Uncertainties in synthetic Meteosat SEVIRI infrared brightness temperatures in the presence of cirrus clouds and implications for evaluation of cloud microphysics

Synthetic brightness temperatures of five infrared Meteosat SEVIRI channels are investigated for their sensitivities on cirrus radiative properties. The operational SynSat scheme of the regional German weather prediction model COSMO-DE is contrasted to a revised scheme with a special emphasis on consistency between the model-internal ice-microphysics and infrared radiation in convective situations. In particular, the formulation of generalized effective diameters of ice, snow and graupel as well as subgrid-scale cloud cover has been improved. Based on the applied modifications, we first show that changed assumptions on the cirrus radiative properties can lead to 10K warmer brightness temperatures. Second, we demonstrate that prescribed relative changes of 20% in cloud cover and particle size induce maximum changes of around 4 to 5K. The maximum sensitivity appears for semi-transparent cirrus having brightness temperatures around 240 and 260K and total frozen water path around 30gm−2 for viewing geometries over Central Europe. We further consider the known COSMO-DE cold bias to discuss the problem of inconsistencies in model-internal and external formulations of cloud microphysical and radiative properties. We demonstrate that between 35% and 70% of the cold bias can be attributed to the radiative representation of cirrus clouds. We additionally discuss the use of window-channel brightness temperature differences for evaluation of model microphysics and hypothesize that the amount of COSMO-DE ice is overestimated in convective situations. •Revised scheme for calculation of synthetic Meteosat SEVIRI brightness temperatures•Improved consistency between model-internal and external microphysical formulations•Up to 10K systematic changes in brightness temperatures for semi-transparent cirrus•Crystal size perturbations of 20% induce up to 4K brightness temperature deviations.•Cold COSMO-DE bias is partly attributed to cirrus-radiation uncertainties.