Predicting precipitation on the decadal timescale: A prototype climate service for the hydropower sector

• Published in: Climate services Vol. 32; p. 100422
• Main Authors: Tsartsali, E.E., Athanasiadis, P.J., Materia, S., Bellucci, A., Nicolì, D., Gualdi, S.
• Format: Journal Article
• Language: English
• Published: Elsevier 01.12.2023
• Subjects: Climate services; Decadal predictions; Hybrid forecasts; Hydropower
• ISSN: 2405-8807; 2405-8807
• DOI: 10.1016/j.cliser.2023.100422

Decadal predictions present an emerging opportunity for various socio-economic sectors affected by climate variability. However, the development of associated climate services is still in an incipient stage. This study focuses on developing a prototype climate service for an end-user in the hydropower sector. The service aimed at predicting precipitation in three drainage basins (Guadalquivir, Ebro and Po) for the next ten years, and was developed in close collaboration with the user. In this paper we do not provide the real-time forecasts, but we focus on describing and evaluating the methods and the models used. Using a European multi-model ensemble, the predictive skill for precipitation is found to vary with the calendar season, the forecast range and the drainage basin considered, though it is generally low for the purposes of supporting such a climate service. To overcome this deficiency, a hybrid approach was developed making combined use of the good skill in predicting the North Atlantic Oscillation (NAO) and the observed dominant influence of the latter on the decadal variability of precipitation in the areas of interest. Implementing this hybrid approach, which combines predictive information from the dynamical models with statistical information from observations, brings significant skill improvements in all basins during the extended cold season (November-March) for the first 10 forecast years. The hybrid model outperforms the direct multi-model ensemble output, exhibiting statistically significant skill for all basins. Our results suggest that utilising large-scale predictors can significantly improve regional predictions, and provide usable information for the hydropower sector.