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, MünchenE-Mail: angermann@dgfi.badw.de

OutlineOutline

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 DGFIITRS Combination Center at DGFI

General concept: Combination on the normal equation level

Software: DGFI Orbit and Geodetic Parameter Estimation Software (DOGS)

Geodeticdatum

Input data sets for TRF computations (1/2)Input data sets for TRF computations (1/2)

Techn. Service / AC Data Time Period

GPS IGS / NRCan weekly solutions 1996 - 2005

VLBI IVS / IGG 24 h session NEQ 1984 - 2005

SLR ILRS / ASI weekly solutions 1993 - 2005

DORIS IGN - JPL/LCA weekly solutions 1993 - 2005

ITRF2005: Time series of station positions and EOP

ITRF2005 data sets are not fully consistent, the standards andmodels 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)Input data sets for TRF computations (2/2)

Techn. Institutions Data Time Period

GPS GFZ daily NEQ 1994 - 2007

VLBI IGG / DGFI 24 h session NEQ 1984 - 2007

SLR DGFI / GFZ weekly NEQ 1993 - 2007

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 strategyTRF 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)TRF per technique (1/4)

Instrumentation changes Earthquakes Seasonal variations

Analysis of GPS station position time series

ITRF2005. 221 discontinuities in 332 GPS stations (1996 - 2005)

GGOS-D: 95 discontinuities in 240 GPS stations (1994 - 2007)

TRF per technique (2/4)TRF per technique (2/4)

Effect of annual signals ? Equating of station velocities ?

SolutionID

Velocities[mm/yr]

Sol. ID 1Sol. ID 2

4.5 ± 0.98.7 ± 0.7

Δ (2 – 1) 4.2 ± 1.2

Sol. ID 1 Sol. ID 2

obs time span

[yrs]

0.5 1.0 2.0 3.0

velocities

[mm/yr]

-37.5

3.2

-3.7

2.3

2.7

0.8

-0.3

0.5

Velocity differences w.r.t. a linear model

GPS station Irkutsk (Siberia)

GPS station Hofn (Iceland)

1997 2000 2003 20061997 2000 2003 2006

TRF per technique (3/4)TRF per technique (3/4)

Seasonal signals - Comparison with geophysical data

Models consideratmospheric,oceanic and hydrologic mass loads:

NCEP, ECCO,GLDAS

Potsdam

Krasnoyarsk

Bahrain

Correlation coefficient = 0,50

Correlation coefficient = 0,79

Correlation coefficient = 0,73

1997 1999 2 001 2003 2005

[cm]

2

0

-2

2

0

-2

2

0

-2

TRF per technique (4/4)TRF per technique (4/4)

Shape of seasonal signals can be approximated by sine/cosine annual and semi-annual functions

Brasilia Ankara

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)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)Computation of the TRF (2/3)

ITRF2005GGOS-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)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)Realization of the geodetic datum (1/4)

ITRF2005D(DGFI Solution)

GGOS-D

Origin SLR SLR

Scale SLR + VLBI SLR + VLBI + GPS

Orientation NNR conditionsw.r.t. ITRF2000

NNR conditionsw.r.t. ITRF2005

Orientationtime evolution

NNR conditions w.r.t. horizontal motions by using the Actual Plate Kinematic andDeformation Model (APKIM)

Realization of the geodetic datum (2/4)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)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)Realization of the geodetic datum (4/4)

Cdatum = (GT CGPS-1 G)-1Method:

CGPS: Covariance matrix of GPS solution (loose constrained)

G: Coefficients of 7 parameter similarity transformation matrix

Cdatum: Covariance matrix of datum parameters

Datum information of GPS observations (compared to SLR)

Standard deviations for datum parameters [mm]

tx ty tz scale

GPS 0.7 0.7 1.1 0.2

SLR 1.0 0.9 2.7 2.9

Conclusions and outlookConclusions 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).