Impact of seasonal and postglacial surface displacement on global reference frames

Transcription

1 European Geosciences Union, General Assembly 2014 Vienna Austria 27 April 02 May 2014 Impact of seasonal and postglacial surface displacement on global reference frames Hana Krásná 1, Johannes Böhm 1, Matt King 2, Anthony Memin 2, Stanislav Shabala 2, Christopher Watson 2 1 Vienna University of Technology, Austria 2 University of Tasmania, Australia

2 Outline Empirical models of neglected seasonal station motions harmonic model (annual + semi annual period) mean annual model Comparison to displacement series obtained from the GRACE data and to hydrology loading displacement Differences in estimated EOP and celestial reference frames Impact of postglacial uplift on the variability in the TRF scale 2

3 Data in the analysis ~3700 sessions ( ) 5.6 million observations 66 stations (22 datum stations) 871 radio sources (285 datum sources) 3

4 Conventional displacement of stations International Terrestrial Reference Frame considers the position at a reference epoch plus a linear velocity term for station coordinates The actual station movement also includes several tidal and non tidal correction In our reference solution we applied solid Earth tides (IERS Conv. 2010) ocean tidal loading (FES2004) atmospheric pressure loading (GSFC Group) (tidal and non tidal part usual practice in VLBI analysis) pole tide loading (IERS Conv. 2010) ocean pole tide loading (Desai 2002) 4

5 Unmodellednon linear displacements (neglected seasonal station motions) The increasing accuracy of VLBI observations and the growing time span of available data allow the determination of seasonal signals in station positions which still remain unmodelled in the conventional analysis approach We create empirical harmonic models (at annual and semi annual period) for selected stations AND mean annual models by stacking yearly time series of station positions 5

6 Seasonal station displacement models Ten most observing stations ( ). In light red the harmonic model and in blue the mean annual model are shown. In: Krásná H., Z. Malkin, J. Böhm (2014). Proceedings of the Journées

7 Comparison with GRACE series H [mm] harmonic model (annual + semiannual) mean annual model H [mm] GRACE series provided by M. Weigelt and T. van Dam correl. coeff. correl. coeff. FORTLEZA NYALES HARTRAO TIGOCONC HOBART TSUKUB KOKEE WESTFORD MATERA WETTZELL

8 Comparison with a hydrological model harmonic model (annual + semiannual) mean annual model GRACE series provided by M. Weigelt and T. van Dam Hydrology loading displacement provided by GSFC group, D.Eriksson; computed from the monthly GLDAS NOAH model correl. coeff. correl. coeff. FORTLEZA NYALES HARTRAO TIGOCONC HOBART TSUKUB KOKEE WESTFORD MATERA WETTZELL

9 Real VLBI observations difference between solutions with and without harmonic model applied a priori on station coordinates 0.5 [mas] [mas] [ms] [mas] 0.5 Comparison of estimated EOP Artificial VLBI observations > annual amplitude 3 mm in station position 0.5 [mas] 0.5 9

10 Comparison of celestial reference frames 3 global VLBI solutions with estimation of TRF + CRF + EOP Datum sources Sources in at least 2 sessions with more than 20 observations All sources S2: harmonic model S3: mean annual model (annual + semiannual) differences w.r.t. a reference solution Significant changes in the individual source position appear if the source is observed only in a small number of sessions distributed non evenly over the year. 10

11 GIA uplift rates Comparison between the vertical rates in ITRF2008 and the modelled GIA uplift rates at stations included in our global solution The GIA uplift rates are derived by bicubic interpolation of the ICE 5G_VM2_2012 vertical land movement grids obtained from Dick Peltier's website (provided by Matt King). 11

12 GIA uplift rates Analysis of VLBI data within 2 global solutions (estimation of TRF + CRF + EOP) Analysis 1 usual parameterization Analysis 2 GIA uplift rates were applied on station positions as a priori parameters. The reference epoch is the mid point for each site included in the adjustment. Difference of estimated velocities A priori GIA rates The systematic south north shift is most probably caused by the choice of datum stations which are mainly in the northern hemisphere. 12

13 GIA uplift rates Helmert transformation parameters between the two TRF Tx mm 0.74 ± 0.40 Ty mm 0.56 ± 0.40 Tz mm 0.18 ± 0.38 Rx mas 0.01 ± 0.02 Ry mas 0.01 ± 0.02 Rz mas 0.00 ±0.01 scale mm 1.14 ± 0.38 dtx mm/y 0.05 ± 0.40 dty mm/y 0.12 ± 0.40 dtz mm/y 0.09 ±0.38 drx mas/y 0.00 ± 0.02 dry mas/y 0.00 ± 0.02 drz mas/y 0.00 ±0.01 dscale mm/y 0.10 ± 0.38 The mean of the applied uplift rates was 0.10 mm/y (only stations in the global adjustment) The scale rate between the two solutions is at the 0.1 mm/year level. The uncertainty of the scale rate is 0.38 mm/year. 13

14 Conclusions Two kinds of models for unmodelled long period signals in station coordinates were created. For validation they were compared to displacement series obtained from the GRACE data and to hydrology loading corrections. Seasonal station movements do not yield any significant systematic effect on the CRF but can cause a significant change in position of radio sources with a small number of sessions non evenly distributed over the year. Harmonic signal in the station east coordinates with an amplitude of 3 mm propagates into dut1 with an amplitude of 0.01 ms. The scale rate between the two TRF by applying the GIA uplift rates a priori on station positions is 0.10 mm/y with an uncertainty of 0.38 mm/y. 14

15 Thank you for your attention! Impact of seasonal and postglacial surface displacement on global reference frames Hana Krásná, Johannes Böhm, Matt King, Anthony Memin, Stanislav Shabala, Christopher Watson 15

16 Annual and semi annual signal in TRF Harmonic functions Sine and cosine amplitudes are derived from the topocentric station displacement with zero a priori values Estimated in a global adjustment as additional parameters to the default solution Δd REN = Ac REN mjd mjd0 mjd mjd0 sin 2π + AsREN cos P P 2π P period of station movement ( days, days) mjd 0 reference epoch set to J mjd time of observation amplitude A = Ac + REN 2 REN As 2 REN phase Φ REN = arctan As Ac REN REN 16

17 Mean annual model (non harmonic) Approach of Tesmer et al. (2009) was followed Time series of station coordinates in a reference solution were computed Weighted mean value for each year was removed from the time series The time series was stacked into one mean year (local HEN system) Smoothing of the position estimates into a mean annual signal with formal errors as weights was done 17