Sensitivity of meteorological high-resolution numerical simulations of the biggest floods occurred over the Arno river basin, Italy, in the 20th century

During recent years quantitative precipitation forecasts (QPFs), based on NWP models, were continuously increasing their performances and accuracy especially for light and moderate precipitation thresholds; deterministic verification suggests that for high thresholds there wasn't a similar improvement. High thresholds and rare events are particularly difficult to handle and this is a strong limitation for operational activities, particularly for flood forecasting. Where rainfall occurs, when it happens and how much it will be, are information that depends on how the used numerical model is able to determine the size, scale and the evolution of atmospheric systems involved. In this study numerical meteorological simulations of the most important floods occurred along the Arno river basin, Italy, in the 20th century, performed by the Regional Atmospheric Modeling System (RAMS), are analysed with regard to the sensitivity to geometrical and initial conditions. RAMS is presently used to produce operational QPFs in an integrated hydro-meteorological forecasting system for the Arno river basin. The flood events occurred in November 1966 (100-years estimated recurrence, which caused several deaths and catastrophic damages to public and private goods and to the unique artistic and cultural heritage) and in October 1992 (30-years recurrence flood, with extensive damage). The advanced modules for the parameterisation of surface-water-atmosphere exchanges and for cloud and precipitation microphysical processes, included in RAMS, allow the explicit and likely representation (triggering and evolution) of the cloud and precipitation systems, which make this model a good candidate for such sensitivity analyses. The simulations of the floods occurred over the Arno river basin were initialised by means of the NCEP/NCAR global reanalysis data, and the sensitivity of the forecasts to the spatial horizontal and vertical resolution, the representation of the sea surface temperature and the initialisation time is verified against ground data.