IGS Analysis Center Workshop, 2-6 June 2008, Florida, USA GPS in the ITRF Combination D. Angermann, H. Drewes, M. Krügel, B. Meisel Deutsches Geodätisches Forschungsinstitut, München

Outline  ITRS Combination Center at DGFI  Input data for TRF computations  Analysis and accumulation of GPS time series  GPS in the inter-technique combination  Contribution of GPS to the datum realization  Conclusions and outlook

ITRS Combination Center at DGFI  General concept: Combination on the normal equation level  Software: DGFI Orbit and Geodetic Parameter Estimation Software (DOGS) Geodetic datum

Input data sets for TRF computations (1/2) Techn.Service / ACDataTime Period GPSIGS / NRCanweekly solutions VLBIIVS / IGG24 h session NEQ SLRILRS / ASIweekly solutions DORISIGN - JPL/LCAweekly solutions ITRF2005: Time series of station positions and EOP ITRF2005 data sets are not fully consistent, the standards and models were not completely unified among analysis centers Shortcomings concerning GPS: - IGS solutions are not reprocessed (e.g., model and software changes) - Relative antenna phase center corrections were applied

Input data sets for TRF computations (2/2) Techn.InstitutionsDataTime Period GPSGFZdaily NEQ VLBIIGG / DGFI24 h session NEQ SLRDGFI / GFZweekly NEQ GGOS-D: Time series of station positions and EOP Improvements of GGOS-D data compared to ITRF2005:  Homogeneously processed data sets - Identical standards, conventions, models, parameters - GPS: PDR (Steigenberger et al. 2006, Rülke et al. 2008)  Improved modelling - for GPS: absolute instead of relative phase centre corr. - for VLBI: pole tide model was changed GGOS-D: German project of BKG, DGFI, GFZ and IGG funded by BMBF

TRF computation strategy First step: Analysis of station coordinate time series and computation of a reference frame per technique Modelling time dependent station coordinates by - epoch positions - linear velocities - (seasonal signals) - discontinuities Second step: Combination of different techniques by - relative weighting - selection of terrestrial difference vectors (local ties) - combination of station velocities and EOP - realization of the geodetic datum

TRF per technique (1/4) Instrumentation changes Earthquakes Seasonal variations Analysis of GPS station position time series ITRF discontinuities in 332 GPS stations ( ) GGOS-D: 95 discontinuities in 240 GPS stations ( )

TRF per technique (2/4) Effect of annual signals ? Equating of station velocities ? Solution ID Velocities [mm/yr] Sol. ID 1 Sol. ID ± ± 0.7 Δ (2 – 1)4.2 ± 1.2 Sol. ID 1 Sol. ID 2 obs time span [yrs]  velocities [mm/yr]     0.5 Velocity differences w.r.t. a linear model GPS station Irkutsk (Siberia) GPS station Hofn (Iceland)

TRF per technique (3/4) Seasonal signals - Comparison with geophysical data Models consider atmospheric, oceanic and hydrologic mass loads: NCEP, ECCO, GLDAS Potsdam Krasnoyarsk Bahrain Correlation coefficient = 0,50 Correlation coefficient = 0,79 Correlation coefficient = 0, [cm]

TRF per technique (4/4) Shape of seasonal signals can be approximated by sine/cosine annual and semi-annual functions BrasiliaAnkara Estimation of annual signals in addition to velocities ? Advantages: - Improved velocity estimation - Better alignment of epoch solutions Disadvantages / open questions: - More parameters (stability) ? - Seasonal signal geophysically meaningful ? - How to parameterize seasonal signals ?

Computation of the TRF (1/3) Selection of local ties at co-location sites SLR-VLBI (9) SLR-GPS (25) VLBI-GPS (27)

Computation of the TRF (2/3) ITRF2005 GGOS-D Selection of terrestrial difference vectors (1) Three-dimensional differences between space geodetic solutions (GPS and VLBI) and terrestrial difference vectors [mm] = stations in southern hemisphere Krügel et al. 2007: Poster presented at AGU Fall Meeting 2007

Computation of the TRF (3/3) Selection of terrestrial difference vectors (2) Mean pole difference Network deformation Mean pole difference: 35  as (≈1 mm) Network deformation: ≤ 0.3 mm Number of co-locations: 19 ITRF2005: Pole difference: 41  as Deformation: 1.0 mm No. co-locations: 13

Realization of the geodetic datum (1/4) ITRF2005D (DGFI Solution) GGOS-D OriginSLR ScaleSLR + VLBISLR + VLBI + GPS OrientationNNR conditions w.r.t. ITRF2000 NNR conditions w.r.t. ITRF2005 Orientation time evolution NNR conditions w.r.t. horizontal motions by using the Actual Plate Kinematic and Deformation Model (APKIM)

Realization of the geodetic datum (2/4) Station velocity residuals for 56 core stations used to realize the kinematic datum of the ITRF2005D solution w.r.t. APKIM

Realization of the geodetic datum (3/4) Translation and scale estimates of similarity transformations between combined PDR05 and weekly solutions (Rülke et al., JGR 2008)

Realization of the geodetic datum (4/4) C datum = (G T C GPS -1 G) -1 Method: C GPS : Covariance matrix of GPS solution (loose constrained) G:Coefficients of 7 parameter similarity transformation matrix C datum :Covariance matrix of datum parameters Datum information of GPS observations (compared to SLR) Standard deviations for datum parameters [mm] txtx tyty tztz scale GPS SLR

Conclusions and outlook  Discontinuities: Number of jumps reduced due to homogeneous re- processing; discontinuity tables among techniques should be adjusted.  Annual signals: Treatment of seasonal variations in station positions (e.g., by estimating sine/cosine functions) should be investigated.  Co-locations: Discontinuities are critical; at least two GPS instruments should be operated at each co-location site.  Geodetic datum: GPS reprocessing provides stable results for the scale and for the x- and y-component of the origin, the z-component shows large seasonal variations which should be investigated.  GPS reprocessing is essential for the next ITRF (unified standards and models should be applied for different techniques).