A Sensitivity Study of a Bayesian Inversion Model Used to Estimate Emissions of Synthetic Greenhouse Gases at the European Scale

To address and mitigate the environmental impacts of synthetic greenhouse gases it's crucial to quantify their emissions to the atmosphere on different spatial scales. Atmospheric Inverse modelling is becoming a widely used method to provide observation-based estimates of greenhouse gas emissions with the potential to provide an independent verification tool for national emission inventories. A sensitivity study of the FLEXINVERT+ model for the optimisation of the spatial and temporal emissions of long-lived greenhouse gases at the regional-to-country scale is presented. A test compound HFC-134a, the most widely used refrigerant in mobile air conditioning systems, has been used to evaluate its European emissions in 2011 to be compared with a previous study. Sensitivity tests on driving factors like-observation selection criteria, prior data, background mixing ratios, and station selection-assessed the model's performance in replicating measurements, reducing uncertainties, and estimating country-specific emissions. Across all experiments, good prior (0.5-0.8) and improved posterior (0.6-0.9) correlations were achieved, emphasizing the reduced sensitivity of the inversion setup to different a priori information and the determining role of observations in constraining the emissions.The posterior results were found to be very sensitive to background mixing ratios, with even slight increases in the baseline leading to significant decrease of emissions.