Department of Earth System Science

Recent inverse analyses constraining carbon fluxes using atmospheric CO₂ observations have assumed that the CO₂ source from atmospheric oxidation of reduced carbon is released at the surface rather than distributed globally in the atmosphere. This produces a bias in the estimates of surface fluxes. We used a three-dimensional (3D) atmospheric chemistry model (GEOS-CHEM) to evaluate the magnitude of this effect on modeled concentrations and flux estimates. We find that resolving the 3D structure of the atmospheric CO₂ source, as opposed to emitting this reduced carbon as CO₂ at the surface, yields a decrease in the modeled annual mean interhemispheric gradient (N-S) of 0.21 ppm. Larger adjustments (up to −0.6 ppm) are apparent on a regional basis in and downwind of regions of high reduced carbon emissions. We used TransCom3 annual mean simulations from three transport models to evaluate the implications for inversion estimates. The main impacts are systematic decreases in estimates of northern continental land uptake (i.e., by 0.22 to 0.26 Pg C yr⁻¹), and reductions in tropical land carbon efflux with smaller changes over oceans and in the Southern Hemisphere. These adjustments represent a systematic bias in flux estimates, accounting for changes of 9 to 27% in the estimated northern land CO₂ sink for the three models evaluated here. Our results highlight the need for a realistic description of reduced carbon emission and oxidation processes in deriving inversion estimates of CO₂ surface fluxes.