How significant is the dynamic component of the North American vertical datum?

Transcription

1 CGU Annual Scientific Meeting -8 June 212, Banff, Canada How significant is the dynamic component of the North American vertical datum? E. Rangelova, M.G. Sideris (University of Calgary) W. van der Wal (TU Delft)

2 Overview North American Vertical Reference System (NAVRS, 222) an equipotential surface defined by a local Wo realized by a geoid model N for a reference epoch To a model of the temporal geoid variations δn maintained and monitored through the collocated GNSS and AG networks Factors contributing to δn and δwo discussed in this paper decadal climate-related signal in Greenland hydrologic long-term variations glacial isostatic adjustment (GIA) of the geoid and crust Combined GRACE rates and GPS velocities Significance of δn and δh Conclusions 6 June 212 CGU Annual Meeting, Banff, Canada, June -8, 212 2

3 Connection between δh, δh H and δn At At epoch epoch ttthrough point point P: P: Temporal changes of the gravity field and heights at point P of the topographic surface W = W P At At epoch epoch t = t + δt W = W + δw W = W P P o Change in in potential δw = W P W o P o Change in in geoid geoid height height δ N = δw / g Change in in levelled height height δh ( W W ) g = P P / Absolute displacement δ h = δh + δn( + δho) in the past δh = δh δn now 6 June 212 CGU Annual Meeting, Banff, Canada, June -8, 212 3

4 δn due to Greenland ice mass loss GRACE-based ice loss model from Jensen (2) Sea-level equation (Farrell and Clark 1976) solved A compressible earth model is used Rotational feedback is included: a small effect on the sea-level shape The decadal effect for the Atlantic coast of North America is - to 1 mm. 6 June 212 CGU Annual Meeting, Banff, Canada, June -8, 212 4

5 δn due to hydrology Long-term variations in the hydrology model or model deficiencies? Snow over ice. Need to separate secular geodynamic and long-term hydrology signals in GRACE observations. High variance signals Spectral mixing 6 June 212 CGU Annual Meeting, Banff, Canada, June -8, 212

6 δn from GRACE data series Mean geoid rate and standard deviation of the geoid rate 6 k-filtered (Kusche et al.,jg, 29) DDK2 solutions: CSR, GFZ, JPL, DMT-1, AIUB and ITG Time period: January 23 to December 29 Resolution: ~34 km Error: ~ -3% 6 June 212 CGU Annual Meeting, Banff, Canada, June -8, 212 6

7 δh from GRACE & GPS velocities GPS vertical velocities in Canada ~ CBN stations from GEODVEL1b (Argus et al, GJI, 2 and Argus and Peltier, GJI, 2) k-filtered CSR and GFZ GRACE RL4 solutions Resolution: 3 km (DDK1), 34 km (DDK2) & 24 km (DDK3) GLDAS/Noah correction Outliers: (GRACE-GPS) > 6 mm/yr The GLDAS correction overall increases the standard deviation of the differences improves the agreement for southern latitudes and smaller velocities increases the disagreement for northern latitudes and larger velocities The CSR/GLDAS combination better suited for NA than GFZ/GLDAS 6 June 212 CGU Annual Meeting, Banff, Canada, June -8, 212 7

8 δh from GRACE&GPS velocities CSR CSR - GLDAS GFZ GFZ - GLDAS Outliers (9) Bias (-3.3 mm/yr) Slope (1.29) Std (.71 mm/yr) Outliers () Bias (-3.7 mm/yr) Slope (1.29) Std (.6 mm/yr) Outliers (9) Bias (-3.8 mm/yr) Slope (1.39) Std (.89 mm/yr) Outliers (2) Bias (-3.4 mm/yr) Slope (1.2) Std (.6 mm/yr) DDK1 - - Outliers (11) Bias (-2.9 mm/yr) Slope (1.16) Std (.46 mm/yr) - - Outliers () Bias (-3.6 mm/yr) Slope (1.22) Std (.9 mm/yr) - - Outliers () Bias (-3.3 mm/yr) Slope (1.24) Std (.64 mm/yr) - - Outliers (17) Bias (-3.4 mm/yr) Slope (1.2) Std (.2 mm/yr) DDK2 - - Outliers (16) Bias (-2. mm/yr) Slope (1.2) Std (.8 mm/yr) - - Outliers (17) Bias (-2.6 mm/yr) Slope (1.9) Std (.26 mm/yr) - - Outliers () Bias (-2.2 mm/yr) Slope (1.6) Std (.17 mm/yr) - - Outliers (19) Bias (-2.3 mm/yr) Slope (1.) Std (.14 mm/yr) DDK3 - - GRACE [mm/yr] - - GRACE [mm/yr] - - GRACE [mm/yr] - - GRACE [mm/yr] 6 June 212 CGU Annual Meeting, Banff, Canada, June -8, 212 8

9 δh from GRACE&GPS velocities CBN GPS site 6 June 212 CGU Annual Meeting, Banff, Canada, June -8, 212 9

