ReProcessing

The participating Analysis Centers contributing complete solutions to this effort are:

• CODE — Center for Orbit Determination in Europe, Switzerland
• EMR — Natural Resources Canada, Canada
• ESA — European Space Operations Centre (ESOC), ESA, Germany
• GFZ — GeoForschungsZentrum/Potsdam, Germany
• JPL — Jet Propulsion Laboratory, USA
• MIT — Massachusetts Institute of Technology, USA
• NGS — National Geodetic Survey, NOAA, USA
• PDR — GeoForschungsZentrum/Potsdam & Technical University of Dresden, Germany
• SIO — Scripps Institution of Oceanography, USA

In addition, some centers provide weekly SINEX solutions to help densify the terrestrial reference frame, particularly for GPS stations located near tide gauges (associated with the TIGA Pilot Project); work is underway to perhaps include associated satellite orbits and clocks:

• GTZ — GFZ TIGA, Germany
• ULR — University of La Rochelle, France

Summaries of the analysis strategies and procedures for each AC are posted at the IGS Central Bureau. Submitted AC solution files are posted at this GFZ site.

The product combination centers are:

• NRCan — Natural Resources Canada, Canada
  • SINEX combination — as of mid-Feb. 2009, results posted for years 2000-2008 (see below)
• NCL — University of Newcastle-upon-Tyne, UK
  • SINEX combinations
• GFZ — GeoForschungsZentrum/Potsdam, Germany
  • ftp directories for repro1 weekly summary reports & daily product files from orbit combinations
  • a summary of all AC submissions is given in the ftp file “Submission_statistics_*.txt” where the * is the date of last update
• NRL — Naval Research Laboratory, USA
  • clock & timescale combination results not yet available

Essential elements of the analysis procedures used by all participating centers include:

• try to complete archives for missing or useful data sets
  • focus on as many tide gauge stations as possible
  • also try to include NGA/NIMA tracking stations
  • try to include all GPS stations co-located with other techniques
• use “absolute” antenna calibrations for satellite transmit & ground receive antennas
  • background information
  • implemented in IGS operational solutions starting GPS week 1400
  • current antenna calibration file is igs05.atx in ANTEX format
• use IGS05 reference frame (aligned to ITRF2005)
  • background information, including station-specific corrections to account for effect of new antenna calibration model
  • implemented in IGS operational solutions starting GPS week 1400
  • current reference frame file is IGS05_repro.snx in SINEX format
  • currently known or suspected discontinuities in the IGS station positions are tabulated in SINEX format
• IERS Conventions 2003 generally implemented
  • with occasional updates posted by the IERS Conventions Center
• updated model for station displacements due to ocean tidal loading
  • site-dependent load coefficients recommended using FES2004 ocean tide model
  • corrections for counter-balancing center-of-mass motion of the solid Earth should be included in site coefficients
  • whole-Earth center-of-mass corrections should also be applied in generating SP3 orbits as described in updated IERS Conventions Chapter 7
• no non-tidal loading displacements should be applied to station positions
  • see this position paper on principles for conventional contributions to modeled station displacements, presented at the IERS workshop on Conventions, held at BIPM, 20-21 September 2007
• updated models for tropospheric propagation delays
  • refer to revised IERS Conventions Chapter 9
  • a priori meteorological data sources:
    • [1] local sensor met files (which however are only available for a few sites), or
    • [2] barometric pressure values computed using the location- & season-dependent GPT model, or
    • [3] retrieved from a numerical weather model, as for the ECMWF global values provided by the service at the Technical University of Vienna in the form of gridded hydrostatic zenith delays (see below)
  • the a priori hydrostatic delay in the zenith direction should be computed from the surface pressure according to the formula of Saastamoinen (1972) as given by Davis et al. (1985) and shown as eqn (11) in the revised IERS Conventions Chapter 9
    • or retrieved directly in this form based on a numerical weather model, as for the gridded ECMWF values provided by the
      service at the Technical University of Vienna
    • or alternatively one can use an a priori zenith delay consisting of both the hydrostatic and wet components (using surface temperature values from GPT and a standard relative humidity of about 0.5) to ensure tropospheric parameter adjustments that are more nearly zero-mean
  • line of sight delays should be computed using either
    • [1] the location- & season-dependent GMF mapping function model, or
    • [2] or a mapping function based on retrieval coefficients from a numerical weather model, as for the ECMWF global values provided by the service at the Technical University of Vienna for the VMF1 mapping function model
  • residual tropospheric delays should be parameterized in the GPS data analysis on the assumption that they are predominantly due to the wet component of the troposphere (wet mapping function partials) & azimuthal gradients, as described in IERS Conventions Chapter 9
• no common model for higher-order ionospheric corrections has been agreed yet
• the IGS master SINEX template file, igs.snx, is updated daily based on the history of IGS site logs & should be used to validate AC SINEX file metadata for information on historic and discontinued sites igs_with_former.snx should be used
• P1-C1 satellite code biases
  • see site at Astronomical Institute of the University of Bern for general information, bias tables, and receiver types:
    • http://www.aiub.unibe.ch/download/bcwg/cc2noncc
    • ftp://ftp.unibe.ch/aiub/bcwg/cc2noncc
  • see ESA/ESOC site for latest version (v6.2) of cc2noncc.f routine:
    • • http://acc.igs.org/software/CC2NONCC_ESOCv6.5.tar
  • for information about biases before 02 April 2000 see this message from Stefan Schaer (10 Feb 2009): “Pre-2000” GPS P1-C1 bias values for IGS reprocessing

