OceanSODA-UNEXE: A multi-year gridded Amazon and Congo River outflow surface ocean carbonate system dataset

Abstract. Large rivers play an important role in transferring water and all of its constituents including carbon in its various forms from the land to the ocean, but the seasonal and inter-annual variations in these riverine flows remain unclear. Satellite Earth observation datasets and reanalysis products can now be used to observe synoptic-scale spatial and temporal variations in the carbonate system within large river outflows. Here we present the OceanSODA-UNEXE time series, a dataset of the full carbonate system in the surface water outflows of the Amazon (2010–2020) and Congo Rivers (2002–2016). Optimal empirical approaches were used to generate gridded Total alkalinity (TA) and dissolved inorganic carbon (DIC) fields in the outflow regions. These combinations were determined by equitably evaluating all combinations of algorithms and inputs against a matchup database of in situ observations. Gridded TA and DIC along with gridded temperature and salinity data enable the calculation of the full carbonate system in the surface ocean. The algorithm evaluation constitutes a Type A uncertainty evaluation for TA and DIC where model, input and sampling uncertainties are considered. Total combined uncertainties for TA and DIC were propagated through the carbonate system calculation allowing all variables to be provided with an associated uncertainty estimate. In the Amazon outflow, the total combined uncertainty for TA was identified as 36 μmol kg⁻¹ (weighted RMSD 35 μmol kgkg⁻¹ and weighted bias 8 μmol kg⁻¹ for n=82) and for DIC was 44 μmol kg⁻¹ (weighted RMSD 44 μmol kg⁻¹ and weighted bias −6 μmol kg⁻¹ for n=70). The spatially averaged propagated uncertainties for the partial pressure of carbon dioxide (pCO₂) and pH are 85 μatm and 0.08 respectively, where the pH uncertainty is relative to an average pH of 8.19. In the Congo outflow, the combined uncertainty for TA was identified as 29 μmol kg⁻¹ (weighted RMSD 28 μmol kg⁻¹and weighted bias 6 μmol kg⁻¹ for n=102) and for DIC was 40 μmol kg⁻¹ (weighted RMSD 37 μmol kg⁻¹and weighted bias −16 μmol kg⁻¹ for n=77). The spatially averaged propagated uncertainties for pCO₂ and pH are 74 μatm and 0.08 respectively, where the pH uncertainty is relative to an average pH of 8.21. The combined uncertainties in TA and DIC in the Amazon and Congo outflows are lower than the natural variability their respective regions allowing the time varying regional variability to be evaluated. Potential uses of these data would be for assessing the spatial and temporal flow of carbon from the Amazon and Congo rivers into the Atlantic and for assessing the riverine driven carbonate system variations experienced by tropical reefs within the outflow regions.