For high resolution gravity field modelling in Germany, global geopotential models are combined with terrestrial gravity field data and topographic information from high resolution digital terrain models. In this context, it is vitally important to assess the quality of recent global geopotential models from GOCE and other satellite missions, GPS-levelling data as well as high resolution gravimetric quasigeoid models with an aspired accuracy of 1 cm. Within these analyses, the differences of the existing gravity field data sets have to be enlightened.

For this purpose, a unique data set of astrogeodetic vertical deflections was observed in the period from 2006 to 2010 with the zenith camera system TZK2-D. The observations were carried out on 394 stations in total along two profiles in north-south and west-east direction with a spacing of 3-4 km between adjacent stations and total lengths of approx. 600 km. The method of astronomical-topographic levelling was used to compute quasigeoid heights. With regard to the aspired accuracy of 1 cm/100 km, systematic effects within the astrogeodetic gravity field modelling had to be analysed, i.e. particularly, the effects of the underlying reference frames as well as temporal variations of the observations. Commonly used approximations within astrogeodetic gravity field modelling were analysed using rigorous formulas.

Considering the identified systematic effects, the comparison between the astrogeodetic vertical deflections and corresponding values from the gravimetric quasigeoid model EGG2008 (Denker, 2013) reveals the stated accuracy of the astrogeodetic vertical deflections of 0.08". By means of comparisons with the astrogeodetic data, the accuracy of vertical deflections from global geopotential models is estimated to be between 0.18" and 0.40" for the ultra-high resolution model EGM2008 and 0.20" up to a spatial resolution of 100 km for the recent GOCE models of the third generation, being fully compatible with the relevant error estimates of the models. The comparisons between the height anomalies from the astronomical-topographic levellings and GPS-levelling data as well as high resolution gravimetric quasigeoid models in Germany show an agreement of 1.2 cm to 2.9 cm RMS. The short wavelength differences can be assigned to the ellipsoidal heights of the GPS-levelling data, while the analyses of the long wavelength differences of a few centimetres over several 100 km is very complex, as the differences include uncertainties of all involved data sets. In this context, the astrogeodetic quasigeoid solutions provide a valuable additional control. On the whole, the results reveal the high quality of the existing gravity field data sets in Germany, including the astrogeodetic data set determined by this research.