Research On Sea Level Height Inversion Using Shore-based Multi-mode And Multi-frequency GNSS-R Signals

Global warming has caused sea level rise,sea ice melting,and coastal erosion,which seriously threatens the safety of coastal areas.Sea level changes have a profound impact on coastal areas.At the same time,sea level can be used for ocean circulation analysis and tidal model establishment,as well as meteorological model research and tsunami warning,it’s an indispensable physical parameter in marine scientific research.Therefore,the monitoring of sea level in coastal areas is of great significance.GNSS-R(GNSS-Reflections)sea level altimetry has the advantages of all day,low cost,and rich data sources,it makes up for the shortcomings of traditional ocean altimetry methods and spaceborne radar altimetry,and provides a wider range of options for sea level height measurement.This paper takes GNSS-R sea level altimetry as the core content of the research,and improves the GNSS-R sea level altimetry method in terms of inversion accuracy and time resolution.The main contents and conclusions are as follows:1.The related knowledge including GNSS-R altimetry geometric model and reflected signal characteristics are introduced.Build a simulation model of multipath and signal-to-noise ratio(SNR),and discuss the relationship between the two sides: Affected by the multipath effect,the SNR will contain more physical information of the reflector at low satellite altitudes.This conclusion provides a theoretical basis for the inversion of sea level height based on the SNR.2.In view of the problems of poor signal separation and low inversion accuracy caused by poor signal separation and large deviation in the process of SNR inversion of sea level,a sea level inversion method using Gaussian multi-peak fitting for signal separation was proposed.Taking the tide gauge data as the true value,the data of different stations and different satellite systems are selected for verification,and compared with the polynomial and wavelet filtering methods.The results show that the Gaussian fitting has a higher fitting degree to the signal trend term,and the inversion value can better reflect the sea level change.Compared with the inversion results of the polynomial and wavelet filtering methods,the accuracy of the GPS inversion results of the Gaussian fitting at SC02 station was improved by 30% and 20%respectively,and the accuracy of the BDS inversion results of the MAYG station was improved by 29% and 19%.The results showed that all indicators were improved,indicating that the use of Gaussian fitting method for sea level height inversion is feasible.3.Aiming at the problem of poor resolution in the inversion process of GNSS-R sea level height,the idea of sliding window was introduced,and the inversion was carried out by combining two methods of wavelet filtering and Gaussian fitting.The results show that the number of valid points of the inversion results of wavelet filtering combined with sliding window method is increased by 5 times,and the number of valid points of the inversion results of Gaussian fitting combined with sliding window method is increased by 5.4 times,The time resolution has been greatly improved.The inversion results of the two methods are basically consistent with the actual situation of the tide gauge,and both results show strong correlations.At the same time,the accuracy of Gaussian fitting combined with sliding window inversion results is 10% higher than that of wavelet filtering combined with sliding window inversion results,and the correlation coefficient is higher.It shows that the Gaussian fitting windowing method is better than the wavelet windowing method.4.Using wavelet filtering and Gaussian fitting methods,and combined with sliding windows,the joint inversion of sea level height from multi-mode and multi-frequency data is studied.The results show that the indicators of the multi-mode and multi-frequency joint inversion results are between the indicators of the inversion results of the single-frequency data of each system,and the inversion results using Gaussian fitting with and without sliding window are better than those of the inversion results using wavelet filtering.Compared with the single-frequency data inversion results before windowing,the time resolution of the multifrequency data joint inversion results of each system is increased by 2-3 times,and after windowing is increased by 4-8 times.Time resolution is greatly improved.