Expanding urban populations and the significant contribution of cities to global fossil-fuel CO2 (CO2ff) emissions emphasize the necessity of achieving independent and accurate quantifications of the emissions from urban area. In this paper, we assess the utility of total column dry air CO2 mole fraction (XCO2) data retrieved from NASA’s Orbiting Carbon Observatory 2 (OCO-2) observations to quantify CO2ff emissions from cities. Observing System Simulation Experiments 15 (OSSEs) are implemented by forward modeling of meteorological fields and column XCO2. The impact of transport model errors on the inverse emissions estimates is examined for two ―plume cities‖ (Riyadh, Cairo) and a ―basin city‖ (Los Angeles metropolitan region, LA). The pseudo data experiments indicate convergence of emission uncertainties related to transport model errors with increasing amount of observations. The 1-σ uncertainty of emission estimates is constrained to approximately 15%/5% with about 10 pseudo tracks for plume city/basin city. The systematic wind speed biases in simulated 20 wind fields for LA lead to overestimations in total CO2ff emission, which require data assimilation to improve high-resolution atmospheric transport. The contribution of biogenic fluxes gradients in urban and rural area of Pearl River Delta metropolitan region in China are examined by simulations with biospheric fluxes imposed by the Net Ecosystem Exchange (NEE) from multiple terrestrial biosphere models, which show about 24±21% (1σ) and 19±15% (1σ) contributions to the total XCO2 enhancements for the two cases examined. The representations of transport model errors for the emission optimization are 25 examined for Riyadh, Cairo and LA in real cases. The determination of background XCO2 is discussed for LA by using constant and simulated background with biospheric fluxes included, demonstrating the need of careful consideration of the variations in background XCO2 for identifying concentration enhancements due to fossil fuel emissions.