Atmospheric carbon dioxide (CO₂) concentrations have risen significantly since the pre-industrial era due to human activities, contributing to climate change, rising sea levels, and an increased frequency of extreme weather events. Substantial efforts have been made to estimate anthropogenic emissions and monitor industrial facilities to reduce CO₂ release into the atmosphere. However, accurately quantifying emission rates at small scales presents significant challenges, requiring high-resolution observational data and transport models.

In this project, we utilized drone-measured CO₂ concentration data from a controlled point-source release campaign, with a spatial resolution of approximately 1 cm and a sampling frequency of 20 Hz. This dataset was combined with a WRF-LES (Large Eddy Simulation) model at a 10-meter resolution, employed as an adjoint in a Bayesian inversion algorithm to estimate the point-source emission rate. The project is ongoing; with the WRF-LES adjoint models completed and initial inversion results obtained, we are now focusing on refining data processing methods to further enhance the accuracy and reliability of our findings.