In the framework of the ESA Sentinel-5 Precursor Innovations project (S5P+I) a SO2 layer height (LH) retrieval algorithm based on machine learning has been developed, successfully validated and used by ECMWF/CAMS. The SO2LH processor is currently running in a pre-operational quasinear-real-time mode on DLR IMF computers to automatically retrieve the LH from ongoing volcanic eruptions. The generation of SO2LH products and dissemination to ECMWF/CAMS e.g. after a volcanic eruption significantly improves the CAMS SO2 forecasts since the accurate deter - mination of the location, height and loading of SO2 plumes emitted by volcanic eruptions is essential. This project allows the continuation of this automated service until the SO2LH algorithm is implemented into the operational framework.
