Understanding the errors in ocean tide models for improved GRACE gravity field recovery

The Gravity Recovery And Climate Experiment (GRACE) and its successor, the GRACE Follow-On (GRACE-FO), observe temporal variations in the Earth’s gravity field caused by mass redistribution. The primary observation is range and range rate (Level-1), which are processed to reach global maps of mass change (Level-3). Various background models are used in Level-1B processing; however, these are not perfect, and their uncertainties affect the gravity field solutions. Among the time-variable gravity signals, high-frequency atmospheric and non-tidal ocean variations, as well as ocean tides, are generally undersampled, resulting in short-period signals aliasing onto longer periods in the recovered gravity field, which becomes a major error component. In this study, we analyzed the impact of various ocean tide models on gravity field recovery in both spatial and spectral domains. Furthermore, we evaluated the differences in the ocean tide models in terms of tidal amplitudes and gravity potential, which is computed along the footprints of GRACE-A at the orbital pass-time scale. The derived inter-model differences from orbital simulations are used to compute uncertainty information that can be utilized later in the stochastic modeling of the gravity field. The results indicate that significant differences persist among ocean tide models in accurately resolving tidal signals, especially at higher degrees. Furthermore, we find that modelling aliasing error at monthly mean scale is providing a conservative estimate compared to modelling errors along the orbit while considering Ocean tides at the satellite pass-time.  The error in the global geoid height ranges ±1.2 mm when considering orbit pass-time information, compared to ±0.12 for the case where we compute errors at monthly mean scale. The major differences were observed over polar regions, ice sheets, shallow waters, and coastal areas. Hence, we expect that modelling ocean tide de-aliasing errors at orbit pass-time scales instead of at monthly scales could reduce GRACE(-FO) uncertainties.