22020.00+GP  NONE
   TRM29659.00     NONE
   TRM29659.00     TCWD
   TRM33429.00+GP  NONE
   TRM33429.20+GP  NONE
   TRM41249.00     NONE
   TRM55971.00     NONE
   TRM57971.00     NONE
   TRM59800.00     NONE
   TRM59800.00     SCIS
   TRM59800.80     NONE
   TRM59800.80     SCIS
However, the correction values for 41 antenna types with robot-based 
values remained unchanged:
   ASH700228D      NONE
   ASH700718B      NONE
   ASH700936B_M    SNOW
   ASH700936C_M    SNOW
   ASH701945C_M    PFAN
   ASH701945C_M    SNOW
   ASH701946.3     SNOW
   LEIAR10         NONE
   LEIAR25         NONE
   LEIAR25         LEIT
   LEIAR25.R3      NONE
   LEIAR25.R3      LEIT
   LEIAR25.R4      NONE
   LEIAR25.R4      LEIT
   LEIAR25.R4      SCIT
   LEIAT302-GP     NONE
   LEIAT303        NONE
   LEIAT303        LEIC
   LEIAT504        OLGA
   LEIAT504        SCIS
   LEIGS15         NONE
   NOV702          NONE
   NOV702GG        NONE
   NOV750.R4       NONE
   NOV750.R4       NOVS
   TPSCR.G3        SCIS
   TRM23903.00     NONE
   TRM29659.00     OLGA
   TRM29659.00     SNOW
   TRM33429.20+GP  TCWD
   TRM55971.00     TZGD
   TRM57970.00     NONE
   TRM59800.00     SCIT
   TRM59800.80     SCIT

+ GLONASS-specific corrections from robot calibrations were added, if 
available. Those will be used to generate the GLONASS products of the 
IGS. In case that GLONASS-specific values are not available for a 
certain antenna type, the correction values for the GPS frequencies will 
be used instead.

+ As the correction values of the IGS reference antenna AOAD/M_T got 
updated, all converted calibrations slightly changed. This concerns 87 
antenna calibrations converted from NGS field results and 14 types 
converted from igs_01.pcv.

+ To improve internal consistency the correction values of 3 antenna 
types are now converted from NGS field calibrations, rather than from 
   ASH700228C      NONE

+ The calibrations of 2 antenna types were removed due to naming 
inconsistencies between IGS and NGS:
   SPP571908273    NONE
   SPP571908273    SPKE

+ The calibrations of 2 antenna types remained untouched due to problems 
with unmodeled subtypes having the same name. The Antenna Working Group 
will try to approach this problem in the near future:

Some remarks on the magnitude and impact of the changes

"AOAD/M_T        NONE" as an example

Although the Dorne Margolin antenna AOAD/M_T of Allen Osborne 
Associates, Inc. was the reference antenna of the IGS, as long as 
relative phase center corrections were used, its "type mean values" 
contained in igs05.atx were based on the calibration of one single 
antenna only, namely SN 404 (serial number). However, the igs05.atx 
correction values were a mean of 42 individual calibration runs. For 
igs08.atx, additional 20 calibration runs of a second antenna (SN 393) 
were considered.

As the number of individually calibrated antennas is small, the phase 
center corrections change significantly, if additional calibrations are 
taken into account. The differences between calibration results for 
antennas of the same type contain the errors of the calibration 
procedure, but also the effect of deviations of the individual antenna 
from the ideally manufactured one. The latter effect could only be 
compensated by using individual rather than type mean calibrations. 
However, individual calibrations are not considered within the IGS so far.

If the number of individual calibrations to derive the type mean values 
was bigger, the transition from one phase center model to the next would 
be smoother. On the one hand, this would have advantages for the 
maintenance of the reference frame, but on the other hand, the stability 
of the phase center corrections would somehow be illusive, as actual 
deviations from the type mean hardly showed up anymore.

In the case of AOAD/M_T, the phase center correction changes due to the 
update from igs05.atx to igs08.atx are on the sub-mm level. The biggest 
effect shows up in the up component of the L1 PCO (0.6 mm) as well as in 
the phase center variations (PCVs) for L1 (up to 0.6 mm below an 
elevation of 10°). If these corrections are applied to derive the 
ionosphere-free linear combination, the effect is amplified. Thus, IGN 
could find coordinate changes for stations equipped with AOAD/M_T of 
about 2-3 mm (see coefficient "a" of the model for up position shifts) 
that were too small to be considered for IGS08:

As the AOAD/M_T antenna is used to convert relative field calibrations, 
also all those correction values change in the same manner. In the 
future, coordinate changes due to calibration updates will get smaller 
and smaller, whereas the problem of individual antenna deviations 
persists. Unfortunately, their impact on station coordinates could 
easily reach the 1 mm level.

Biggest phase center correction changes

By checking the coefficients "a" of IGN's latitude-dependent correction 
models (ftp://igs-rf.ensg.eu/pub/IGS08/new_calib/lat_models.txt), one 
can easily detect those antenna types that suffer the biggest changes 
(differences of more than 1 cm in the station coordinates):

+ ASH700228C      NONE
Whereas the former relative IGS model igs_01.pcv only contained one set 
of phase center corrections for all antennas of the type ASH700228x, NGS 
provides three different sets (ASH700228A/B, ASH700228C, ASH700228D/E). 
Therefore, igs08.atx contains values converted from relative NGS field 
calibrations that differ significantly from those converted from 
igs_01.pcv (contained in igs05.atx).

+ ASH700718A      NONE
Both NGS and igs_01.pcv provide identical correction values for 
ASH700718A and ASH700718B. Therefore, the robot-based values for 
ASH700718B were copied. The big offset component in North direction only 
shows up in the igs_01.pcv values (used for igs05.atx), but not in the 
NGS or Geo++ results.

+ ASH701941.B     SCIS
For that antenna/radome combination no calibration results were 
available so far. Therefore, the correction values for the antenna 
without radome (NONE) were applied within the IGS. So, the dramatic 
coordinate change in vertical direction demonstrates the effect of 
ignoring the radome.

+ LEIAT202-GP     NONE
As both NGS and igs_01.pcv provide identical correction values for 
LEIAT202-GP and LEIAT302-GP, robot-based values were copied from 
LEIAT302-GP. The values contained in igs05.atx were converted from 

According to NGS, LEISR299_INT should be identical to LEISR399_INTA, but 
not to LEISR399_INT as was the case in igs_01.pcv. This might be related 
to the confusion as regards these antenna names within rcvr_ant.tab.

+ TRM39105.00     NONE
In this case, converted field calibrations could be replaced by 
robot-based values.

Biggest impact of robot-to-robot updates

In most cases the coordinate changes caused by robot calibration updates 
are small. The "a" coefficients from 
ftp://igs-rf.ensg.eu/pub/IGS08/new_calib/lat_models.txt are below 2 mm 
for most antenna types that already had robot-based correction values in 
igs05.atx. The calibration differences are smaller, but amplified by the 
ionosphere-free linear combination.

The biggest coordinate changes can mainly be explained by significant 
changes in the number of individually calibrated antennas. Detailed 
information on the number of individual antennas and calibration runs is 
given in the antenna-specific comments contained in igs08.atx:

+ ASH700936E      NONE (up component)     1 ->  2 individual antennas
+ TRM29659.00     NONE (east component)  12 -> 18 individual antennas
+ ASH700936D_M    NONE (up component)     1 ->  4 individual antennas
+ LEIAT504GG      NONE (up component)     5 -> 25 individual antennas

More information about the IGSMail mailing list