Atmospheric Modeling and Verification of Point Source Fossil Fuel CO2 Emissions

Abstract:

Emissions from large point sources (electricity generation and large-scale industry) of fossil fuel CO₂ (CO₂ff) emissions are currently determined from self-reported “bottom-up” inventory data, with an uncertainty of about 20% for individual power plants. As the world moves towards a regulatory environment, there is a need for independent, objective measurements of these emissions both to improve the accuracy of and to verify the reported amounts.

“Top-down” atmospheric methods have the potential to independently constrain point source emissions, combining observations with atmospheric transport modeling to derive emission estimates. We use the Kapuni Gas Treatment Plant to examine methodologies and model sensitivities for atmospheric monitoring of point source fossil fuel CO₂ (CO₂ff) emissions. The Kapuni plant, located in rural New Zealand, removes and vents CO₂ from locally extracted natural gas at a rate of ~0.1 Tg carbon per year. We measured the CO₂ff content in three different types of observations: air samples collected in flasks over a period of a few minutes, sodium hydroxide solution exposed the atmosphere, and grass samples from the surrounding farmland, the latter two representing ~1 week integrated averages. We use the WindTrax Lagrangian plume dispersion model to compare these atmospheric observations with “expected” values given the emissions reported by the Kapuni plant. The model has difficulty accurately capturing the short-term variability in the flask samples but does well in representing the longer-term averages from grass samples, suggesting that passive integrated-sampling methods have the potential to monitor long-term emissions. Our results indicate that using this method, point source emissions can be verified to within about 30%.

Further improvements in atmospheric transport modelling are needed to reduce uncertainties. In view of this, we discuss model strengths and weaknesses and explore model sensitivity to meteorological conditions. In particular, we look at how much of the variation in our observations can be attributed to changing emissions levels and how much is a result of natural climate and atmospheric transport variations that remain unconstrained by our model and input data.