Based on the needs for accurate analysis of the wind field data of the wind radar (WindRAD) onboard Fengyun-3E (FY-3E), this paper uses multi-source data from buoy stations, numerical prediction model and satellites to evaluate the quality of FY-3E WindRAD wind field products with all wind speeds in multiple dimensions. The additive, multiplicative and random errors of 4 independent observation sources are analyzed by using the triple collocation (TC) method. Taking the buoys as an unbiased observation source, it can be found from the error analysis of multi-source data that the negative deviation of the additive error of FY-3E WindRAD C-band wind speed is larger, the deviation of Ku-band and dual-band wind speed is similar, the multiplicative error of C-band is the smallest, and their random errors are similar. Among the independent sources, the root mean square errors (RMSEs) of the buoys are the smallest, the two satellite products are similar and larger than those of the buoys, and the RMSEs of the model are the largest. In summary, FY-3E WindRAD can provide high-precision wind field products, but the main problem is that the consistency of products in different datasets needs to be optimized. The research results of the accuracy of satellite products based on multi-source data in this paper can provide more comprehensive and reliable information for satellite product users, and provide a basis for objective calibration methods for predicting sea surface wind field by numerical models.