A systematic bias in float pH leads to overestimation of derived pCO2 and underestimation of carbon uptake by the Southern Ocean

Abstract The carbon flux estimated from biogeochemical Argo float data indicates a lower annual carbon uptake by the Southern Ocean compared to fluxes derived from other observations (e.g., ship and aircraft measurements). The root cause of this discrepancy remains controversial, with growing evidence suggesting that potential biases in float-derived pCO2 may be a plausible explanation. Here, we perform a multi-variable comparison of vertical profiles between float- and ship based-data and reveal consistent discrepancies in pH, pCO2 and dissolved inorganic carbon, which are not found in other variables such as dissolved oxygen, nitrate and total alkalinity. Our findings are consistent with a previously unrecognized negative bias in float pH driving a positive offset in float-derived pCO2. The float-derived surface pCO2 is, on average, biased high by 15 ± 3 µatm compared to ship data, representing a larger magnitude of bias than previously recognized. Biases exist in both surface and deep waters, including old deep waters containing minimal anthropogenic carbon. A more sophisticated adjustment for float pH, involving multiple cross-reference depths, may be required for accurate estimation of air-sea CO2 exchange in the Southern Ocean.