Test Results

To evaluate analysis and combination procedures, the ACs have provided test solutions for the first quarter of year 2000. Results from those tests are posted here.

Objective

Before proceeding with the full reprocessing campaign, a 3-month test period at the beginning of year 2000 was chosen to evaluate the analysis procedures and performances of the Analysis Centers (ACs), as well as exercise the reprocessing combinations.

Test Sinex Frame Results

The ACs provided test solutions for the first quarter of year 2000 (GPS weeks 1042-1059; the ESA solutions cover the 14 weeks from 1042-1055). Statistics for the station position residuals (average weighted mean & average standard deviation) from the combination of the weekly SINEX files from each AC, after removing separate Helmert transformations (see following), are tabulated below.

The mean & standard deviation of the Helmert parameters for the weekly SINEX file alignments to IGS05 are tabulated below for each AC during this test period (GPS weeks 1042-1059).

There are 319 stations in the test cumulative solution file (SINEX format), which is available in file IG000P17.ssc.Z (unix compressed) or uncompressed as IG000P17.ssc.

In addition, Remi Ferland has produced the following postscript (ps) plots concerning the test SINEX combinations:

• Table 5-1 — number of stations & variance factors for each AC
• Table 5-2-1 — station residual statistics w.r.t. IGS05
• Table 5-2-2 — station residual statistics w.r.t. weekly combination
• Table 5-2-3 — station residual statistics w.r.t. cumulative combination
• Table 5-3-1 — Helmert transformation parameter estimates to IGS05
• Table 5-3-2 — Helmert transformation parameter standard deviations to IGS05
• Table 5-4 — combined apparent geocenter offsets & standard deviations

Test Sinex ERP Results

Earth rotation parameter statistics (mean & standard deviation) for each AC during the test period (GPS weeks 1042-1059) are tabulated below. The PDR ERPs have been rejected due to the use of continuity over-constraints; see this site for further information. The SIO ERPs have the same problem but have not be excluded here; this explains their unrealistically small rate variations.

In addition, Remi Ferland has produced the following postscript (ps) plots concerning the test SINEX ERP combinations:

• Table 5-5-1 — weighted averages of ERP residuals
• Table 5-5-2 — standard deviations of ERP residuals

Test Orbit Results

Orbit combination statistics (mean & standard deviation) for each AC during this test period (GPS weeks 1042-1055) are tabulated below.

