Apport du VLBI à l ITRF: forces et faiblesses

Transcription

1 Apport du VLBI à l ITRF: forces et faiblesses Zuheir Altamimi Paul Rebischung Laurent Métivier Xavier Collilieux IGN, France zuheir.altamimi@ign.fr Présenté par David Coulot

2 Introduction: Points-clé La réalité d une terre déformable et des erreurs systématiques des techniques Construction de l ITRF ITRF2014 et comment modéliser le mouvement non-linéaire des stations: Signaux périodiques Déformation post-sismiques (Post-Seismic Deformation - PSD) pour les stations sujettes à des tremblements de terre majeurs: un modèle de correction ITRF/ITRF2014 : contribution VLBI : Forces et faiblesses Quelques illustrations

3 "Motions" of the deformable Earth Nearly linear motion: Tectonic motion: mainly horizontal Post-Glacial Rebound: Vertical & Horizontal Non-Linear motion: Seasonal/Periodic: Annual, Semi & Inter-Annual caused mainly by loading effects Rupture & Post-seismic deformation caused by earthquakes, Volcano Eruptions,

4 Reference Frame Representations Long-Term Secular Frame: mean station positions at a reference epoch (t 0 ) and station velocities: The indispensable Frame for science applications The ITRF is a Long-Term Secular Frame "Quasi-Instantaneous" Frame: mean station positions at a "short" interval: Daily or weekly representations <= Regularized Position Nonlinear motion embedded in their time series Inputs to ITRF by 4 techniques (VLBI, SLR, GNSS, DORIS) Still rely on the ITRF for at least the orientation definition

5 DORIS GPS SLRVLBI ITRF Construction Time series stacking Step 1 Long-term Solutions DORIS GPS SLR VLBI X V, EOPs Local ties Velocity equality at co-location sites Combination ITRF X V, EOPs Step 2 ITRF Specifications: Origin: SLR Scale: SLR & VLBI Orientation: Alignment to previous ITRF

6 ITRF2014 & VLBI contribution

7 ITRF2014: Input data Service/Technique Number of Solutions Time span IGS/GNSS/GPS 7714 daily (21 yrs) IVS/VLBI 5328 daily (35 yrs) ILRS/SLR 244 fortnightly 1147 weekly (35 yrs) IDS/DORIS 1140 weekly (22 yrs)

8 VLBI Contribution to ITRF daily/session-wise solutions 407 sessions with two stations (not used) 86% of the sessions include a small number of stations between 3 & sessions with stations 8 sessions with 20 stations two sessions exceptionally include 21 and 32 stations ==> VLBI schedule not really designed for TRF

9 ITRF2014 Network

10 ITRF2014 Network: VLBI & SLR

11 Current VLBI sites (observed in 2014)

12 Current Co-locations Tahiti

13 Modelling nonlinear station motions: Motivations Position time series of all stations exhibit periodic signals More than 100 sites are subject to Post-Seismic Deformation due to major earthquakes Red Stars: EQ Epicenters (58) Green circles: ITRF2014 sites (117) Precisely modeling the above leads to more robust secular frame and site velocities.

14 ITRF Combination model Station positions, velocities and transformation parameters Earth Orientation Parameters Periodic signals PSD: applied as a correction model before stacking

15 Periodic Signals Annual & semi-annual terms estimated, using: total sum of all frequencies number of frequencies period of the i th frequency ==> 6 parameters per station & per frequency, i.e. a & b along each X, Y, Z axis.

16 Periodic Signals Annual & semi-annual terms estimated, using: total sum of all frequencies number of frequencies period of the i th frequency ==> 6 parameters per station & per frequency, i.e. a & b along each X, Y, Z axis.

17 Impact of estimating seasonal signals Negligible impact on horizontal velocities Up to 1 mm/yr change in vertical velocities, for stations with large seasonal signals, large number of discontinuities, or/and data gaps in time series

18 Estimating seasonal signals vs applying non-tidal atmospheric loading (NTAL) model? Three IGS cumulative solutions: STD: Periodic signals NOT estimated, NTAL model NOT applied NTAL * : Non-tidal atmospheric loading model applied before stacking F2: Annual and semi-annual signals estimated * NTAL model provided by Tonie van Dam

19 Wettzell (Germany) Residuals STD NTAL F2 wrms: 1.4 mm wrms: 1.3 mm wrms: 1.3 mm wrms: 1.4 mm wrms: 1.3 mm wrms: 1.3 mm wrms: 5.0 mm wrms: 4.0 mm wrms: 4.7 mm