10 Calibrated errors of N, h, and H SuperNet v..3.1 ellipsoidal heights ITRF2(26.) Corrected for GIA Nov7 orthometric heights Only measurements after 1981 True orthometric height Error w.r.t. Rimouski CGG2 geoid heights SNv31 Nov7 - CGG2 Variance-component estimation #224 GPS-on-BMs in Eastern and (h-h-n) Statistics Central Canada Min [m] Max [m] Mean [m] Std [m] 2 outliers removed after 3-sigma test Estimated mean: Nov7 datum shift Estimated scale factors: SuperNet v..3.1:.8±.62; Nov7:.61±.16 CGG2:.39±.46 6 June 212 CGU Annual Meeting, Banff, Canada, June -8, 212

11 Decadal δn/ N/σN N and δh/ H/σH Geoid height Orthometric height Error Error Decadal signal/error Decadal signal/error 6 June 212 CGU Annual Meeting, Banff, Canada, June -8,

12 Conclusions δn due to GIA Scientific datum Value 1.2 mm/yr (peak value in HB) Update frequency yearly Greenland -. to.1 mm/yr yearly Hydrology < 1 mm/yr yearly National datum Value 1.2 cm per decade -. to.1 cm per decade < 1 cm per decade Update frequency decadal? no For scientific and high accuracy applications the geoid height can be updated accordingly to the required accuracy using the δn grid. Since the Canadian levelling network will become obsolete nationally in the future, the geoid stability can be assessed at CACS stations (links to ITRF and IGS) and CBN (crustal deformation monitoring) networks. Maintain the consistency in the vertical component of an accurate geocentric spatial reference system for North America. 6 June 212 CGU Annual Meeting, Banff, Canada, June -8,

13 Acknowledgements GSD, NRCan for geodetic, orthometric and geoid heights data sets D. Argus (JPL) for GEODVEL1b solution Eva Börgens (Universität Bonn and University of Calgary) for the k-filtered GRACE solutions from different processing centres Laura Jensen (Universität Bonn) for the Greenland model GFZ and CSR GRACE solutions from the International Centre for Global Earth Models (ICGEM), GLDAS/Noah model ftp://agdisc.gsfc.nasa.gov/data/s4pa/gldas_subp/gldas_noah_m/ This work is part of the ESA STSE-GOCE Height System Unification project in the WPs (1) unification of the North American height systems with consultants GSD, NRCan, Canada and NGS, NOAA, USA. (2) recommendations for considering temporal variations of heights and geoid for maintenance of a unified height system. 6 June 212 CGU Annual Meeting, Banff, Canada, June -8,

14 Error Back up slides Geodetic height Decadal signal/error 6 June 212 CGU Annual Meeting, Banff, Canada, June -8,

15 Conclusions Error in the observed GIA rate from GRACE reaches.3 mm/yr in the areas with signal maxima Greenland ice loss disturbs the geoid by - to 1 mm/decade. Not significantly large for including in the δn model. Difficulties to separate GIA from hydrologic signals. Hydrologic model deficiencies obscure the secular GIA signal. δh and δn should be modelled by combining GRACE and GPS (AG). GRACEalone rates are subject to leakage errors, errors in applied corrections, theoretical approximations, etc. For vertical datum update, on a decadal scale δn becomes significant after years in the areas south of Hudson Bay. However, change in orthometric heights close to Rimouski is significant compared to the calibrated error. 6 June 212 CGU Annual Meeting, Banff, Canada, June -8, 212

16 Combining GRACE &Terrestrial Data Example: Vertical displacement rates in the Great Lakes area Data sets GRACE DDK2 CSR RL4 rates no GLDAS correction GEODVEL1b GPS vertical velocities joint TG/Altimetry absolute rates Differences in the epochs and reference systems Least-squares adjustment - no constraints Estimated biases w.r.t. the GPS sites GRACE: 2.8 ±.1 mm/yr TGA: 2.4 ±.2 mm/yr TGA station GPS site 9 December 211 AGU Fall Meeting 211, San Francisco, USA 16

17 Combining GRACE &Terrestrial Data 9 December 211 AGU Fall Meeting 211, San Francisco, USA 17

18 Secular Trend and Hydrology Signal (cont) CSR RL4 km Least-squares fitting Standard PCA/EOFs Rotated PCA/EOFs 9 December 211 AGU Fall Meeting 211, San Francisco, USA 18

19 Secular Trend and Hydrology Signal Leastsquares GLDAS Rotated PCA/EOFs Standard PCA/EOFs 9 December 211 AGU Fall Meeting 211, San Francisco, USA 19

20 Combining GRACE &Terrestrial Data (cont) Ndot and vertical crustal displacement rates computed from an optimal combination of GRACE, GPS and absolute gravity 1. Gravity rates 2. Gravity-to-height ratio Trend component: multiquadrics GRACE-observed displacement GPS velocities Terrestrial gravity rates INPUT 1. Gravity rates 2. Gravity-to-height ratio 3. Signal covariance matrices from GPS data 1. Gravity rates 2. Gravity-to-height ratio 3. GPS data Signal component: Stokes integration Levelled vertical displacement Absolute vertical displacement Terrestrial gravity rates Geoid rates OUTPUT 1. Gravity rates 2. Gravity-to-height ratio 3. GPS data 4. GRACE-observed displacement VCE and outlier detection 9 December 211 AGU Fall Meeting 211, San Francisco, USA 2