DATE2022-05-12 16:22:53
TITLEPredicting precipitation at decadal timescale — Developing a climate service for the energy sector in Southern Europe.
AUTHORSEirini E. Tsartsali (1,2) ,Panos J. Athanasiadis (1) ,Stefano Materia (1,3) ,Alessio Bellucci (1,4) ,Dario Nicolì (1) ,Silvio Gualdi (1)
  1. 1) Centro Euro-mediterraneo Sui Cambiamenti Climatici Foundation (cmcc), Bologna (Italy) ,2) Department Of Physics And Astronomy, University Of Bologna, Bologna (Italy) ,3) Currently At: Barcelona Supercomputing Center (bsc), Barcelona (Spain) ,4) Currently At: Istituto Di Scienze Dell’atmosfera E Del Clima - Consiglio Nazionale Delle Ricerche (isac-cnr), Bologna (Italy)
ABSTRACTDecadal predictions have rapidly evolved in the last decade, and now are produced operationally worldwide to bridge the gap between seasonal predictions and climate projections. Skillful decadal predictions present an emerging opportunity for the development of climate services to assist planning and decision making by governments and businesses in various socio-economic sectors. On these grounds, the EU Copernicus Climate Change Service (C3S) aims at revealing the potential benefits of decadal predictions for different industries and at developing real-time, sector-specific decadal prediction products. Four European institutions participated in the C3S_34c tender, each one developing a decadal climate service working with an end user from a specific sector (insurance, agriculture, infrastructure and energy). CMCC developed a prototype climate service based on decadal predictions for ENEL Green Power in the energy sector, focusing on hydropower investments and operations in three European drainage basins, Guadalquivir and Ebro in Spain and Po in northern Italy. The end user was particularly interested in the amount of water available in the reservoir, which is primarily affected by the basin-integrated precipitation, and the safety of the dams related to precipitation extreme events that can cause dam failure and flooding. After discussing with the end user different forecast possibilities and challenges, it was decided to proceed with forecasting precipitation amounts over each of the three catchment areas. Using initialised experiments from four decadal prediction systems (DePreSys4, EC-Earth3, CMCC-CM2-SR5, MPI-ESM-HR), we first assessed the direct multi-model output for different calendar seasons and forecast year-ranges. Even though statistically significant skill was found in some cases, it was too low for the purposes of the climate service. Therefore, in order to meet the needs of the end user we explored the possibility of using large-scale predictors that can influence the precipitation in the basins. The North Atlantic Oscillation (NAO), one of the leading modes of atmospheric variability in the Euro-Atlantic sector, can drive climatic anomalies in the mediterranean area and strongly controls precipitation in the three catchment areas. Focusing on the extended cold season, from November to March, we found that NAO drives a large part of the precipitation variability in the drainage basins with high (anti-)correlations in decadal timescales. Moreover, statistically significant skill was obtained for the NAO index during the extended cold season, especially after integrating over more forecast years. The skilful NAO predictions and the high (anti-)correlations between NAO and the aggregated precipitation anomalies over each basin, allow the usage of the NAO index as a statistical predictor for precipitation. Therefore, we built a hybrid model to predict precipitation using the dynamically predicted NAO from the multi-model ensemble and the statistical (linear) relationship between the observed NAO and precipitation in each basin. Taking into account both the available skill and the needs of the end user we focused on the extended cold season for the forecast range 1-10 years. Using this hybrid approach, the precipitation predictive skill was increased significantly in all drainage basins. Last, we investigated whether developing similar hybrid models could lead to skillful predictions of precipitation extremes and indices of interest for the end user. No statistically significant skill was found for the precipitation extremes, defined as the 95th percentile of the aggregated precipitation in each basin. However, good skill was obtained for the number of wet days (number of days with at least 0.1 mm of rain) during the extended cold season which can be valuable information for the end user, since it is a good indicator of the distribution of the precipitation events during the specific season. Climate services on multi-annual timescales are still at an early stage, and various challenges remain to be addressed, e.g. finding ways to increase skill and the spatial resolution of the forecasts. However, this study shows the high potential of the decadal predictions to become useful to end users for their operations and planning. Large-scale predictors, can significantly improve the regional predictions and provide useful sector-specific information.