Based on the partial least square regression, the nearsurface wind speed predicted by Zhejiang WRF ADAS Real Time Modeling System in winters of 2013-2016 is corrected and evaluated. The forecast product is used as the independent variable while the corresponding observational data of the automatic weather stations are used as the dependent variable. At most of the 956 stations, the correction of wind speed has been improved. Through quantitative analysis, we find that the improvement effect is best for the stations in the western region of Zhejiang, up to 91.7%. The improved stations in the central part account for 86.5%. The improvement of the stations in the eastern coastal region is slightly worse, with 67% obviously improved stations. The improvement percentage of wind speed correction, indicates the station performance of each city is good. Except for Zhoushan (49.9%), the rate of wind speed improvement is more than 50% in other cities. Based on the analysis of the gale process in the northeastern coast Zhejiang on 20 January 2017, the corrected wind speed is closer to the observation, especially at the nearshore stations.