Improvements and challenges of modeling air pollutants by assimilating Sentinel-5p TROPOMI observations

Within the framework of the CAMS Evolution (CAMEO) project, we implement a three-dimensional variational (3D-Var) data assimilation system in the Danish Eulerian Hemispheric Model (DEHM) to improve simulations of key atmospheric pollutants in Europe including sulfur dioxide (SO₂), ozone (O₃), carbon monoxide (CO), and formaldehyde (HCHO). The data assimilation framework integrates Sentinel-5p (S5p) TROPOMI satellite observations with model predictions to provide more accurate estimates of these species. The 2023 Mount Etna eruptions, captured in S5p observations, provide an opportunity to evaluate the performance of the modeling system under extreme emission scenarios. Of particular interest is the ability of the system to capture not only the substantial SO₂ plumes from volcanic eruptions, but also their cascading effects on other pollutants – including the formation of CO through magmatic processes, and perturbations in O₃ concentrations due to complex gas and heterogeneous chemical processes. Our approach confronts several key challenges, including the representation of highly localized and dynamic pollutant distributions, interactions of different chemical species, and the refinement of error covariance structures for both regular and extreme episodes. Evaluations with both satellite and ground observations show enhancements of SO₂ concentrations especially at upper layers (e.g. 2-4 km) but also show challenges to improve ground-level concentrations compared to observations. Sensitivity analyses are conducted to assess the impact of assimilation frequency, observation error specifications, and the inclusion of supplementary ground-based data. Results demonstrate improvements in the capability of DEHM to simulate atmospheric transport and chemical processes across various temporal and spatial scales, from regional background conditions to intense emission events. The study highlights the potential of near-real-time satellite data assimilation in enhancing vertical distribution and provides insights into optimizing model performance during dynamic emission events. The findings also provide insights into optimizing model performance for varying spatial and temporal scales of atmospheric phenomena.