Refined computation strategies for the new German Combined Quasigeoid GCG2016
20/09/2016 | 17:15 | Session 3: Local/regional geoid determination methods and models
Author(s): Joachim Schwabe, Gunter Liebsch and Uwe Schirmer
Joachim Schwabe, Gunter Liebsch and Uwe Schirmer
"For over one decade, the German Combined Quasigeoid (GCG) models have been the official transformation surface between ellipsoidal heights from GNSS and leveling heights in Germany. These models are based on the solutions of two data and computation centers (BKG Leipzig, Institut für Erdmessung (IfE) at the University of Hannover), where gravimetric quasigeoids are individually derived and fitted to GNSS leveling data w.r.t. the national reference frames.
The new model GCG2016 is to be released and adopted in late 2016. It will be consistent with the latest official reference frame realizations in Germany: DREF91(R2016) – which is to be adopted as new national realization of the ETRF2000 in Germany – and DHHN2016 (new leveling network).
We give an overview on preliminary results computed at the BKG. Improvements regarding method and underlying gravity data are summarized. Special focus is given to the following major modifications in the applied remove-restore strategy:
- use of a high-resolution global gravity field model (GGM) at full scale (EIGEN-6C4)
- improved long-wavelength stability of the residual quasigeoid solution by introducing the GGM spherical harmonic coefficients as direct observations in the pointmass estimation
- residual terrain corrections based on a joint 1 arcsec digital elevation model (DEM), where the national DEM has been augmented with bathymetry from the GEBCO model and, for the first time, lake bathymetry data of Lake Constance
Thanks to these modifications, and to a new and highly and consistent GNSS leveling dataset, the agreement between gravimetric quasigeoid and GNSS leveling could be further increased. With respect to a three-parameter (plane) fit, a standard deviation of 1.0 cm could be achieved, and the remaining residuals are limited to a few centimeters. These residuals are subsequently interpolated onto a corrector grid. Based on a simple kriging approach the predicted interpolation error over the German mainland ranges from 2 to 8 mm. Therefore, we conclude that the GCG2016 can be used to realize GNSS leveling at the 1 cm level or better.