Impact of submesoscale variability in estimating the air-sea CO₂ exchange: Results from a model study of the POMME experiment

A high-resolution ocean biogeochemical model is used to estimate oceanic pCO₂ and air-sea CO₂ flux in the NE Atlantic. The model is validated against shipboard and Carioca drifting float data acquired during the POMME experiment. Between winter and spring, the seasonal variability is characterized by a rapid drawdown of pCO₂ of ∼20 μatm associated with the phytoplankton bloom driving CO₂ uptake by the ocean. The model reveals that this uptake propagates northward in response to the northward propagation of the bloom. More remarkably, this study demonstrates intense variability of the carbon system at the submesoscale. Our model predicts filamentary structures of pCO₂ that show gradients of 25 μatm over 20 km, consistent with observations from Carioca drifting floats. This submesoscale variability is similar in magnitude to the mean seasonal drawdown. Lagrangian diagnostics suggest that pCO₂ small-scale structures are shaped by horizontal stirring of large-scale gradients created by the bloom northward propagation. We compared air-sea flux derived from model pCO₂ and from observed pCO₂ and estimated the error due to data undersampling to ∼15 to 30%. Results from a simulation at coarser resolution showed that the impact of model resolution on air-sea CO₂ flux is only ∼5%. This suggests that the submesoscale variability of pCO₂, although large in amplitude, accounts for small modulation of the net air-sea CO₂ flux in this region.