The operational numerical forecasting system used for predicting road conditions is presented. The core of the system is the Danish road condition model (RCM). The atmospheric forcing for the RCM derives from a limited area model (HIRLAM). For the initialisation of the forecasts observations from the road stations and cloud cover from synoptic stations for the previous three hours are taken into account applying flux corrections for the individual stations. This assures that the initial states for the forecasts are in agreement with the observed state in a consistent way.
Five hour forecasts are made every hour for the nearly 250 road stations in Denmark. The data assimilation and the treatment of the physical processes in the RCM are described. The quality of the model output is illustrated through verification of operational results. Finally, problems associated with accurate forecasts of road condition during winter is discussed.