CO 2 is one of the most significant trace gases in the atmosphere due to its role in global climate change (IPCC, 2013). Thus, estimating accurate carbon budgets is an important task in understanding the carbon cycle (Friedlingstein et al., 2019). In an atmospheric inversion of CO 2 , a transport model is used as a forward model with prescribed or forecasted meteorology to transform surface CO 2 flux information into atmospheric CO 2 concentrations (or mole fractions of dry air) (Ciais et al., 2010). The recent availability of high horizontal resolution transport models at global and regional scales (e.g., Agustí-Panareda et al., 2019;Feng et al., 2016;Kim et al., 2020) now permits the variability of atmospheric CO 2 to be simulated at higher spatial and temporal resolutions than before. At the same time, the CO 2 observation network has expanded considerably so that high resolution models constrained by a dense network of surface, aircraft, and satellite-based measurements which cover broad and high temporal frequencies may be able to retrieve surface CO 2 fluxes at higher spatio-temporal scales with an inverse model. However, systematic differences in transport models continue to play an important role in errors and uncertainties of estimated surface CO 2 fluxes obtained through atmospheric inversions (