Research On Spaceborne Multimode GNSS Reflectometry Sea Wind Sensing Signal Processing

Wind field observations are the basis of the weather forecast,the early warning of natural disasters and the climate change.With the development of the space technology,human find out a new way to observe sea wind field,the satellite remote sensing.Remote sensing technologies enable human to observe the change of sea wind field.GNSS-R(Global Navigation Satellite System Reflectometry)is a new spaceborne technology to retrieval the sea wind,which utilizes the reflected GNSS signal over sea to measure wind.Up to now,the successful spaceborne GNSS-R projects include TDS-1 of Europe and CYGNSS of USA.This dissertation focuses on the wind retrieval signal processing technology of a spaceborne multimode GNNS-R receiver developed by China.This dissertation first discusses the basic theory of wind retrieval based on GNSSR and divides the development of GNSS-R wind field sensing signal processing into three phases.The phase one is to establish a simulation and processing system of GNSSR reflection signal.The aim of the system is to valid the signal processing algorithm,predict wind speed and provide the basic simulation data of the spaceborne GNSS-R receiver.The simulation results show the importance of developing the multimode GNSS-R receiver.Phase two is the development of the spaceborne GNSS-R signal processing algorithm.This dissertation,based on the basic structure of GNSS signal and the feature of the reflected signal,completes the GNSS-R signal processing algorithm and analyzes the effects of each signal structure on the GNSS-R technology.The signal processing algorithm and the key parameters are analyzed in detail.This dissertation improves the GNSS-R receiver in three aspects.The improvements include the prediction of the specular point,the high-sampling rate code delay DDM(delay doppler mapping)and the multimode processing.A specular point prediction method based on self-adaptive learning rate gradient descent algorithm is proposed.The specular point prediction accuracy is greatly improved while the algorithm maintains real time performance.To increase the delay bin sampling rate,a subarray-based adjustable bin interval algorithm is proposed,the algorithm uses nonuniform sampling to increase delay sampling rate while keeps the ability to tolerate the tracking error.This dissertation realizes a multimode receiver that is compatible with BDS,GPS and GALILEO.Based on the statistics of the noise and the order statistics theory,a new method to estimate the quality of DDM is proposed.This method is a better estimation than the traditional SNR based estimation.This method is the basis of improving data utilization.Phase three is spaceborne data based wind retrieval algorithm validation.This dissertation analyzes the DDM data of TDS-1 to establish quality control algorithms for the GNSS-R receiver.Five types of anomalous data are found and the algorithm to detect these data is proposed which provides the baseline for the quality control algorithms.Finally,a wind retrieval algorithm based on machine learning algorithm is proposed and the problems caused by data are analyzed.The mean square error is not reliable sometimes,therefore,Kullback-Leibler divergence is suggested to evaluate the model performance.Analyzation shows that the noisy data cause the bias in the retrieval wind and a bias modification algorithm is proposed.A location-time-based training method which improves the final result is proposed.The proposed algorithm improve the data utilization without degrading the precision.