Abstract: To enhance the forecasting accuracy of near-surface wind speed under complex terrain conditions, this paper firstly utilizes wind field observation data from 60 national and regional meteorological observation stations in the Chongli Zone of Zhangjiakou for the 2022 Beijing Winter Olympics, adopts the statistical deviation correction method to obtain the hourly average deviation of wind field background field within the forecast time limit of 1-12 h from September 2021 to January 2022. Then, the hourly average deviation field is used to correct the deviation of the Hebei Provincial Meteorological Bureau 50 m resolution fusion analysis and proximity forecasting system (INCA-50 system) background field, and obtain the wind field forecast results by using the corrected background field. Finally, the forecast effect of refined wind field after deviation correction is verified by comprehensive comparison of wind field and case test. The results are as follows. (1) Compared with INCA-50 grid forecast products, the systematic deviation of wind speed after deviation correction is obviously reduced, and the absolute error of wind speed is reduced by up to 36% within 12 h forecast time limit, and the improvement rate of average absolute error of stations is up to 30%. (2) The test results of an example show that, the deviation correction method can effectively make up for the shortage of sparse observation stations. After the deviation correction, the problem that the wind speed prediction result of INCA-50 system is too large can be effectively improved, and the predicted wind speed is closer to the observed wind speed.