A water mass transformation method applied to diagnosing ocean carbon uptake

The ocean is a strong sink for anthropogenic CO₂, absorbing around a quarter of emissions since the industrial era. Quantifying the ocean carbon sink is necessary for constraining the global carbon budget; however, discrepancies remain between estimates of the ocean carbon sink over the last 30 years from observation-based data products and those from numerical models. Moreover, larger regional uncertainties highlight the need for a better understanding of the drivers of ocean carbon sink variability, to help improve models and to better constrain future climate projections. A comprehensive understanding of the sink must include knowledge of (1) the air-sea flux of CO₂, (2) the accumulation of carbon in the ocean interior, and (3) how it is redistributed within the ocean by changes in the physical circulation. This characterisation is typically achieved using numerical models, which are constrained by resolution and the need to parameterise processes including physical mixing at the sub-grid scale.

We present a novel method for characterising the ocean carbon sink from a combination of oceanographic datasets, and for reconciling our knowledge of the ocean’s uptake of CO₂ with that of interior carbon storage rates. Our Optimal Transformation Method (OTM) uses a water mass framework to diagnose the transport and mixing of tracers such as heat, salt, and carbon consistent with observed interior changes and estimates of boundary forcings. The water mass framework has the advantage that the transport and mixing of conservative tracers are diagnosed exactly, with no need for parameterisation. We validate OTM using outputs from a data-assimilating biogeochemical ocean model and demonstrate its ability to recover the model’s ‘true’ air-sea CO₂ fluxes when initialised with biased priors. OTM reduces root-mean-squared errors between diagnosed air-sea CO₂ fluxes and the model truth from prior to solution by up to 71%, while simultaneously estimating inter-basin transports of heat, freshwater, and carbon consistent with the model. Following successful validation, we apply OTM to a combination of observational data products to diagnose estimates of the ocean’s uptake and redistribution of carbon since 1990, utilising reanalyses of air-sea heat and freshwater fluxes, interior temperature and salinity, air-sea CO₂ fluxes, and machine-learning reconstructions of interior ocean carbon.