S. Sperati

Characterization of forecast errors and benchmarking of renewable energy forecasts

Abstract The accuracy of the wind and solar power predictions strongly depends on the quality of meteorological forecasting, improvements of which are principally related to the increasing computational power, to the larger availability of meteorological observations, and to effective data assimilation techniques. Benchmarking exercises, where different prediction models and techniques are compared, represent a useful method to assess the state of the art of the wind and solar power predictions and to address the most recent developments in this evolving scientific topic. This chapter presents and discusses two benchmarking exercises organized by the European R&D project ANEMOS (ENK5-CT-2002-00665) and by the European Cooperation in Science and Technology (COST) Action ES1002 Weather Intelligence for Renewable Energies (“WIRE”). The first exercise involved 11 state-of-the-art models that were run for six test cases across Europe with different terrain characteristics and were evaluated under an appropriate protocol. The goal was to establish the quality level of the most advanced wind power prediction systems available in the first half of the 2000s. The benchmark exercise organized by the COST action WIRE verified the performance of a wide range of modeling approaches in use during the first half of the 2010s for both wind and solar power forecasting. These benchmarks allowed obtaining a better knowledge of the state of the art in both wind and solar power forecasting, with an overview of the main approaches in use, and assessing the evolution of their performances during the period 2000–14.