Detailed information of the observation datasets.

There are abundant wind energy resources along the coast of China. Understanding spatial-temporal characteristics of wind speed is significant in meteorology, coastal engineering design and maritime industries. Reliable wind products such as reanalysis data, coupled with accurate wind speed measurements, are essential for elucidating the primary characteristics of the wind field. In this study, we evaluated hourly 10 m and 100 m wind speed data from the fifth-generation ECMWF atmospheric reanalysis (ERA5) by comparing it with direct wind measurements obtained from 19 wind tower located across the coastal waters of China. The results are as follows: 1) the basic statistical characteristic between ERA5 reanalysis and observed wind speeds demonstrate good consistency. However, the ERA5 tends to underestimate wind speed, particularly at high speeds during extreme conditions. 2) Compare ERA5 data with observations from each station using a frequency distribution-based score method, hourly scores of most stations are between 0.8 to 0.9. It shows the higher simulation skill in the northern region than the southern due to the influence of high-frequency typhoon in the South China Sea. 3)Distribution function parameters, mean values, variability, and wind threshold frequencies were analyzed for this ensemble of observation, providing an overall description of wind characteristics. Generally speaking, there is no clear linear relationship between scores and the other variables. On longer time scales (6–24 hours), the score and correlation between ERA5 and observations further increased, while the centered root-mean-square error (CRMSE) and standard deviation decrease. 4) Hourly wind data with a regular spatial distribution in ERA5 reanalysis provides valuable information for further detailed research on meteorology or renewable energy perspectives, but some inherent shortcomings should be considered.中国沿海蕴藏着极为丰富的风能资源。明晰风速的时空特征，在气象学、海岸工程设计以及海洋产业中均具有重要意义。可靠的风力产品（如再分析数据）结合精准的风速实测数据，是阐明风场核心特征的必要基础。本研究针对欧洲中期天气预报中心（ECMWF）第五代大气再分析数据（ERA5）所提供的逐时10米与100米风速数据，选取中国近海水域19座测风塔的直接实测风速作为对照，开展了对比评估。研究结果如下：1）ERA5再分析数据与实测风速的基础统计特征展现出良好的一致性，但ERA5整体存在风速低估现象，在极端工况下的高风速段尤为明显。2）采用基于频率分布的评分方法对各站点的ERA5数据与实测数据进行比对，多数站点的逐时评分介于0.8至0.9之间。受南海高频台风活动影响，北方区域的模拟表现优于南方区域。3）针对本次实测数据集，我们分析了分布函数参数、平均值、离散度以及风速阈值频率，以此全面刻画了区域风特征。总体而言，评分与其余变量之间未呈现明确的线性相关关系。在更长的时间尺度（6–24小时）下，ERA5与实测数据间的评分及相关性进一步提升，而中心化均方根误差（CRMSE）与标准差则有所降低。4）ERA5再分析数据所包含的具有规则空间分布的逐时风速数据，可为气象学或可再生能源领域的后续深入研究提供极具价值的参考，但同时也需考虑其固有的局限性。