Short-latency monitoring of continental, ocean- and atmospheric mass variations using GRACE intersatellite accelerations

GRACE K-band intersatellite range observations have been processed in combination with precise orbit trajectories from reduced dynamic modelling and GNSS processing. These data have been inverted to mass equivalent quantities (water heights) by using physically motivated constraints in order to reduce well-known instabilities. In a first instance, the satellite observations are reduced using geophysical background modelling and implicitly continued downward using a rigorous formulation in terms of reproducing kernel functions which map time variable mass variations from Earth’s fluid envelops to remotely sensed observations. The space-wise observation equations are well suited for a Kalman-filter measurement update approach which enables an enhanced temporal resolution of subweekly to daily time-series. For stochastic signal prediction and process model derivation, spatio-temporal correlations of mean hydrological and atmospheric signals have been used. It is shown that the GRACE dynamic measurements are converted to in situ observables, thus facilitating the combination with other data sources such as, for example higher resolution remote sensing data and/or available ground data. Through the consideration of full signal and measurement error covariances in a Kalman filter process, the effects of aliasing from poorly modelled or insufficiently available background information can be mitigated. We present the theoretical rational of the method, together with results for daily Kalman filtered solutions, for the full duration of the GRACE mission and numerical tests and comparisons to available operational GRACE products and inundation volumes. Finally, the major findings of these computations are listed.