Issues to be Resolved

Based on the test results, there are several outstanding issues and questions that should be resolved:

• all ACs need to update their analysis summaries
  • including operational & reprocessing ACs
  • CODE last updated 12 Mar. 2002
  • EMR last updated 23 Jan. 2002
  • GFZ last updated 27 Feb. 2003
  • JPL last updated 13 Apr. 2004
  • MIT last updated 28 Jan. 2004
  • SIO last updated 31 Oct. 2005
• new ESA & NGS solutions generally much improved over original solutions
• not all metadata used by ACs is consistent with the epoch of the observational data
  • MIT SINEX comparisons to the IGS cumulative solution indicate an inconsistency in the N component metadata
• RMS global consistency of AC SINEX frames not quite as good as in routine operational solutions
• differences in alignment of AC SINEX frames sometimes exceed 2 cm in their weekly TZ geocenter offsets
  • the ESA solutions dominate the combined geocenter value due to very small formal errors
• differences among AC polar motion rate solutions sometimes exceed 200 µas/d
  • NGS LODs are significantly offset & have poor stability
• differences among AC orbit solutions sometimes exceed 1 cm in their DZ origin components
  • NGS orbit DZ origin offset most
• differences among AC orbit solutions sometimes exceed 100 µas in their rotational orientation
  • SIO orbit rotations largest & most unstable
• the relative agreement among the participating ACs is generally fairly good, but it is not at the level of the current IGS operational Final products
• constraints on the Earth rotation rate parameters are not handled consistently among ACs
  • affects the PDR & SIO polar motion rates
• only two ACs provide usable clock estimates
  • not sufficient for a robust combined clock product
• verify that all necessary IGS metadata files are current & complete
  • include metadata for all historic & discontinued stations in the IGS master SINEX template file, igs.snx
  • master file of P1-C1 satellite code biases back to 1994
• organize product archiving & file naming structures

Steps Ahead

The following schedule is suggested:

• Feb. 2008 — begin full reprocessing phase
  • start with data for week 1459 (23-29 Dec 2007)
  • then work backwards in time, hopefully back to 1994.0
• 02-06 June 2008 — IGS workshop
  • evaluate progress for initial phase of reprocessing

Issues to be solved

There are several remaining issues to be addressed:

• all ACs need to maintain updated analysis summaries, including operational & reprocessing ACs
  • SIO last updated 31 Oct. 2005
• verify that all necessary IGS metadata files are current & complete
  • all historic & discontinued stations in the IGS master SINEX template file igs_with_former.snx

The following filenaming and archiving conventions have been adopted for this first and future reprocessing campaigns:

• IGS reprocessing campaign numbering
  • Since over the coming years we expect that there will be successive reprocessings, we will declare that each has a 1-character number to designate it. The present campaign will be “1”. It will run from now till at least the end of 2009, probably.
  • Do we really need to think about more than 9 reprocessings now? That will be for a period of probably about 20 years. Many other things will likely change over that time.
• AC & combination file deposits
  • ACs should deposit standard Finals-like reprocessed solution files to the CDDIS Global Data Center (only) in their normal file delivery areas but under new subdirectories .../repro.
  • The AC and combination filenames must indicate the reprocessing campaign sequence number by using names of the type:
        CC#WWWWD.typ.Z
    where the standard center code “CCC” is changed to “CC#” and “#” is the particular reprocessing sequence number.
  • Those AC files are automatically moved to the standard public product archive directories but in new subdirectories
        /gps/products/WWWW/repro#
    where # is the particular reprocessing sequence number. During the reprocessing, the combination products will also be put in these subdirectories.
  • As with standard operational files, if an AC or combination center resubmits a newer version of any reprocessed solution file it will overwrite any older version.
• move to operational status
  • When the current 1^st reprocessing campaign reaches its conclusion and the results are certified to be of a quality at least as good as the original operational solutions, then the present operational files will be moved downward to
        /gps/products/WWWW/orig
    subdirectories for permanent archiving.
  • At the same time, the 1^st reprocessed files will be symbolically linked from
        /gps/products/WWWW/repro1/CC1WWWWD.typ.Z
    back up to the standard product directories to become the new operational products using the current standard IGS product names, from the earliest reprocessed week through 1459:
        /gps/products/WWWW/repro1/CC1WWWWD.typ.Z
-> /gps/products/WWWW/CCCWWWWD.typ.Z
  • The present operational product files for weeks 1460 onward will remain in place and should be consistent with the 1^st reprocessing products for weeks 1459 and earlier. (In fact, the original operational AC solutions during 2008 — weeks 1460 through 1512 — will also be recombined at the SINEX level to ensure stricter consistency in the handling of time series discontinuities and some other details. These recombinations are posted in repro1 subdirectories too, and will be moved to the operational directories at the end of the campaign. However, there will be no other recombined products nor new AC solutions.)
  • In this way, users should always be able to access the latest, most consistent set of IGS products by always downloading files from the standard product directories using the standard names.
  • On the other hand, users must take care to check for processing consistency when mixing recently downloaded files with older files. README and other advisory information will be posted and distributed to caution users on this point.
• future reprocessings
  • The same approach will be used with any future reprocessing campaigns, by creating sequentially new
        /gps/products/WWWW/repro#
product directories, etc.

