Analog versus multi-model ensemble forecasting: A comparison for renewable energy resources. (March 2023)

• Record Type: Journal Article
• Authors: Pappa, Areti; Theodoropoulos, Ioannis; Galmarini, Stefano; Kioutsioukis, Ioannis
• Is Part Of: Renewable energy. Volume 205(2023)
• Page Start: 563
• Page End: 573
• Publication Date: 2023-03
• Subjects: Analog ensemble -- Multi-model ensemble -- Wind speed -- Wind power -- Solar radiation -- Solar power; AnEn Analog Εnsemble -- AnEn-B Analog ensemble of the best model -- AnEn-M Analog ensemble of the mme -- mme multi-model ensemble mean -- mmeW weighted multi-model ensemble mean -- HM Hybrid Model -- NWP Numerical Weather Prediction; Renewable energy sources -- Periodicals; Power resources -- Periodicals; Énergies renouvelables -- Périodiques; Ressources énergétiques -- Périodiques; 333.794
• DOI: 10.1016/j.renene.2023.01.030
• Languages: English
• ISSNs: 0960-1481

Abstract

Abstract: To satisfy the energy demand from renewable sources, accurate weather predictions are necessary. The Analog ensemble (AnEn) technique aims to correct a (weather) prediction given historical observational data. In this work, the AnEn is applied to the wind speed and solar radiation predictions used in the AQMEII multi-model ensemble, spanning a whole year, to produce probabilistic forecasts over Europe. The skill of each deterministic model in forecasting the wind speed, the solar radiation and the respective renewable energy potential is compared to the skill of the AnEn as well as to the skill of the multi-model ensemble mean, either unconstrained (mme) or analytically optimized (mmeW). Results show that the AnEn significantly improves the wind (radiation) forecast skill of the numerical models in the range 25–43% (13–24%), being larger for moderate or low skill models. Compared to mme, the AnEn improvement is larger across all quartiles except the upper one. AnEn and mme are mostly comparable with the mmeW at intermediate values of wind speed and solar radiation. At higher values, the AnEn should benefit from additional auxiliary inputs and a larger dataset. A hybrid model combining the advantages of AnEn and mmeW and providing even more accurate forecasts is proposed.