Evaluation of Cloud Representation of the high-resolution mesoscale model HARMONIE using Meteosat Second Generation observations

Abstract

This study consists of two parts. In the first part we evaluate the cloud cover representation of the high-resolution mesoscale model HARMONIE using MSG (Meteosat Second Generation) observations. We assess the performance of different model versions. We also examine anomalies in time and space. Cloud cover is a difficult to model parameter and is dependent on several processes, like turbulence and convection. Verification of cloud cover in HARMONIE using satellite data has not been done systematically and is therefore needed. Although MSG cloud cover is approximately 4% higher compared with ground based observations, a significant underestimation of cloud cover in HARMONIE above sea is observed. The degree of underestimation is dependent on weather conditions. The latest version of HARMONIE correlates better with MSG than its predecessors. More advanced verification methods show a systematic underestimation of the cloud area and an increase in performance during the night.

In the second part, the impact of cloud cover schemes is studied. Subgrid variability of temperature and humidity are important for cloud formation. Parametrizations are needed to link the subgrid variability of temperature and humidity to cloud cover and are called cloud schemes. Two types of cloud schemes are assessed: a relative humidity cloud scheme and a statistical cloud scheme. The relative humidity cloud scheme directly links relative humidity to cloud cover, while the statistical cloud scheme provides a probability distribution of the temperature and humidity around the mean, which can be converted to a fractional cloud cover. Each sophistication of the statistical cloud scheme contributes to better cloud cover forecasts. During the day, sophistications to the cloud scheme have a larger impact than during the night, which is due to the lack of convective and turbulent conditions during the night. Relative humidity cloud schemes perform worse than the statistical cloud schemes. The relative humidity cloud scheme does provide information about possible changes to the statistical cloud scheme.