A Robust Method for Calculating Carbon Dioxide Emissions From Cities and Power Stations Using OCO-2 and S5P/TROPOMI Observations

Abstract

We introduce a new method for calculating the carbon dioxide (${\text{CO}}_2$) emissions from point sources (e.g., power stations) and cities using the cross-sectional flux method constrained by space-based ${\text{CO}}_2$ and nitrogen dioxide (${\text{NO}}_2$) observations. First, we derive a proxy estimate for ${\text{CO}}_2$ enhancements from ${\text{NO}}_2$ observations through linear regression near the plume cross-section. Then, we fit a Gaussian function to the resulting ${\text{NO}}_2$-based ${\text{CO}}_2$ enhancement data. We apply this method to data from the Orbiting Carbon Observatory-2 (OCO-2) and the Sentinel-5 Precursor TROPOspheric Monitoring Instrument (S5P/TROPOMI) starting from May 2018. The method is tested on the Matimba and Medupi power stations in South Africa, as well as the cities of Madrid (Spain), Las Vegas (USA), and Baghdad (Iraq). The corresponding mean ${\text{CO}}_2$ emission estimates are 49 $\pm$ 17 Mt/yr, 22 $\pm$ 10 Mt/yr, 30 $\pm$ 8 Mt/yr, and 41 $\pm$ 19 Mt/yr, respectively. The results show that the method is robust and can be applied to challenging cases where ${\text{NO}}_2$ data helps constrain the fit. The proxy approach allows evaluation of 17 additional scenes (out of 53), reducing the average error of individual emission estimates from approximately 30%–40% to 22%–25% compared with the traditional method. Furthermore, we highlight the significant potential of satellite data to uncover discrepancies in reported emission estimates, for example, by identifying under-reported or missing emission sources. The proposed approach can be extended to other case studies and applied to future satellite missions with joint ${\text{CO}}_2/{\text{NO}}_2$ observations, such as CO2M, GOSAT-GW, TanSat-2, and TANGO.