Item

Abstract

De-aliasing based on geophysical background models (BM) allows for a reduction of high-frequency, high-amplitude signal components in GRACE and GRACE-FO data processing. These are primarily related to the ocean tides (OT) and non-tidal variations within the atmosphere and the oceans (AO), and would otherwise superimpose the target signals stemming e.g. from the hydro- or the cryosphere. In the course of previous studies, it was shown both in closed-loop simulations (Abrykosov et al. 2021) as well as in real data processing (Panafidina et al. 2023, in preparation) that the incorporation of OT BM errors in terms of error variance-co-variance matrices (VCM) yields an enhanced gravity retrieval performance. In this contribution, we discuss the obvious next step, which is the consideration of stochastic properties of the underlying imperfections of the AO models. This task is not as trivial as in case of OT, since the spatial error features significant time variations. Thus, we first derive a static error VCM for the AO model and then refine it by adding temporal variations. The added value of both approaches is assessed in numerical closed-loop simulations by rigorously adding the error information into the data processing scheme. As the time-variable error VCM may quickly become too large to be handled, explicit investigations towards determining the optimal decorrelation period are carried out and discussed with regard to trade-off between computational efficiency and gravity retrieval accuracy.