Uncertainties in the simulation of XCO2 plumes from power plant emissions: A comparison between 6 high-resolution atmospheric transport models

Abstract

Power plants are a major source of CO2 globally. Although their emissions are routinely monitored in many countries especially in the developed world, these numbers are often not publicly available and a complete global record is still far from reality. An important goal of Europe's planned Copernicus CO2 satellite mission CO2M is therefore to provide an independent quantification of power plant emissions worldwide. Emissions may be estimated from satellite XCO2 observations by simulating the plumes with an atmospheric transport model and finding those emissions that minimize a cost function of the differences between simulation and observations. Here we present a comparison of CO2 plume simulations from six high-resolution models, three Large Eddy Simulation models, two mesoscale Eulerian models, and one Lagrangian particle dispersion model. Simulations were conducted for two large coal-fired power plants, Belchatow in Poland and Jänschwalde in Germany, which were extensively observed with aircraft in situ and remote sensing measurements during the CoMet campaign in May-June 2018. The observations provide a unique opportunity to study the capability of the models to simulate such plumes in a realistic manner and to design optimal modelling and emission quantification strategies. The Belchatow plume was sampled under highly convective and turbulent conditions whereas the Jänschwalde plume was observed in a more stable weather situation. The models are able to reproduce these differences by simulating a highly structured turbulent plume for Belchatow and a more Gaussian-shaped plume for Jänschwalde. However, the models differ in many details including the horizontal and vertical spread of the plumes, suggesting that in addition to resolution the specific choices of turbulence and advection scheme have a significant impact on the results. Our findings suggest that estimating emissions from individual images is particularly challenging for turbulent plumes. Since turbulence intensity evolves with the build-up of the convective boundary layer, a satellite overpass well before noon would likely be an advantage.