Accurate quantification and monitoring of urban CO₂ emissions are essential to formulating emission mitigation strategies and assessing their effectiveness. Space-borne high-resolution mapping of atmospheric CO₂ concentrations is expected to help objectively understand carbon emissions, in support of constructing a global greenhouse gas monitoring system. In this work, we evaluate the capability of constraining fossil-fuel CO₂ (FFCO₂) emissions for a medium-sized city, Xiamen, China, using the total column CO₂ mole fraction (XCO₂) retrievals from the Orbiting Carbon Observatory-3 (OCO-3) in Snapshot Area Map (SAM) mode. Based on five OCO-3 overpasses filtered from nearly four years of data, the top-down inversion method is used to infer the urban CO₂ emissions, which combines XCO₂ measurements and atmospheric transport modeling using the WRF-Chem model at a horizonal resolution of 1 km, coupled with prior anthropogenic CO₂ emissions and biogenic fluxes. An intercomparison is performed over the inversely estimated emissions and the estimates from seven inventories based on downscaling and bottom-up methodologies and activity data at different levels. The results suggest an ensemble mean of the inversely estimated emissions of 1.85×10⁴ ± 1.60×10³ tC/d over Xiamen, which is higher than the average of the inventories by more than 20 %. The posterior error reduction varies from case to case, which can be limited by the number of observational constraints, transport model errors, and unfavorable meteorological conditions that diminish the representativity for urban emissions by the XCO₂ sampling. The different quantification methods yield an average of 1.60×10⁴ tC/d, with the spread of 45 % (±1σ), which suggests that evaluations of the emission inventories with more constraints from atmospheric CO₂ measurements are necessary to acquire an objective emission estimate at the city scales. 

CO2 emissions,anthropogenic sources,inversion,Remote sensing (RS)