Air quality modeling of sulfur dioxide (SO2) concentrations remains challenging due to the high variability of both natural and anthropogenic emission sources, as well as the complexities associated with its multiphase chemistry. The data assimilation (DA) of satellite observations is a promising technique for constraining model uncertainties by combining the strengths of high-resolution and dense satellite retrievals with physical consistent model outputs. However, existing SO2 DA applications have primarily focused on volcanic events while this study addresses them together with other emissions. The implementation of an SO2 DA framework within the MINNI regional chemical transport model using an Ensemble Adjusted Kalman Filter (EAKF) via the DART framework is presented. The performances of the DA assimilation framework were tested using Sentinel-5P/TROPOMI SO2-COBRA total column retrievals over continental Europe for August 2023. The filter constrained the ensemble variance to capture plumes from power plants and volcanic activity. The ensemble considered 20 members and perturbations of emissions and boundary conditions. On a monthly basis, the mean correction for the total column averaged over the domain was 2 × 10−5 mol m−2, with localized maximum adjustments reaching 3.3 × 10−4 mol m−2. At the surface level, domain-averaged corrections of concentrations reached up to 2.6 µg m−3. Despite current limitations related to ensemble size, static vertical localization, and the typical temporal fading of initial condition corrections, validation against in situ data confirmed the system’s ability to transfer column information to near-surface levels. These results demonstrate the feasibility and added value of integrating mixed-source SO2 satellite retrievals into regional air quality simulations, contributing to more accurate, observation-driven atmospheric monitoring.