The following schedule is suggested:

• Feb. 2008 — begin full reprocessing phase
  • start with data for week 1459 (23-29 Dec 2007)
  • then work backwards in time, hopefully back to 1994.0
  • the operational Final products from week 1460 forward should be consistent with the reprocessing so that the continuous span from 1994 onward will eventually become the official IGS product set
• 02-06 June 2008 — IGS workshop
  • evaluate progress for initial phase of reprocessing
• January 2009 — prepare IGS submission for ITRF2008
  • all available solutions (with sufficient AC redundancy) submitted on a sliding schedule up January 2009 will be used to form a preliminary time series of weekly repro1 SINEX combinations:
    • earlier — years 2007-2005 to be submitted to CDDIS
    • 14 Nov. 2008 — year 2004 to be submitted to CDDIS
    • 28 Nov. 2008 — year 2003 to be submitted to CDDIS
    • 12 Dec. 2008 — year 2002 to be submitted to CDDIS
    • 26 Dec. 2008 — year 2001 to be submitted to CDDIS
    • 09 Jan. 2009 — year 2000 to be submitted to CDDIS
  • the preliminary repro1 time series will then be submitted to the IERS ITRF Product Center to be included in the ITRF2008 multi-technique combination
  • it was originally expected that the IGS time series would cover at least the period 2000 – 2008 (inclusive)
  • this period was extended in May (see below) to cover the additional years 1997 – 1999 (inclusive), which were submitted by late July
• 10 February 2009 — submission deadline for ITRF2008 input solutions
  • all technique combined solutions due to ITRF Product Center of the IERS
  • first preliminary ITRF2008 results & discussions at EGU General Assembly, 19-24 April 2009 in Vienna
  • final ITRF2008 solution to be released 15 July 2009
• May – July 2009 — extend IGS submission for ITRF2008
  • due to delays in finalizing submissions from the other techniques, further time was available for the IGS to extend its solution
  • years 1997 through 1999 were added to the original IGS submission
  • and the first half of 2009 included also
  • total IGS solution period for ITRF2008 is thus years 1997 – 2009.5 (inclusive)
• IGS solution for ITRF2008 — some assessment results are given below
• later 2009 — finish repro1 products
  • form full set of repro1 orbit & SINEX combination products using all available AC solutions during period 1994 – 2007
  • repro1 products will be merged with standard operational IGS Final products for period since week 1460 (since 30 December 2007) for complete coverage from 1994 to present
  • schedule for completion of repro1 reanalysis work by ACs (as of 22 Jul 2009):
    • 14 Aug 2009 — year 2007 to be submitted to CDDIS
    • 21 Aug 2009 — year 2006 to be submitted to CDDIS
    • 28 Aug 2009 — year 2005 to be submitted to CDDIS
    • 04 Sep 2009 — year 2004 to be submitted to CDDIS
    • 11 Sep 2009 — year 2003 to be submitted to CDDIS
    • 18 Sep 2009 — year 2002 to be submitted to CDDIS
    • 25 Sep 2009 — year 2001 to be submitted to CDDIS
    • 02 Oct 2009 — year 2000 to be submitted to CDDIS
    • 09 Oct 2009 — year 1999 to be submitted to CDDIS
    • 16 Oct 2009 — year 1998 to be submitted to CDDIS
    • 23 Oct 2009 — year 1997 to be submitted to CDDIS
    • 30 Oct 2009 — year 1996 to be submitted to CDDIS
    • 06 Nov 2009 — year 1995 to be submitted to CDDIS
    • 13 Nov 2009 — year 1994 to be submitted to CDDIS
  • final cutoff dates for any replacement AC solutions (updated 12 November 2009):
    • 16 Nov 2009 — last submissions for years 2001-2007 at CDDIS
    • 23 Nov 2009 — last submissions for years 1997-1999 at CDDIS
    • 30 Nov 2009 — last submissions for years 1994-1996 at CDDIS
    • it remains to be seen if enough usable solutions will be available for 1994
    • final weekly SINEX combinations are expected by the end of Dec 2009

