The RMI Road Weather Forecasting System

Slippery road and highway conditions can cause accidents if necessary maintenance actions are not undertaken. These include the clearing of snow and prevention of ice by salting. Belgium is characterised by a “marginal” winter environment, where the air temperature commonly fluctuates around the freezing point. A few snow events usually occur during winter, while days with frost occur more often. For these types of conditions, great potential in cost savings for winter maintenance lies in the accurate prediction of ice formation. To aid decision making, road weather information systems have long been used. These consist of road weather stations (RWS) that gather meteorological and road observations, combined with models that forecast the road condition and tools to communicate the relevant information to end users for decision support. For forecasting purposes, the use of dedicated road weather models (RWM’s) has become popular approach. At the Royal Meteorological Institute of Belgium (RMI), a road weather forecasting system was developed in collaboration with the Royal Netherlands Meteorological Institute (KNMI). The KNMI RWM is a physical model, which was previously validated and compared with the Finnish RoadSurf model over the Netherlands. This model was taken as the basis to develop a RWM for Belgian highways in 2018, leading to the deployment of the RMI “GMS” system (“Gladheidmeetsysteem” in Dutch). This system has been operational since the winter of 2018-2019, and is being further improved in close collaboration with the regional road maintenance authorities Belgium. GMS forecasts are updated every hour and communicated to users through a GIS-based interface. Users can consult forecasts of road surface temperature (RST), road surface condition and various other meteorological variables for all RWS locations (about 140), in addition to RWS observations. Other map layers include overlays of weather radar images for precipitation and satellite images for cloud cover. Finally, users can consult a static thermal map and webcam images for highways in Flanders, and geolocated weather reports generated by citizens through the RMI weather app for the whole country. The physical RWM has been tuned for Belgian highways, and adapted to make use of various available numerical weather prediction (NWP) models as input. Road surface condition information from RWS is also used for better initialization. Since the winter of 2022-2023, the model is also run with a mini-ensemble input of four NWP models. This information is currently used to present an uncertainty interval around the RST forecasts, but more applications are foreseen. We present the operational system, GIS-interface and preliminary validation results of the ensemble road forecasts for the past winter. We also comment on specific use cases and present avenues for future improvements to the GMS system.