How microphysical choices affect simulated infrared brightness temperatures

- New two-moment cloud ice scheme reduces simulated infrared brightness temperature bias distinctly.
- Choice of heterogeneous ice nucleation scheme and consideration of cloud ice sedimentation together have largest impact.
- Other changes including exact aerosol number concentrations play minor role.

Numerical weather prediction (NWP) today relies more and more on satellite data, both for assimilation and for evaluation. However, process-based analyses of the biases between observed and simulated satellite data, which go beyond a mere identification of the biases, are rare. The present study investigates a long-known bias (Böhme et al., 2011) between brightness temperatures (BTs) simulated from the regional NWP model COSMO-DE forecasts via RTTOV (Radiative Transfer for TOVS) and those observed by Meteosat Second Generation (MSG) Spinning Enhanced Visible and Infrared Imager (SEVIRI). The pivotal question is whether a novel two-moment cloud ice scheme, developed by Köhler (2013) primarily to improve the representation of ice nucleation processes, exhibits an improved performance with respect to this bias and, if that is so, to provide a process-based analysis which identifies the reasons for the improved behaviour.It is shown that the new two-moment cloud ice scheme reduces the BT bias distinctly and can therefore be considered an improvement in comparison to two standard schemes, the two-category ice scheme and the three-category ice scheme. The improvement in simulated BTs is due to a vertical redistribution of cloud ice to lower model levels. Sensitivity studies identify two of the introduced changes in the two-moment cloud ice scheme to be hand-in-hand responsible for most of the improved performance: the choice of heterogeneous ice nucleation scheme and the consideration of cloud ice sedimentation. Including only cloud ice sedimentation without changing the heterogeneous ice nucleation scheme has no distinct effect on cloud ice. Further sensitivity studies with varying aerosol number densities reveal a comparably small sensitivity, indicating that the use of a physically reasonable heterogeneous ice nucleation scheme is far more important than the exact knowledge of the actual aerosol number densities.