Martin Weissmann, Hans-Ertel-Centre for Weather Research, LMU Munich

Using MSG SEVIRI infrared satellite observations for convective-scale ensemble data assimilation

Florian Harnisch, Leonhard Scheck


The limited predictability of convective systems requires the assimilation of frequent and spatially dense observations in convective-scale data assimilation systems. Measurements from geostationary satellites are therefore a potentially powerful data set. So far, mainly variational methods have been used to assimilate satellite radiances, while using these observations in ensemble data assimilation is still in its infancy. Clear-sky infrared channels were used successfully in several modelling systems, but assimilating cloud-affected data poses significant challenges. To facilitate the direct assimilation of MSG SEVIRI observations in the experimental km-scale ensemble data assimilation (KENDA) system of Deutscher Wetterdienst (DWD), a variety of problems need to be addressed. These are, among others, an effective treatment of clear-sky and cloudy areas and the correction of systematic differences between observations and model equivalents. Following studies at ECMWF, we developed an error model for the assimilation of infrared humidity channels in KENDA in an all-sky approach. The model accounts for the increased uncertainty in cloudy conditions, but avoids the need for additional derived cloud products. First assimilation experiments show that the use of the error model in KENDA leads to reasonable increments in both clear-sky and cloudy conditions. Furthermore, the radiance assimilation leads to improved cloud and water vapour fields.

File available here