Evaluation of National and International Gridded Meteorological Products for Rainfall-Runoff Modelling in Northern Italy

• DOI: 10.2139/ssrn.4919712
• Publication Date: 2024-08-08T05:18:37Z
• Authors: Sarigil Gökhan; Mattia Neri; Elena Toth
• Keywords: Physical geography; QE1-996.5; Indirect validation; Gridded meteorological products; Rainfall-runoff modeling; Indirect validation; E-OBS; ERA5-Land; CHIRPS; Northern Italy; Gridded meteorological products; Rainfall-runoff modeling; ERA5-Land; Geology; E-OBS; CHIRPS; GB3-5030

Study region: Northern Italy. Study focus: Gridded meteorological products provide spatially-distributed meteorological forcings, facilitating hydrological modeling in large-scale experiments. However, their accuracy, in particular as far as precipitation is concerned, varies considerably in space and time, and rigorous validation of these products is essential before their application. This study conducts a large-scale evaluation of five meteorological datasets in Northern Italy through i) a direct comparison of precipitation and temperature estimates and ii) an indirect validation, assessing their ability to reproduce streamflow when used to force the CemaNeige-GR6J hydrological model. The tested datasets include two gauged-based products, namely SCIA (the reference gridded dataset from the Italian Institute for Environmental Protection and Research) and E-OBS, two products based on reanalyses (the global ERA5-Land and the national MERIDA) and a gauged-corrected global satellite precipitation product (CHIRPS). New hydrological insights for the region: Gauge-based datasets provide the best streamflow simulations when the underlying station density is high: SCIA, based on a uniform and dense gauge network across the entire study area, confirms to be the best choice as the climatic reference dataset, while the use of E-OBS is not recommended in Piedmont due to the low number of stations. In areas with low station density, reanalyses may yield to more accurate results: among reanalysis-based products, the Italian MERIDA dataset outperforms ERA5-Land. Finally, CHIRPS results to be the least accurate precipitation dataset.