Variations of the three-dimensional atmospheric CO2; implications for carbon budget from model simulations and aircraft measurements

Three-dimensional structure of atmospheric carbon dioxide (CO2) is essential for quantification of surface fluxes, and to illustrate effect of transport model error while estimating those by inverse modeling. We performed forward transport simulations using multiple transport models and variants for comparison with the atmospheric CO2 data obtained from an aircraft measurement network: CONTRAIL, covering a wide range of latitude, longitude and height. Four transport models (NICAM-TM, ACTM, NIES-TM, MJ98-CDTM) and three sets of flux data (one TransCom inversion flux and two bottom-up-based fluxes) are used in this transport model intercomparison (TMI).TMI simulations and CONTRAIL CO2 concentration for the period of 2006—2007 are analyzed. While general features in the three-dimensional CO2 concentration fields are captured well by the models, there exist statistically significant differences deciphering salient characteristics of the models. Large impacts of the surface flux on three-dimensional CO2 fields are found, confirming the implications of the observations in the free troposphere for deriving surface fluxes. Our results suggest remote sensing observations from middle-upper troposphere layers are likely to be useful, along with the ongoing and planned satellite missions. In the summer, all the models transporting inverted flux overestimated latitudinal gradients in the free troposphere compared to the CONTRAIL measurement, while there was almost no systematic bias in the marine boundary layer. We infer that terrestrial northern hemispheric sink is weaker in the inversion flux, derived using the in situ measurement network and 12 TransCom models. Our results also suggest that land-ocean partitioning of global total CO2 flux is robustly constrained by the newly developed forward models and the aircraft measurements.