Lessons learned after two years of operational high-resolution (1,5 km) WRF simulations in Catalonia and a plan to increase their skill

The WRF-ARW has been the flagship mesoscale model in the Meteorological Service of Catalonia (SMC) since 2012. Several operational runs are performed daily (initialised at 00 and 12 UTC), using both the GFS and the IFS model for initial and boundary conditions, to account for uncertainties in the synoptic evolution. To provide more accurate forecasts to end-users, a convection-allowing simulation with a grid spacing of 1,5 km was added to the operational chain, starting in the summer of 2019. 

However, the verification results show that the improvement over its mother domain (a 3 km simulation with parameterised convection) is irregular because it does not happen for all the variables. For instance, the 2 m temperature forecasts are more reliable for the highest resolution domain but the wind speed at 10 m has a comparable skill. Regarding the precipitation, there is a very slight improvement only for high daily precipitation rates (50 or 80 mm) during some seasons; nevertheless, the results are worse in forecasting the occurrence of precipitation (that is, when considering low daily precipitation quantities). The comparison of the verification results among different model configurations (with various resolutions and initial conditions) can be easily performed by using a skill score table. This table and its design will also be presented in this session. 

Certainly, these results help to conceive strategies to enhance the skill of the 1,5 km simulations for some of the variables that arise as more inaccurate. For instance, it is evaluated to what extent using alternative static fields (changing the model topography or the land category) improves the forecasts of temperature, humidity or wind near the surface. Furthermore, the sensitivity of precipitation forecasts to several physics schemes is tested, seeking an enhancement of their skill.