A method for estimating localized CO₂ emissions from co-located satellite XCO₂ and NO₂ images

Abstract

Carbon dioxide (CO2) is the most important anthropogenic greenhouse gas. Its atmospheric concentration has increased by almost 50 % since the beginning of the industrial era, causing climate change. Fossil fuel combustion is responsible for most of the atmospheric CO2 increase, which originates to a large extent from localized sources such as power stations. Independent estimates of the emissions from these sources are key to tracking the effectiveness of implemented climate policies to mitigate climate change. We developed an automatic procedure to quantify CO2 emissions from localized sources based on a cross-sectional mass-balance approach and applied it to infer CO2 emissions from the Bełchatów Power Station (Poland) using atmospheric observations from the Orbiting Carbon Observatory 3 (OCO-3) in its snapshot area map (SAM) mode. As a result of the challenge of identifying CO2 emission plumes from satellite data with adequate accuracy, we located and constrained the shape of emission plumes using TROPOspheric Monitoring Instrument (TROPOMI) NO2 column densities. We automatically analysed all available OCO-3 overpasses over the Bełchatów Power Station from July 2019 to November 2022 and found a total of nine that were suitable for the estimation of CO2 emissions using our method. The mean uncertainty in the obtained estimates was 5.8 Mt CO2 yr−1 (22.0 %), mainly driven by the dispersion of the cross-sectional fluxes downwind of the source, e.g. due to turbulence. This dispersion uncertainty was characterized using a semivariogram, made possible by the OCO-3 imaging capability over a target region in SAM mode, which provides observations containing plume information up to several tens of kilometres downwind of the source. A bottom-up emission estimate was computed based on the hourly power-plant-generated power and emission factors to validate the satellite-based estimates. We found that the two independent estimates agree within their 1σ uncertainty in eight out of nine analysed overpasses and have a high Pearson's correlation coefficient of 0.92. Our results confirm the potential to monitor large localized CO2 emission sources from space-based observations and the usefulness of NO2 estimates for plume detection. They also illustrate the potential to improve CO2 monitoring capabilities with the planned Copernicus Anthropogenic CO2 Monitoring (CO2M) satellite constellation, which will provide simultaneously retrieved XCO2 and NO2 maps.