The Bayesian inverse method, combined with measurements of atmospheric carbon dioxide (CO2) and a transport model, can serve as an independent verification approach to improve the precision of emission estimates. This study utilized the Bayesian inverse model, along with ground- and space-based measurements, to validate CO2 emissions in Seoul. A Bayesian inverse modeling framework was developed, integrating crucial input data such as anthropogenic CO2 emissions, biogenic CO2 fluxes, atmospheric CO2 measurements, a Lagrangian transport model, and error covariances for both prior emissions and observations. The averages of posterior emissions decreased after the inversion run, with a correction of approximately -8.69%. This suggests that the prior emissions were overestimated. There was an average 9.7% reduction in posterior emission uncertainties compared to prior uncertainties. The most substantial reductions in uncertainty were observed in areas with concentrated observation sites. The performance of the inverse model was thoroughly investigated through sensitivity analysis, encompassing different background representations, prior uncertainty levels, temporal and spatial uncertainties, and observational network configurations. Additionally, we quantified spatiotemporal changes in CO2 emissions due to COVID-19. The abundance of ground and space observations in Seoul provided robust constraints on urban CO2 emissions, allowing for an objective evaluation of the effectiveness of carbon reduction policies.

This work was supported by Korea Environmental Industry & Technology Institute (KEITI) through "Project for developing an observation-based GHG emissions geospatial information map", funded by Korea Ministry of Environment(MOE)(RS-2023-00232066).