Enhancing severe convection forecasts with microwave radiance assimilation in limited‐area models

Accurate representation of atmospheric humidity in convection-scale data assimilation is essential for forecasting severe precipitation. Satellite radiances, especially from microwave water-vapour channels, provide valuable information on tropospheric moisture globally, yet remain largely underused in convection-permitting limited-area models (LAMs). This study demonstrates the benefit of assimilating such radiances in LAMs for intense convection forecasting. We use the ICOsahedral Non-hydrostatic model over Italy at 2.2-km resolution (ICON-2I), coupled to the local ensemble transform Kalman filter, following the operational configuration of Arpae Emilia-Romagna and ItaliaMeteo. The severe convective event of September 15, 2022 in the Marche region, Central Italy, serves as a case study; this produced over 400 mm rainfall, substantially underestimated by the operational forecast assimilating only conventional and radar data. We introduce the assimilation of clear-sky microwave water-vapour channels from the polar-orbiting Microwave Humidity Sounder (MHS) as main focus, complemented by all-sky infrared radiances from the geostationary Spinning Enhanced Visible and InfraRed Imager (SEVIRI) following the German Weather Service operational configuration. Each instrument alone improves the Marche event precipitation forecast, but their joint assimilation yields the best performance, indicating positive synergy between microwave and infrared water-vapour channels. Inspection of forecast initial conditions shows that joint MHS+SEVIRI assimilation introduces consistent modifications to the pre-convective atmospheric state, with increased low- to mid-tropospheric humidity favouring deep convection. The relative contribution of each platform is assessed with the partial analysis increments algorithm, highlighting the complementary impact of the two satellites throughout the troposphere. Extended verification over five days confirms overall improvement in root-mean-square error (RMSE) and bias for temperature, humidity, and wind at surface and upper levels, particularly in the mid-troposphere. The fractions skill score and dichotomous scores also show benefit for intense rainfall forecasts from MHS assimilation. These results motivate the operational introduction of satellite radiances in the Arpae Emilia-Romagna and ItaliaMeteo systems.