Evaluating High-Resolution Model Forecasts of European Winter Storms by Use of Satellite and Radar Observations

Abstract

The forecasting performance of the operational Lokal Modell (LM) of the German Weather Service (DWD) is evaluated for two severe winter storms that crossed central Europe in December 1999. Synthetic satellite images constructed from model output fields are compared with observed imagery from a Meteosat satellite. Also, synthetic radar images constructed from forecast precipitation fields are taken for validation against observed precipitation as represented by the European radar composite of the DWD. Comparisons are performed by visual inspection of satellite and radar imagery and by calculating statistical measures such as frequency histograms of observed and synthetic brightness temperature. Whereas the visual inspection allows detection of even finescale details in both synthetic satellite and radar imagery, for example, the position of fronts and rainbands, the statistical analysis reveals model deficiencies with respect to the representation of upper-level cloudiness. The operational LM did not incorporate a cloud-ice parameterization scheme at the time that these storms occurred. Additional experiments were performed to investigate the impact of the cloud-ice parameterization on brightness temperature. As expected, the longwave radiative impact is seen to be strongly influenced by the presence of cloud ice. Furthermore, changing the value of the ice-to-snow autoconversion threshold in the microphysics scheme by a factor of 2 leads to a significant improvement in synthetic brightness temperature as compared with observations. The results suggest that synthetic satellite and radar images could be used to perform quality control on numerical weather forecasts in real time. Basic ideas are proposed for an automated quality-control system using an image-matching tool developed at the German Aerospace Center (DLR).