• IGS solution for ITRF2008 — combined solution files submitted to IERS ITRF Product Center
  • SINEX combination results prepared by Rémi Ferland at NRCan
  • description of IGS submission in February 2009 for the period 2000 – 2008
  • extension of IGS submission in May 2009 to add the period 1997 – 1999
  • extension of IGS submission in July 2009 to add first half of 2009
  • summary report for IGS submission to ITRF2008 by Rémi Ferland at NRCan
  • plots of combination statistical results
  • file of discontinuities used for ITRF2008 submission
• Consistency of the IGS contribution to ITRF2008 (2009)
  • presentation by Rémi Ferland at the European Geosciences Union 2009 General Assembly
• Reprocessed polar motion results from the ITRF2008 combination
  • analysis of IGS GPS-only polar motion results submitted for ITRF2008, illustrating the time evolution of the quality of the IGS polar motion time series over the 12.5 year period from 1997.0 through 2009.5
• Quality assessment of the ITRF2008 (2009)
  • presentation by Z. Altamimi, X. Collilieux, and L. Métivier at the Fall 2009 American Geophysical Union Meeting
• ITRF2008 station position residual time series
  • station position residuals from long-term linear stackings of solutions from each contributing technique, including IGS
  • NOTE: the adjustment of weekly Helmert alignment parameters to form the long-term frames has probably absorbed some portion of the non-linear variations of individual stations (e.g., due to time-varying surface loading), especially from the vertical component

Repro1 preliminary product combinations for evaluation:

• SINEX TRF & ERP preliminary combination results
  • provided by Natural Resources Canada (NRCan), Canada
  • summary report prepared by Rémi Ferland at NRCan (10 Dec 2009)
• ftp directories for preliminary weekly orbit combination summary reports
  • provided by GeoForschungsZentrum/Potsdam (GFZ), Germany

Repro1 final product combinations:

• SINEX TRF & ERP final combination results
  • provided by Natural Resources Canada (NRCan), Canada
  • release message prepared by Rémi Ferland at NRCan (05 Mar 2010)
  • summary tables and plots
  • Combination of the reprocessed IGS Analysis Center SINEX solutions (2010), presentation by R. Ferland at IGS 2010 Workshop, Newcastle upon Tyne, UK, 30 June 2010
• Availability of reprocessed products
  • IGS Mail 6136 by G. Gendt & R. Ferland (26 Apr 2010)
  • clocks are aligned to GPS Time, not yet to the IGS time scale, but this has no impact on purely geodetic applications
• ITRF2008 station position residual time series
  • station position residuals from long-term linear stackings of solutions from each contributing technique, including IGS
  • note: the adjustment of weekly Helmert alignment parameters to form the long-term frames has probably absorbed some portion of the non-linear variations of individual stations (e.g., due to time-varying surface loading), especially from the vertical component

Assessments & use of repro1 products:

