A Comparative Study of Near‐Surface Wind Speed Observations and Reanalysis Datasets in China Over the Past 38 Years

ABSTRACT Near‐surface wind speed (NSWS) reanalysis products have been widely used in NSWS research, yet there is a lack of evaluation for NSWS reanalysis products. Based on the 10 m wind speed observations data from stations during 1980–2017, this article evaluated the climatic characteristics of NSWS over mainland China at various spatiotemporal scales using four reanalysis datasets: ERA5, MERRA‐2, NCEP‐2 and JRA‐3Q. Compared with station observations, ERA5 and NCEP‐2 showed positive biases in the climatology of NSWS, MERRA‐2 showed negative biases and JRA‐3Q showed significantly negative biases. Regarding the spatial distribution of the climatology, ERA5 performed the best with a spatial correlation coefficient of 0.55 with stations. At the interannual scale, ERA5 showed the best performance with a correlation coefficient of 0.68 with stations, significantly better than the other three reanalysis datasets. At the interdecadal scale, MERRA‐2, NCEP‐2 and JRA‐3Q can generally reproduce the decreasing trend in NSWS from 1980 to 2011. However, none of the four reanalysis datasets can capture the increasing trend from 2011 to 2017. In terms of the annual cycle, only ERA5 can well reproduce the seasonal characteristics in different regions, including the maximum NSWS in the north and the Qinghai–Tibet Plateau in spring and the south of the Yangtze River in summer. Considering the evaluation results at different spatiotemporal scales, ERA5 exhibited good performance regarding the spatial distribution of climatology, interannual variability and annual cycle, but failed to reproduce observational features at the interdecadal scale. JRA‐3Q showed significant advantages in terms of interdecadal variability, whereas neither MERRA‐2 nor NCEP‐2 showed prominent advantages in various aspects.