Verification of mesoscale model products

Verification of operational forecast models is done primarily through comparison of model analysed fields or through a comparison of selected surface stations with nearby grid points. The new generation of operational mesoscale forecast models with a horizontal resolution of around 10 km does not contain its own data assimilation system in general and consequently does not produce its own analysis field. Further, the horizontal resolution is still too rough to compare instantaneous values of surface stations especially over complex terrain with model grid points with success.

The recently developed model independent mesoscale analysis tool VERA (Vienna Enhanced Resolution Analysis) is used to allow for a verification of the forecasts. Hourly forecast surface fields of the three forecast models ALADIN, LM and ECMWF are compared to the also hourly available model independent analysed VERA fields.

The resulting fields (difference between VERA Analysis and model forecast) are useful for detecting deficits in the actual model run at a very early stage. This makes them to a very useful tool for nowcasting purposes. Currently, this method is applied at the Austrian aviation Met service.

A systematic investigation of the differences may give some hints about general model problems, e.g., speed of frontal movement along the northern Alpine barrier, pressure gradients across the Alps during Foehn situations or systematic differences of the wind direction between model and analysis.

First results of this systematic investigations for the parameters: surface winds, potential and equivalent potential temperature and reduced pressure will be presented in the talk.