• Quality assessment of the ITRF2008 (2009)
  • presentation by Z. Altamimi, X. Collilieux, and L. Métivier at the Fall 2009 American Geophysical Union Meeting
• Quality assessment of GPS reprocessed terrestrial reference frame (2009)
  • presentation by X. Collilieux, Z. Altamimi, L. Métivier, and T. van Dam at the Fall 2009 American Geophysical Union Meeting (ppt)
• Reprocessed polar motion results from the ITRF2008 combination
  • analysis of IGS GPS-only polar motion results submitted for ITRF2008, illustrating the time evolution of the quality of the IGS polar motion time series over the 12.5 year period from 1997.0 through 2009.5
• Reprocessed polar motion assessments — for period 2005 – 2007
  • results shown are for the latest AC solutions, which may not be the same as some earlier solutions used for the ITRF2008 submission; also not all ACs had solutions available at the time of ITRF2008 submission; all AC solutions will be used in the final repro1 combination
• Validating Earth orientation series with models of atmospheric & oceanic angular momenta (2009)
  • presentation by R. Gross at the Fall 2009 American Geophysical Union Meeting
• ITRF2008 ERP comparisons with IGS repro1 (25 March 2010)
  • comparison by Z. Altamimi of preliminary ITRF2008 combinations by IGN & DGFI with IGS repro1 (IG1) series
• ITRF2008 & IGS repro1 polar motion comparisons with AAM+OAM (28 March 2010)
  • comparison by J. Kouba of preliminary ITRF2008 combination results for polar motion by IGN & DGFI as well as by IGS repro1 (IG1) with AAM+OAM excitation
• Preliminary reprocessed orbit assessments — for period 2005 – 2007
  • results shown are for preliminary AC solutions, which may not be the same as some earlier solutions used for the ITRF2008 submission or final solutions submitted later; also not all ACs had solutions available at the time of ITRF2008 submission; all AC solutions will be used in the final repro1 combination
• Reprocessed orbit assessments — for 3 equal periods from Aug. 1999 – Dec. 2007
  • results shown are for the latest AC solutions, which may not be the same as some earlier solutions, and the final IG1 combination; the analysis of day-boundary discontinuities considers conventional XYZ 1D magnitudes, as well as along, cross, and radial components
• Preliminary analysis of IGS reprocessed orbit and polar motion estimates (2009)
  • presentation by J. Ray and J. Griffiths at the European Geosciences Union 2009 General Assembly
• Assessment of the orbits from the 1st IGS reprocessing campaign (2009)
  • presentation by J. Griffiths, G. Gendt, Th. Nischan, and J. Ray at the Fall 2009 American Geophysical Union Meeting
• Multi-technique reprocessing and combination using “space-ties” (2009)
  • presentation by T. Springer, F. Dilssner, D. Escobar, M.Otten, I. Romero, and J. Dow at the Fall 2009 American Geophysical Union Meeting
• Quality of GRACE orbits using the reprocessed IGS products (2009)
  • presentation by Z. Kang, B. Tapley, S. Bettadpur, and H. Save at the Fall 2009 American Geophysical Union Meeting
• Assessment of 3D hydrologic deformation using GRACE and GPS (2009)
  • presentation by C. Watson, P. Tregoning, K. Fleming, R. Burgette, W. Featherstone, J. Awange, M. Kuhn, and G. Ramillien at the Fall 2009 American Geophysical Union Meeting (ppt)
• Results from the new IGS time scale algorithm (2009)
  • poster presentation by K. Senior and J. Ray at the Fall 2009 American Geophysical Union Meeting (ppt)
• Results from the new IGS time scale algorithm (2010)
  • poster presentation by K. Senior and J. Ray IGS 2010 Workshop, Newcastle upon Tyne, UK, 28 June 2010 (ppt)
• External comparison of EOP results (2010)
  • presentation by R.S. Gross at IGS 2010 Workshop, Newcastle upon Tyne, UK, 30 June 2010
• IGS reprocessing — Summary of orbit/clock combination and first quality assessment (2010)
  • presentation by G. Gendt, J. Griffiths, Th. Nischan, and J. Ray at IGS 2010 Workshop, Newcastle upon Tyne, UK, 30 June 2010
• ITRF2008 and the IGS contribution (2010)
  • presentation by Z. Altamimi at IGS 2010 Workshop, Newcastle upon Tyne, UK, 28 June 2010
• Status of IGS core products (2010)
  • presentation by J. Ray and J. Griffiths at IGS 2010 Workshop, Newcastle upon Tyne, UK, 28 June 2010

• Interactions of the IGS reprocessing and the IGS antenna phase center model (2009)
  • presentation by R. Schmid, P. Steigenberger, U. Hugentobler, R. Dach, M. Schmitz, and Florian Dilssner at the Fall 2009 American Geophysical Union Meeting
• Preparations for the 2nd IGS Reprocessing Campaign (2009)
  • poster presentation by J. Ray at the Fall 2009 American Geophysical Union Meeting