Within this project a new high-quality gravity field solution should be generated for Austria, which overcomes the inconsistencies between previous geoid solutions and geoid heights derived from GPS/leveling campaigns. The computation will optimally combine the complementary data types of satellite observations and all available terrestrial gravity field measurements in Austria and neighbouring countries. For this purpose, the Remove-Compute-Restore technique will be adapted in order to avoid a double consideration of the topographic-isostatic masses when performing long- and short-wavelength signal reductions. Furthermore, the atmospheric mass effects on gravity observations will be considered within this project.
The Least Squares Collocation (LSC) approach will serve as reference method for the gravity field computation. Improvements are aimed for with regards to the choice of a proper universal covariance function and the optimal weighting of the different observation types. In this context, further developments concerning the integration of the now available full variance/covariance information of the recent global gravity field models will be necessary.
The unique gravity gradient observation type of GOCE's gradiometer instrument delivers information of the gravity field particularly in the medium wavelengths, where the spectral overlap of global satellite data and local terrestrial data occurs. For an optimum regional gravity field determination it is therefore reasonable to use the GOCE gradients as in-situ observations and directly combine them with terrestrial data. However, there are still some issues that are aimed to be clarified on the course of this project for a practical applicability using GOCE gradients as in-situ observations.
Alternatively to LSC, a Gauss-Markov model with parametrization as Radial Basis Functions will be implemented. The optimum weighting of the observation groups will be performed by Variance Component Estimation (VCE). By the new methodological developments of this proposed project, an increasing number of observations can be included in the calculations and a downsampling of the available data, as it is required in LSC due to computational limitations, will no longer be necessary. Compared to the standard LSC approach, the information content which can be incorporated in a new geoid solution can therefore be increased drastically.
The achieved results will be verified by cross-validation methods and by comparing with independent GPS/leveling observations.