Operational all‐sky assimilation of geostationary infrared water‐vapour channels in the regional ICON‐D2 model with an ensemble Kalman filter

Abstract The water‐vapour‐sensitive channels of infrared imager instruments on board geostationary satellite instruments, such as the spinning enhanced visible and infrared imager (SEVIRI), deliver valuable information on humidity and clouds in the upper and mid‐level troposphere. In past decades it was common practice to assimilate these channels in clear‐sky only, due to a number of challenges of the all‐sky assimilation, such as the forward modelling (i.e., the computation of the model‐equivalent brightness temperatures) of cloudy scenes. In recent years, however, more efforts have been taken by research institutions and weather services to overcome these challenges. In this article we describe which steps were necessary to assimilate the two water‐vapour‐sensitive channels of Meteosat‐SEVIRI at 6.2 and 7.3 m in our convective‐scale regional forecasting system ICON‐D2 at the German Weather Service within a local ensemble transform Kalman filter framework in all‐sky conditions with positive forecast impact. After finding an optimal set‐up how to use these channels in combination with the already operationally assimilated visible channel at 0.6 m including vertical height assignment and localization, observation error modelling and data reduction, a clear positive impact is achieved. Upper air moisture, the cloud cover of high and mid‐level clouds, and thereby also direct and diffuse radiation at the surface and the 2‐m temperatures and humidity verified against SYNOP stations is improved in forecasts for more than 24 hr.