Carbon dioxide (CO₂) emitted from thermal power plants accounts for approximately 40% of energy-related CO₂ emissions in the world. CO₂ emission datasets for thermal power plants have large uncertainties, especially those from developing countries. We performed CO₂ column measurements with solar-viewing portable Fourier transform spectrometers and wind measurements with a Doppler lidar at sites approximately 3–7 km from a thermal power plant in Japan during field campaigns in October 2018 and October/November 2019 to evaluate the uncertainty in CO₂ emission estimates based on ground-based atmospheric measurements. Under the assumption that the CO₂ plume from a thermal power plant disperses according to a Gaussian plume model, CO₂ emission rates were estimated simultaneously with the dispersion coefficient of lateral plume width and wind direction by an optimal estimation method. The estimated CO₂ emission rates were in good agreement with those determined from hourly fuel consumption data, with a mean bias of 2.8% and a standard deviation of 16.8%. Analyses of 27 plumes observed for approximately one month during the 2019 campaign when the actual CO₂ emission rate was nearly constant revealed that monthly CO₂ emissions can be quantified with a precision (standard error) of 6.9% at the 95% confidence level. Evaluations using different wind data demonstrated that accurate data for determining the wind field are of importance in the column measurement-based emission estimates.