On the relationship between tropospheric CO and CO₂ during KORUS-AQ and its role in constraining anthropogenic CO₂

Abstract. While the complementarity of CO data in monitoring CO₂ from fossil-fuel combustion (ffCO₂) is widely known, a rigorous demonstration of its use in reducing uncertainties on top-down regional ffCO₂ emissions is still warranted. Here, we report a case study investigating the regional covariation of observed and modeled abundances of CO, CO₂, and ffCO₂ and demonstrating its implication to joint CO:CO₂ inversions. We use data from a recent aircraft field campaign (KORUS-AQ) conducted over Korea and neighboring regions on May 2016 for this case study. We use the Community Atmosphere Model with Chemistry (CAM-Chem) to simulate CO, CO₂, ffCO₂ and associated source tags, using a posteriori fluxes from global CO₂ flux inversions and CO emissions independently calibrated against CO data. Among other model-data comparisons, CAM-Chem simulations show an underestimation in CO₂ (1 ppm), CO (24 ppb) and ffCO₂ (1 ppm) against aircraft measurements. These are all within the range of model and data uncertainties. Although the overall observed enhancement ratio, ΔCO⁄ΔCO₂ (~ 13.3 &om; 0.21 ppb/ppm), is well captured by CAM-Chem (~ 13.8 ± 0.23 ppb/ppm), we find an overestimation (29 ppb/ppm) for air samples between 2 to 3 km, where East Asian influence is substantial (35 %). The contribution of ffCO₂ from Korea and Japan is smaller (30 %) and localized below 3 km, suggesting that regional ffCO₂ and background and non-ffCO₂ cannot be neglected in interpreting observed enhancements in this region. These spatial variations translate in the joint CO:CO₂ inversion to increases in a posteriori ffCO₂ estimates from East Asia (27 % ± 24 %) and Korea and Japan (9 % ± 17 %). This is consistent (albeit larger in 1-sigma uncertainty) with our estimate using ¹⁴CO₂ data (27 % ± 9 % and 10 % ± 3 %, respectively). In contrast, the inversion using only CO₂ data shows a decrease by ~ 5 % ± 27 % in East Asia and ~ 6 % ± 19 % in Korea and Japan. Our results show that inversions using both CO₂ and CO can be an effective approach in constraining ffCO₂ when the regional variations of CO and CO₂ relationships are appropriately accounted for. Although this further points to the potential of augmenting current observing system of CO₂ with CO for global inverse analyses of ffCO₂ from different regions of the globe, we highlight the need to verify the spatiotemporal distribution of the covariation of CO with CO₂ in both regional and global models. We caution its use for constraining local ffCO₂, unless the spatiotemporal a priori flux distribution and surface processes are reasonably represented, as they may confound the analysis. These have important implications on inversion studies using columnar data from satellite observations, especially for regions lacking necessary verification measurements.