A Comparative Performance Study of WRF, COSMO and ICON Atmospheric Models for the Italian Peninsula at Very High Resolution

This study seeks to estimate the precision of three different high-resolution atmospheric models in the simulation of two sets of short-term weather forecasts for a duration of two-weeks for the Italian Peninsula. The following models were used: WRF (Weather Research and Forecasting Model), COSMO (Consortium for Small-scale Modelling) and ICON (Icosahedral Nonhydrostatic Model). The capability of these weather forecasting systems has been evaluated using their optimum-configurations, which were obtained from a tuning procedure at a spatial resolution of about 2 km over Italy. The models share the same forcings given by the Integrated Forecasting System (IFS) analyses provided by the European Centre for Medium-Range Weather Forecasts (ECMWF) with a gridpoint distance between neighbouring points of approximately 0.081° (∼9 km). This study was performed over two periods: from 01/01/2019 to 15/01/2019 and from 16/08/2020 to 30/08/2020. These periods were chosen to characterise the models’ performance over the winter and summer seasons respectively. The precision of these weather forecasting systems were evaluated by taking their best-configurations, which were obtained from a tuning procedure at a spatial resolution of 0.018° (∼2 km) over the domain specified. The ERA5-Land reanalysis, which was provided by the Copernicus Climate Change Service (C3S) and the gridded SCIA observed data (SCIA - Sistema nazionale per la raccolta, elaborazione e diffusione di dati Climatologici di Interesse Ambientale). The results for each model reveal that the variables analysed here are all consistent with respect to those observed, as they capture the main features that characterise the summer and winter weather conditions investigated here. Those differences observed among the models may be related to the complex parameterization schemes used in WRF, COSMO, and ICON, that could affect the models’ performance.