Sulphur dioxide emissions during volcanic eruptions can have a significant impact on air traffic. Indeed, sulphur dioxide is a precursor of sulphuric acid which is highly corrosive and damages aircraft engines when they pass through a plume of SO2. Currently, to predict the concentration of sulphur dioxide of volcanic origin in the French chemistry-transport model MOCAGE, two emission inventories from Carn are used, one for active and one for passive outgassing. However, these inventories are based on annual satellite measurements and do not cover all the volcanoes in the world. Moreover, active degassing is variable over the course of a year. A first way to improve SO2 forecasting is to use accurate information on released SO2. However, this information is not available in near real time. A second way is to use satellite information and assimilate these data in MOCAGE. TROPOMI on Sentinel-5p is able to measure SO2 columns, even for a weak signal, with high horizontal resolution from the surface to the top of the atmosphere, but it cannot give information on the total SO2 column during the night. The IASI instruments on Metop B and C measure total SO2 columns with a lower resolution than TROPOMI and do not give any information on the presence of volcanic SO2 below 5km altitude, but they can measure the total SO2 column during the day and night. The total SO2 columns from these satellites were assimilated into MOCAGE using different settings to constrain the height of the SO2 plume in the model. This presentation will give a comparison between the assimilation using some settings for the eruption of three volcanoes: the Hunga Tonga-Hunga Ha’apai , the Soufriere Saint Vincent and the Cumbre Vieja.