Toward robust and consistent regional CO2 flux estimates from in situ and spaceborne measurements of atmospheric CO2

We evaluate the robustness and consistency of global and regional posterior CO2 flux estimates for 2010 inferred from two versions of bias-corrected CO2 column retrievals from the Japanese Greenhouse Gases Observing Satellite (GOSAT). Six satellite-based inversions, generated from three atmospheric transport models and two independent Bayesian inference algorithms, facilitate a rigorous investigation of the uncertainty of the inverted fluxes. This ensemble shows hemispheric and regional differences in posterior flux estimates that are beyond 1 sigma uncertainties and in some regions are unrealistic. We recognize the importance of these satellite data in further understanding the contemporary carbon cycle but we argue that more resources should be invested in characterizing the errors of the prior fluxes, the systematic errors of the retrievals, and the systematic errors of the transport models, to improve confidence in the resulting posterior fluxes. Key Points Nine global CO2 atmospheric inversions are compared The CO2 fluxes inferred from the satellite data show limited realism The satellite-inferred fluxes show large sensitivity to all input information