A Critical Assessment of Precipitation Datasets Over Italy

Italy, because of its complex orography and geography, is prone to extreme precipitation events, which may result in enormous damages and losses. It is therefore fundamental to monitor precipitation across the Italian territory and across time. Even though Italy has been a pioneer in developing meteorological observations, its rain gauge network – as those of many other countries in the World –is characterised by an uneven density and with records not always freely available online. Satellite and reanalysis precipitation datasets have the potential to overcome some of the issues characterising ground-based monitoring networks, but their performances vary widely across climate, topography and time scale. In this work we assess the performance of six remote sensing and reanalysis datasets (ERA5-Land, CERRA-Land, CHIRPS, CMORPH, IMERG, PERSIANN-CCS-CDR) in observing precipitation across the entire Italian peninsula, using the ground-based national records of the SCIA dataset as ground-truth. For our analysis, we compute common continuous and categorical metrics across different time scales (daily, monthly, annual). We then provide the best performing dataset at different spatial scales (i.e. watershed, administrative province, administrative region, nation-wide, Köppen-Geiger climate zone), providing useful insights for hydrological studies of various purposes. Our results show that at the national level, the two reanalysis datasets (CERRA-Land and ERA5-Land) outperform the satellite-based observations, having overall higher and consistent performances across the different climatic zones. Among the satellite datasets analysed, the most-performing is IMERG, while the least-performing in all the Italian climatic regions are CMORPH and PERSIANN-CCS-CDR, with the worst performances in the alpine and cold semi-arid climates, respectively.