Quantifying CO2 emissions of power plants with the CO2M mission

In this study, we analyse the capability of the Copernicus CO₂ monitoring (CO2M) satellite mission to quantify the CO₂ emissions of individual power plants, which is one of the prime goals of the mission. The study relies on synthetic CO₂ and NO₂ satellite observations over parts of the Czech Republic, Germany and Poland and quantifies the CO₂ and NO_x emissions of the 15 largest power plants in that region using a data-driven mass-balance approach.

The synthetic observations were generated for six CO2M satellites based on high-resolution simulations of the atmospheric transport model COSMO-GHG. To identify the emission plumes, we developed a plume detection algorithm that identifies the location, orientation and extent of multiple plumes from CO2M's NO₂ observations. Afterwards, a mass-balance approach was applied to individual plumes to estimate CO₂ and NO_x emissions by fitting Gaussian curves to the across-plume signals. Annual emissions were obtained by interpolating the temporally sparse individual estimates applying a low-order spline fit.

Individual CO₂ emissions were estimated with an accuracy <65% for a source strength >10 Mt CO₂ yr^-1, while NO_x emissions >10 kt NO₂ yr^-1 were estimated with <56% accuracy. NO₂ observations were essential for detecting the plume and constraining the shape of the Gaussian curve. With three CO2M satellites, annual CO₂ emissions were estimated with an uncertainty <30% for source strengths larger than 10 Mt yr^-1, which includes an estimate of the uncertainty in the temporal variability of emissions. Annual NO_x emissions were estimated with an uncertainty <21%. Since NO_x emissions can be determined with better accuracy, estimating CO₂ emissions directly from the NO_x emissions by applying a representative CO₂:NO_x emission ratio  seems appealing but this approach was found to suffer significantly from the high uncertainty in the  CO₂:NO_x emission ratios determined from the same CO2M observations.

Our study shows that CO2M should be able to quantify the emissions of the 400 largest point sources globally with emissions larger than 10 Mt yr^-1 that account for about 20 % of global anthropogenic CO₂ emissions. However, the mass-balance approach used here has relatively high uncertainties that are dominated by the uncertainties in the estimated CO₂ background and the wind speed in the plume, and uncertainties associated with the sparse temporal sampling of the varying emissions. Estimates could be significantly improved if these parameters can be better constrained, e.g., with atmospheric transport simulations and independent observations.