RANS and LES face to face for forecasting extreme precipitation events in the Liguria region (northwestern Italy)

In the present work, seven different extreme precipitation events, that affected Liguria region (Italy) in the past ten years, were analysed in depth by means of sub-km resolution numerical simulations with the state-of-the-art non-hydrostatic Weather Research and Forecasting (WRF) mesoscale model. The study aims at evaluating the impact on quantitative precipitation forecast of an explicit description of atmospheric boundary layer turbulence through a Large-Eddy Simulation (LES) approach versus the classical Reynolds-Averaged Navier–Stokes (RANS) modelling framework. To this purpose, three different sets of simulations were carried out. In the first set, reaching a resolution of 1.1 km through three nested domains, the best performing planetary boundary layer (PBL) parameterization scheme identified in a preliminary sensitivity analysis, was imposed on each domain. This modelling setup was chosen as the reference one. Then, a further nested domain was introduced with horizontal resolution of 367 m, on which both RANS and LES simulations were performed. Quantitative precipitation forecasts were compared with observed data coming from the regional rain gauge network, composed of about 200 professional stations. In four out of the seven considered events, results show an evident LES contribution to improve model performances, regarding both intensity and/or location of precipitation. This improvement is observed both in cases of underestimation and overestimation by the reference simulations and is mostly associated to a better description of low-level dynamics as well as convection triggering and intensity. In other cases, the effect of LES is not remarkable and simulations closely resemble the reference ones. Further investigation is necessary to strengthen these conclusions, but the results of this study can suggest the possibility to integrate LES in operational simulations.