Evaluation of precipitation forecasts by combined information content of polarimetric cloud and weather radar as well as disdrometer data

Over the last years, weather services have developed a new generation of high resolution mesoscale numerical weather prediction (NWP) models with the aim to improve quantitative precipitation forecasts, especially for convective situations. The skill of forecasts with such models has to be assessed by measurements. This is often done using surface rain gauge data although modern remote sensing and in-situ instruments are able to provide data showing the evolution of convective precipitation in high spatial and temporal resolution. Differences revealed by comparisons of such data with results from NWP models are then useful to identify model deficiencies and improve the model physics. In this study, precipitation forecasts by the NWP model COSMO-LMK which is in operational use at German Weather Service are discussed in the light of plentiful data measured simultaneously by several instruments: the Karlsruhe cloud radar (36 GHz) providing information on cloud properties before the onset of precipitation, the DLR polarimetric Doppler radar POLDIRAD (5 GHz) showing the development of different types of precipitation hydrometeors, 12 vertical pointing micro rain radars (MRR, 24 GHz) delivering profiles of drop size distributions up to heights of about 3 km agl and many rain gauges distributed in the area of interest. The event analyzed took place during the AQUARADAR experiment in summer 2006 in southern Bavaria. The evaluation of model performance is done in terms of various observed quantities. In doing so, synthetic radar quantities are derived from the model forecast employing the polarimetric radar forward operator SynPolRad. SynPolRad was originally developed for polarimetric C-Band radar and has been adapted to also simulate signatures from cloud radar and MRR's. The study focuses on differences between the observations and model forecasts in timing and intensity of clouds and precipitation as well as on the microphysical properties of the precipitation as revealed by polarimetric radar and the disdrometer data.