A revised cloud microphysical parameterization for operational numerical weather prediction using the COSMO model Axel Seifert Deutscher Wetterdienst, Offenbach, Germany Susanne Crewell University of Cologne, Cologne, Germany Quantitative precipitation forecasting (QPF) is one of the major applications of limited-area numerical weather prediction (NWP) models. With a high-resolution NWP model, like the 7-km COSMO-EU which is operational at DWD, the detailed orography and the explicit simulation of mesoscale dynamical structures should lead to an increased forecasting skill compared to global models with coarser horizontal resolution. Unfortunately, the last years have shown some prevalent problems with the precipitation forecasts of the COSMO model. For example, an overestimation of orographic precipitation and, as in many operational forecast models, a too frequent occurence of very light precipitation (drizzle). Together with a model evaluation against cloud radar measurements which revealed that the model often predicted too low values of liquid and ice water content, these deficiencies point towards problems in the microphysical parameterization. Therefore a revised version of the COSMO-EU microphysics scheme has recently been developed. The grid-scale microphysics parameterization of COSMO-EU is a standard one-moment bulk scheme that explicitly predicts cloud droplets, rain drops, cloud ice and snow with most details taken from Rutledge and Hobbs (1983). To improve the mesoscale precipitation structures predicted by COSMO-EU several modifications have been made recently: First, the Kessler-type autoconversion/accretion scheme has been replaced by the parameterization by Seifert and Beheng (2001), simplified to a one-moment scheme, i.e. assuming a constant cloud droplet number concentration. Second, the constant snow intercept parameter has been replaced by a new parameterization as a function of temperature and snow mixing ratio using observations of Field et al. (2005). Third, some modifications of the ice/snow microphysics like a temperature dependent sticking efficiencies and a reduced fall speed have been introduced. The results show a significant improvement of the above-mentioned QPF problems. Drizzle events during spring and autumn are now forecasted much better with less false alarms. Orographic precipitation structures, e.g. at the Black Forest mountain ridge, are represented much more realistically. And also a validation of the ice water content (IWC) with retrievals from cloud radar measurements shows also a clear improvement. While the old scheme produced too low values of IWC the revised scheme agrees well with the radar-derived estimates of IWC in frontal clouds.