Wind product datasets are widely used in research across different temporal scales, such as long-term wind energy potential assessments and studies on global climate change, and short-term numerical modeling of extreme weather events. However, the performance of wind products across interannual, seasonal, and synoptic scales remains inadequately understood. This study addresses these gaps by evaluating blended and reanalysis wind products using in-situ buoy data along the southeast coast of China during typical El Niño (2015/2016) and La Niña (2017) years. Four commonly used datasets were evaluated: the blended wind product Cross-Calibrated Multi-Platform version 3.1 (CCMP V3.1) and the reanalysis wind products Climate Forecast System Version 2 (CFSv2), the fifth generation of ECMWF atmospheric and oceanic reanalysis (ERA5), and the Japanese 55-year Reanalysis (JRA-55). The main findings are as follows: (1) CCMP V3.1 emerges as the most accurate dataset, showing reductions of 5.25%–23.70% in RMSE of wind speed, 3.71%–17.33% in RMSE of wind direction, and 26.46%–84.26% in RMSE of wind power density (WPD) relative to the other three datasets. (2) ERA5 ranks next in wind speed accuracy, with RMSEs 13.10% and 19.47% lower than those of CFSv2 and JRA-55, respectively. Conversely, its WPD RMSE exceeds those of CFSv2 and JRA-55 by 112.87% and 367.26%, respectively. This contrast highlights that wind speed accuracy alone does not guarantee reliable assessments of wind energy potential. (3) On the interannual scale, the bias in wind products shows a positive correlation with the El Niño–Southern Oscillation (ENSO) index, indicating better performance during La Niña years. The average WPD across the three sites is 18.32% higher during La Niña years compared to El Niño years, a variation accurately captured by CFSv2 (14.70%) and JRA-55 (14.62%). (4) On the synoptic scale, as tropical cyclones (TCs) approach and depart from buoy stations, wind products exhibit two bias peaks compared to buoy data. These peaks are asymmetrical, reflecting the inherent asymmetry in TC structure. Additionally, high TC translation speeds lead to larger biases due to the insufficient temporal resolution of the wind products.
Zhang et al. (Fri,) studied this question.