Abstract: To evaluate the accuracy of the ERA5 wind speed data and promote its application value, this study conducted a comparative analysis between the ERA5 reanalysis data and the LiDAR wind profile data obtained during the spring voyage in the Southeast Atlantic in 2023. Results indicate that the ERA5 reanalysis data performs well in predicting low-level wind speeds but exhibits significant deviations in the boundary layer and at higher altitudes, particularly under complex weather conditions such as turbulence, wind shear, strong convection, and precipitation. To solve the above problems, we developed a deep learning-based correction method, which utilizes LiDAR-measured wind profile data as a calibration reference to optimize the ERA5 reanalysis wind speed data. It shows that the corrected ERA5 data agree well with the LiDAR observations. Under complex meteorological conditions, the correlation coefficient between the corrected ERA5 wind speed data and the measured wind speed increased by 57.80%, with the root mean square error and mean absolute error decreasing by 33.26% and 67.10%, respectively, and the bias being reduced by 75.25%. This study shows that the deep learning correction method based on LiDAR-measured data can effectively improve the accuracy of reanalysis wind speed data, providing new insights and references for numerical wind field forecasting and research on air-sea interactions.