20 POVE (Brazil) Residuals STD NTAL F2 wrms: 2.3 mm wrms: 2.4 mm wrms: 1.6 mm wrms: 1.8 mm wrms: 1.8 mm wrms: 1.4 mm wrms: 12.3 mm wrms: 12.2 mm wrms: 5.2 mm

21 Station WRMS Diffs (NTAL F2) Positive values mean better performance for seasonal terms estimation Negative values mean better performance for NTAL 59% 84% 75%

22 Post-Seismic Deformations

23 PSD Correction Regularized Position (ITRF2014) Observed Position

24 Post-Seismic Deformations Fitting parametric models using GNSS/GPS data at major GNSS/GPS Earthquake sites Apply these models to the 3 other techniques at Co-location EQ sites Parametric models: Logarithmic Exponential Log + Exp Two Exp Earthquake Post-seismic deformation

25 How to use ITRF2014 PSD models? Regularized Position (ITRF2014) Local Frame PSD Subroutines in Fortran available at ITRF2014 Web site:

26 GNSS Tsukuba Trajectory VLBI

27 VLBI contribution to the ITRF scale 27

28 ITRF2014 Co-locations (VLBI, SLR & DORIS, Co-located with GNSS)

29 DORIS, SLR & VLBI scales wrt ITRF2014 Full time series of scale factors Scale factors of SLR and VLBI solutions selected to define ITRF2014 scale DORIS SLR VLBI

30 ITRF2014 Co-locations (VLBI & SLR co-locations, NO GNSS) N : number of tie vectors if > 1

31 VLBI vs SLR Scale Difference Solution Scale at ppb Comments ITRF ± 0.10 All Tie SNX files properly weighted Rate 0.02 ± 0.02

32 VLBI vs SLR Scale Difference Solution Scale at ppb Comments ITRF ± 0.10 All Tie SNX files properly weighted Using best ties: Residuals < 5 mm Using best ties: Residuals < 5 mm Rate 0.02 ± ± LT vectors, properly weighted 1.54 ± LT vectors, equally weighted 3mm per component Using best ties: 1.68 ± LT vectors, equally weighted 1mm per component (not reliable ==> large residuals for some sites)

33 VLBI vs SLR Scale Difference Solution Scale at ppb Comments ITRF ± 0.10 All Tie SNX files properly weighted Using best ties: Residuals < 5 mm Using best ties: Residuals < 5 mm Rate 0.02 ± ± LT vectors, properly weighted 1.54 ± LT vectors, equally weighted 3mm per component Using best ties: 1.68 ± LT vectors, equally weighted 1mm per component (not reliable ==> large residuals for some sites) VLBI & SLR colocations, No GPS 1.20 ± sites (good distribution): 13 LT vectors, properly weighted

34 VLBI Weaknesses & Systematic Errors Sparse sessions, not all designed for the reference frame (see next animation for sessions in 2011) Usually 6-8 stations, twice a week rarely ~20 stations Axis offset errors, (Sarti et al., 2011) Elevation-dependent antenna deformations, esp. for large antennas (can be ~1 cm height effect), (Sarti et al., 2009)

35 In the following slides : Animation of VLBI site distribution Per session during February 2011: 14 sessions duration: 14 seconds And then: Per month during 2011: duration: 12 seconds

36 VLBI session February 01, 2011

37 VLBI session February 02, 2011

38 VLBI session February 03, 2011

39 VLBI session February 07, 2011

40 VLBI session February 08, 2011

41 VLBI session February 09, 2011

42 VLBI session February 10, 2011

43 VLBI session February 14, 2011

44 VLBI session February 17, 2011

45 VLBI session February 21, 2011

46 VLBI session February 22, 2011

47 VLBI session February 23, 2011

48 VLBI session February 24, 2011

49 VLBI session February 28, 2011

50 And now VLBI observed sites per month, during 2011

51 VLBI observed sites January

52 VLBI observed sites February

53 VLBI observed sites March

54 VLBI observed sites April

55 VLBI observed sites May

56 VLBI observed sites June

57 VLBI observed sites July

58 VLBI observed sites August

59 VLBI observed sites September

60 VLBI observed sites October

61 VLBI observed sites November

62 VLBI observed sites December

63 Conclusion ITRF2014 innovation: modelling of nonlinear station motions Seasonal signals Precise modeling of post-seismic deformations Current VLBI sessions are not designed for the reference frame: sparse sessions/stations VLBI contribution to the ITRF scale: Persistent scale offset between VLBI & SLR Antenna deformation modelling?