Cloud statistics and NWP-model validation based on long term measurements of a 35 GHz radar

Since November 2003 the Lindenberg observatory is operating continuously a 35.5 GHz coherent and polarimetric cloud radar to measure vertical profiles of reflectivity, Doppler velocity, spectral width and the Linear Depolarisation Ratio (LDR) between 250 m and 15 km height. The radar is designed for long term measurements and has a vertically pointing antenna. The mean annual data availability varies between 83% and 99% during the last seven years. These data have been used for a representative cloud statistic and for validating the numerical-weather-prediction models (GME, COSMO-EU, COSMO-DE) of the German Meteorological Service (DWD). Before the radar data have been analysed with respect to cloud cover and cloud fraction non-hydrometeors were removed by a combination with co-located ceilometer measurements.

The cloud cover shows a significant annual cycle with a maximum in winter and a minimum in summer time, which is mainly caused by the cycle of low clouds. High and middle level clouds are much less variable over the year, whereas high level clouds have a minimum in April and September. Remarkable is an excellent agreement to the expert observed cloud cover, both in total and in the low atmospheric level. The models tend to underestimate clouds in lowest levels and to overestimate middle and especially high level clouds. It is shown that cloud fractions greater than 95% are not predicted by the models with the same frequency as observed by the radar above about 5 km, which can be explained by errors in the empirical parameterisation of the cloud fraction of ice clouds. A lack of low level clouds in the models can be observed especially in spring and summer from the morning until the afternoon. The cloud statistics will be completed by relationships between cloud fraction and humidity at heights of the cloud base.