| When retrieving sea surface wind speed using Global Navigation Satellite System Reflectometry (GNSS-R) signals, the Normalized Bistatic Radar Cross Section (NBRCS) and Leading Edge Slope (LES) are first extracted from the Delay-Doppler Map (DDM). Subsequently, corresponding Geophysical Model Functions (GMFs) relating them to sea surface wind speed are established respectively. Then, the Minimum Variance Estimator (MVE) and Particle Swarm Optimization (PSO) methods are employed to combine the wind speed retrieval results from the NBRCS GMF and LES GMF, with the aim of achieving better accuracy than that obtained from a single observation.A comparative analysis of the wind speed retrieval results from the NBRCS GMF, LES GMF, MVE method, and PSO method shows that both the MVE and PSO combined retrieval methods achieve an overall root mean square error (RMSE) of 1.79 m/s and a correlation coefficient of 0.75. However, the systematic bias of the PSO method is slightly larger than that of the MVE method. Compared to the NBRCS GMF and LES GMF, the two combined retrieval methods reduce the RMSE by approximately 5.3% and 16.0%, respectively.The results indicate that the combined retrieval method, which fuses multiple observations, is superior to the traditional single-observation GMF method, and it can enhance both the reliability and the accuracy of GNSS-R wind speed retrieval. |
