Since March 2002 the two GRACE satellites observe with unprecedented high precision the time varying gravity signal caused by the sum of all mass variations within the Earth system. Since the water mass variation of the continental hydrosphere is the most uncertain component of the global water cycle (see e.g. the 4th assessment report of IPCC, 2007) it is one of the most challenging problems to separate, to identify, and quantify the mass variations of the individual subsystems ocean, atmosphere, and continental hydrosphere.
Using empirical orthogonal functions (EOF) derived from geophysical models it has now been proven that the decomposition is possible. The result is by no way obvious, because atmospheric mass variations overlay the two other subsystems, ocean and continental hydrosphere, and these two disjoint subsystems cause a leakage into each other. The ability to separate the individual mass variations was first proven by a closed loop simulation (Schmeer et al., 2008a). Starting with geophysical models, GRACE observations were synthesized, and the separation of these integral observations led to estimated time coefficients (so called Principal components, PC) coinciding with those of the predefined models. The coincidence remains excellent if other geophysical models are used.
In a second step the separation was performed with real, unfiltered GRACE data (Schmeer et al., 2008b). EOFs were derived from monthly time series of (i) the ECMWF surface pressure data, (ii) the Ocean Model for Circulation and Tides (OMCT, Thomas, 2002), and (iii) the WaterGAP Hydrology Model (WGHM, Döll et al. 2003). A time series of 67 monthly GRACE gravity field models with atmospheric and oceanic background models reconstructed were taken as input (Flechtner, 2005). For hydrology and atmosphere the reconstruction shows a good agreement with high correlations (88.3 and 95%) between original and the reconstructed PCs (see Figure 1). For hydrology there is a clear phase shift for the annual signal with GRACE observing the extrema about 1 month later than predicted by WGHM. It is also to be investigated why the correlation for the subsystem ocean (not shown) is worse and remains below 50%.