Investigating the sensitivity of high-resolution mesoscale models to microphysical parameters by the use of polarimetric radar observations

An improved methodology for investigating mesoscale model microphysics is presented and discussed for a case study. Polarimetric radar data are used to assess numerical weather prediction (NWP) model's skill in reproducing the microphysical features of severe rainfall. To this aim, an event of deep convection, developed on 20 May 2003 in the Po Valley (Italy), is analyzed. During the selected case study, two weather radars, sited in Gattatico and San Pietro Capofiume (near Bologna, Italy), detected a deep-convective and hail cell with a large inner graupel core which reached the ground, as was reported by local weather authorities and citizens. A hydrometeor classification algorithm, based on a Bayesian approach and a radar simulator model, are used to retrieve the vertical structure of the storm and characterize its ground effects. These products are used for evaluating the sensitivity of NWP models with respect to the graupel density, described in terms of the intercept parameter of the graupel size distribution and its depositional velocity. To this purpose two mesoscale NWP models, specifically COSMO-LAMI and MM5-V3, are used at high spatial resolution. Their ability in reproducing the vertical and the horizontal structure and the microphysical distribution of the major convective cell is evaluated. Both models show large sensitivity to different microphysical settings and a capability to reproduce fairly well the observed hail cell. Ground-radar reflectivity fields and the hydrometeor vertical structure are correctly simulated by both NWP models as opposed to a failure in reproducing the graupel distribution near the ground.