Evaluating bias correction methods for wind power estimation using numerical meteorological models

Enhancing our understanding of the meteorological factors influencing renewable energy is crucial in the energy transition, as inherent biases in widely used meteorological numerical models reduce their reliability in accurately simulating essential variables for electricity modeling. This study examines five bias correction methods for estimating wind power capacity factors, utilizing ERA5 reanalysis, Weather Research and Forecasting Model (WRF) simulations, and experimental data from multiple anemometric towers. Areas influenced by large-scale effects, such as the interaction between large-scale atmospheric circulation and orography, were accurately reproduced; however, regions with complex terrain exhibited larger errors. In some cases, the constraints imposed by large-scale features on near-surface winds are strong enough to make bias correction unnecessary. The Weibull quantile mapping and the quantile percentile method produced the lowest errors, however the latter preserved bi-modality. The mean state, linear scale, and quantile mapping Rayleigh methods produced the highest errors in 72% of the cases examined. Analysis of ERA5 revealed the dependence of its ability to reproduce the capacity factors on the conditions around the site. Bias correction alters the probability distribution’s shape, significantly impacting CF estimates through its interaction with the power curve. • Orography plays a crucial role in bias magnitude and behavior. • Correcting bias improves wind energy estimations in complex terrains. • The quantile mapping method shows lower errors.