This Master’s thesis analyses the forecasts of meteorological variables (precipitation and temperature) in the northwestern Italian regions of Piedmont and Aosta Valley, with the aim of evaluating the accuracy along the year through a statistical analysis of forecast errors. In the preliminary phase of the analysis a systematic classification was performed of available forecast and observational datasets, based on criteria such as lead time, spatial and temporal resolution and extent, and data accessibility. This step allows the identification of most suitable datasets for the study purposes, that are: the forecast dataset derived from the ICON-2I numerical model, including its 20 ensemble members, made available through the MeteoHub platform by Meteo Italia, and the observational dataset NWIOI provided by ARPA Piemonte.