Item

Abstract

A profound understanding of clouds and precipitation is crucial to reduce the largest uncertainties of current weather and climate predictions (IPCC, 2013). Relevant processes occur on scales of less than 1 km and cannot be explicitly simulated by today’s climate simulations with horizontal resolutions on the order of 100 km and weather forecasts on the order of 1 km. Large Eddy Simulations (LES) for huge domains with realistic forcing arean emerging tool to bridge this gap. Nevertheless, physical consistency and realism as the prerequisite for model-based studies have to be ensured. For this, an overarching evaluation with new evaluation techniques is developed considering the demands of LES. This conceptis applied to various simulations of the novel ICOsahedral Non-hydrostatic (ICON) LES model with realistic forcing data. The added value is explored through comparisons of the ICON LES with the cloud-resolving COSMO model in terms of basic atmospheric parameters and wind gusts.Twelve days of Germany-wide ICON LES with different resolutions of down to 156 m and 2.8 km resolved COSMO simulations are used for the evaluation of the basic atmospheric state (e.g. wind, temperature, humidity). In situ observations from, for example, weather station networks and remote-sensing measurements are used as reference data. Cloud evaluation is conducted by two months of ICON LES with a resolution of down to 156 m and a circular domain of 220 km in diameter. The model output is compared with compre-hensive cloud measurements and by means of the Cloudnet target classification, providing information about the cloud structure and phase. A novel cloud classification algorithm based on the direct model output is developed and applied. Additionally, physically consistent forward simulations of cloud radar, microwave radiometer, and lidar observations are performed to generate a forward-simulated cloud classification. The added value of LES regarding wind gusts compared to cloud-resolving models is explored by a one-day ICON LES case study around Hamburg with six nests down to 20 m and 20 Hz wind measurements of a boundary layer tower. The basic atmospheric state is well represented by the ICON LES even though the well-tuned COSMO model is slightly better for most parameters and no added value is seen. Incontrast to the expected higher accuracy due to the higher resolution, the errors are often largest for the finest resolved ICON LES. Overall, the simulated clouds by the ICON LESagree well with the observations at supersites. Nevertheless, frozen hydrometeors are overestimated by the ICON LES above 5 km with an ice water content of up to half an order of magnitude larger than the measurements. Additionally, liquid hydrometeors are over-estimated below 5 km, detectable by an overestimated liquid water content of up to one order of magnitude. The cloud classification based on the direct model output is more practicable than the forward simulated approach with remaining technical issues. The diurnal cycle of the wind gust profiles of the 20 m resolved ICON LES show a clear added value by a good match with the observations even though not all wind gusts are explicitly resolved. A new wind gust parameterisation based on the turbulence spectrum for LES is developed and reduces the error of the simulated wind gusts by up to 60% compared to the non-parameterised model output.