Reconciling ocean mass changes from 20 years of GRACE and GRACE Follow On observations

The total mass change of the Earth's land surface precisely offsets the combined changes in the atmosphere and oceans, resulting in a net-zero change for the entire system (land+ocean+atmosphere). Closing the ocean mass budget is crucial for understanding current and future sea-level changes. Recent efforts to reconcile ocean mass observations from GRACE and GRACE-Follow On satellites (hereafter unitedly referred to as GRACE) with both steric-corrected altimetry and and land/ice to ocean estimates have revealed a discrepancy in the mass budget (Wang et al, 2022; Barnoud et al, 2022). This finding indicates a concerning misalignment in our global observation system or understanding of earth mass transport. This study uses GRACE-independent estimates/models of land surface mass changes to validate 20 years of GRACE observations. By calculating the monthly Gravitational, Rotational, and Deformational (GRD) response to 20 years of land mass changes, we reconstruct the global, regional, and seasonal ocean mass changes observed by GRACE from 2003 to 2022. Over the 20-year period, the ocean mass reconstruction aligns well with the GRACE observations. However, a significant deviation emerges after 2020, with the reconstruction showing a larger ocean mass change than GRACE. We demonstrate that this deviation is likely caused by an underestimation of Western Africa precipitation in the ERA5 reanalysis, commonly used by hydrological models to estimate changes in land water storage. Land mass observations from GRACE further confirmvthis underestimation and shows great alignment between models and observations when excluding sub-Saharan Africa. Our results show a global agreement between GRACE and GRD-induced ocean mass changes, suggesting that the misalignment between GRACE and steric-corrected altimetry is likely due to errors in the ARGO observing system. A reported 'salinity-drift' is the primary source of error, and together with an error in the wet path delay originating from drift in the radiometer of the Jason-3 satellite explains most of the post-2016 difference between GRACE and steric-corrected altimetry is identified. The remaining differences likely originate from GIA and/or Argo-biases.