Point source emissions are a significant feature of industrial output, with thermal power plants serving as prominent examples. Globally, CO₂ emissions from thermal power plants make up a substantial portion of energy-related emissions. Effective monitoring of these emissions aids in refining top-down carbon estimates worldwide. The portable EM27/SUN spectrometer, known for its mobility and ease of use, provides highly reliable and sensitive measurements and is adaptable across various environments. It has been widely adopted for satellite data validation and greenhouse gas quantification at multiple scales. However, thermal power plants, typically situated in or near urban areas, experience CO₂ plume dispersion influenced by local surface features, such as buildings, which can affect EM27/SUN observation accuracy, a factor seldom addressed in previous studies. To overcome this, we focus on Hefei Wanneng Power Plant. Through multiple ground-based observations using the EM27/SUN instrument, we analyze the effect of nearby buildings on plume dispersion and assess how the spatial relationship between the measurement line and plume geometry influences site selection. This approach enables a better understanding of CO₂ quantification capabilities and associated uncertainties under varying conditions, providing a foundation for optimizing measurement methods and enhancing accuracy.

To improve CO₂ emission estimation accuracy from industrial point sources using the ground-based EM27/SUN, we explore optimized measurement methods. Using the Lagrangian particle dispersion model, we analyze the influence of nearby buildings on plume dispersion and assess the role of geometric positioning between the measurement line and plume in site selection. By synchronizing EM27/SUN data with TCCON observations and implementing a quality control criterion based on solar intensity, we enhance data reliability. The distance from the emission source is divided into three segments to evaluate the influence of building on measurement and examine plume positioning’s influence on CO₂ column volume fraction data. Our final analysis provides insights into the limitations of current measurement methods and demonstrates how the optimized measurement method can improve emission estimation accuracy.

We begin with a data preprocessing method, using strict quality controls to minimize anomalies and ensure measurement accuracy. Corrected EM27/SUN data from 2021, benchmarked against TCCON, show significant accuracy improvement, with an R² of 0.978 and RMSE of 0.271, compared to pre-correction values of 0.85 and 0.412. Using Hefei Wanneng Power Plant as a case study, repeated measurements from 2021?2023 indicate that downwind $X_{\mathrm{C}{\mathrm{O}}_{\mathrm{2}}}$ increase by 12×10^-6?16×10^-6 relative to background levels, highlighting the influences of CO₂ emissions. Our findings reveal that building-induced wind field complexity near the plant introduces spatial variability in CO₂ plumes and increases estimation uncertainty, with an average deviation of (7.98±10) kg/s in the far section. In both the near and middle sections, deviations in the observation path from the plume’s intersection point influence CO₂ column volume fraction, with rapid changes in column volume fraction observed in the near section and more stable changes in the middle section. The study indicates that, based on an annual average CO₂ emission intensity of 195 kg/s for the power plant, the average deviation in emission intensity estimates in the middle section (1.50?2.24 km) is only (7.13±5.39) kg/s. In contrast, the near section (0.30?1.49 km) has a larger deviation of (28.63±17.57) kg/s, and the far section (2.25?3.50 km) has a deviation of (27.01±17.98) kg/s. In addition, emission estimates for 2022, which primarily rely on middle section measurements, show notably smaller fluctuations than those from 2021 and 2023, with a deviation of just (11±9.67) kg/s. Overall, measurements taken in the middle section of the plume contribute to more reliable results, facilitating a more accurate assessment of the power plant’s CO₂ emission.

Our analysis reveals that nearby buildings and the observation path significantly influence EM27/SUN measurement accuracy. Close to the emission source, CO₂ diffusion is limited, making deviations from the plume axis critical. In contrast, far-section measurements face challenges from complex wind patterns shaped by surrounding buildings, resulting in increased variability in $X_{\mathrm{C}{\mathrm{O}}_{\mathrm{2}}}$ observations. The middle section, however, benefits from relatively uniform CO₂ distribution and moderate diffusion, yielding more consistent results. In the emission estimation results, using Hefei Wanneng Power Plant as an example, the average deviation of CO₂ emission intensity in the middle section (1.50?2.24 km) of this experiment is only (7.13±5.39) kg/s. In contrast, the average deviation in the near section (0.30?1.49 km) is (28.63±17.57) kg/s, and in the far section (2.25?3.50 km), it is (27.01±17.98) kg/s. The dispersion in the near and far sections is significantly greater than that at measurement points within the 1.50?2.24 km range, indicating that measurements taken in the middle section of the plume are more likely to yield reliable results. The experimental and model analysis shows that, in ground-based remote sensing of CO₂ emissions from thermal power plants, the near section (close to the emission source) and the far section (with potential high-rise obstructions) present considerable limitations. Therefore, focusing measurements on the middle section of the plume is recommended to achieve high-precision results.