This master thesis deals with the stochastic modelling of daily GRACE gravity field solutions with the aim of improving the quality of the ITSG-Grace time series. The satellite-based gravity field missions GRACE and GRACE-FO enable the global recording of time-variable mass displacements in the Earth system, whereby high-frequency processes in the atmosphere, ocean and hydrosphere in particular can lead to aliasing artefacts in the monthly solutions. In this work, an approach is pursued in which the uncertainties of the models used are not neglected, but rather specifically integrated into the evaluation. Among other things, the error model AOe07 for AOD1B-RL07, the hydrological model LISFLOOD and ice signals from the ESA Earth System Model (ESM) are used for this purpose. This additional information helps to evaluate the measurement data according to their uncertainty and to better adapt the modelling to the actual physical processes. The use of daily solutions in combination with additional stochastic information therefore promises a reduction in systematic modelling errors. The underlying Kalman smoothing approach allows consistent modelling of the temporal dynamics and is based on the knowledge of statistical properties (e.g. covariances) of the expected gravity field signal. By comparing different model combinations, regularisation approaches and process configurations, the aim is to identify those settings with which the most precise and physically consistent daily GRACE gravity field solutions can be achieved.

Co-author: Univ.-Prof. Dr.-Ing. Torsten Mayer-Gürr