Study On Extraction Method And Application Of Time-variable Gravity Signal Detectd By Satellite

Since the implementation of the new generation of satellite gravity mission,especially with the successful launch of the GRACE satellites,satellite gravity has been widely used in many disciplines.The time-varying gravity field model based on gravity satellite data acquisition can retrieve the surface mass redistribution,thereby enhancing the understanding of the movement of the Earth's systems.In the method of inversion of surface mass redistribution by time-varying gravity field model,the spherical harmonic coefficient method is the most commonly used method in the application of current time-varying gravity field model because of its convenient use and high computational efficiency.However,this method traditionally requires de-correlation filtering and spatial smoothing filtering on the time-varying gravity field model to acquire signals,which leads to the problem of signal weakening or deformation.In the verification of time-varying signals of gravity satellites and other observation methods,GPS measurement has become a more commonly used verification method because of its high accuracy and continuous observation and global distribution of stations,but there are still insufficient verification problems.With the termination of the GRACE mission and the subsequent launch of the GRACE Follow-On satellite,during this period the L-L SST(Low-Low Satellite-to-Satellite Tracking)mode was interrupted to monitor global mass changes,then the H-L SST(High-Low Satellite-to-Satellite Tracking)observations of other low-orbit satellites(especially the Swarm satellites)can whether to fill this data gap is also a question worth exploring.Based on the above three problems,the main work and achievements of this thesis are as follows:(1)The theory and method of inversion of surface mass redistribution using time-varying gravity field are systematically described.The principle of spherical harmonic coefficient method for inversion of surface mass redistribution using time-varying gravity field model is emphasized,and the basic formula of inversion is derived.The filtering methods in the post-processing process are introduced in detail,and the methods such as Gauss filtering,Fan filtering and decorrelation filtering as well as the filtering results are compared and analyzed,and the shortcomings of the traditional filtering methods are pointed out.The error corrections required after extracting geophysical signals,including truncation error,signal leakage error and GIA(Glacial Isostatic Adjustment),are discussed.The methods commonly used for correcting each error are given.Finally,the accuracy of the time-varying gravity field model is evaluated,and the error estimation method for the mass redistribution of time-varying gravity field inversion based on the law of error propagation is described.(2)Aiming at the problems brought by traditional filtering methods in extracting geophysical signals from GRACE time-varying gravity field model,the method of extracting time-varying signals by Kalman filter based on prior information and multi-scale factor method of repairing the extracted signals in the set of post-processing methods are studied.The test results show that this post-processing method can significantly suppress the influence of high frequency noise and strip noise.Because no spatial filtering is needed,Kalman filtering method completely retains the original spatial resolution of the signal,retains more details and improves the signal-to-noise ratio.Compared with the results of the mascon product,both the long-term trend signal and the annual amplitude and phase are consistent.(3)The seasonal surface vertical load deformation based on GRACE satellite observation is obtained by using traditional filtering method,Kalman filtering method and mascon products.The correlation between seasonal vertical displacement and GPS observation is evaluated by using correlation coefficient,WRMS reduction ratio of GPS vertical displacement and annual amplitude reduction ratio of GPS vertical displacement.From the results of 1293 GPS stations tested,it can be seen that about 90% of the GPS observations contain GRACE detection signals.On average,GRACE observation data can interpret more than 50% of the signals.Compared with the other two GRACE data processing methods,Kalman filter method performs the best.The median of the three indicators reaches 0.72,0.26 and 0.51,respectively.Compared with the previous research results,Kalman filter method has a certain improvement.(4)Using a method to approximate the GRACE data released by each organization to obtain the surface load deformation accuracy,the internal precision of the RL05 and RL06 data of the three organizations CSR,GFZ and JPL were estimated respectively.It was found that the data accuracy of the latest versions of each organization was improved by more than 70%,but the time series of surface load deformation obtained by the latest version of GRACE data has no significant improvement compared with the GPS vertical displacement time series.(5)A GPS station selection strategy that takes into account the spatial resolution of GRACE and reduces the interference of subjective factors is proposed,so that the correlation between GRACE and GPS signals can be comprehensively evaluated on a global scale.By gradually increasing the size of the grid to gradually reduce the station data,it is found that the correlation between the two measurement methods is gradually enhanced,the most important reason is the increase in the average observation time of all stations.Therefore,when GPS data is used to verify the observation signal of GRACE on land,the verification result is closely related to the selected GPS station,especially the observation time of the station is an important factor.(6)By comparing and analyzing the degree errors and degree correlation coefficients between the H-L SST models of various LEO satellites and the L-L SST models of GRACE satellites,it is found that the reliable degree of the H-L SST models of CHAMP,GRACE and SWARM satellites for inversion of surface mass redistribution is below the 12 th to 15 th degrees,while the GOCE time-varying model is not suitable for inversion of surface mass redistribution.(7)The method of spectral combination absorbs the advantages of various satellite data,which improves the accuracy of the combined model.A combination of CHAMP,GRACE and SWARM satellite H-L SST models is proposed,which improves the reliability degree of the model from the highest 15 th to 22 th,and thus improves the spatial resolution of the combined model for inversion of surface mass redistribution.By comparing and analyzing the signals from nine regions of the world and using the combined model of the above three satellites,it is verified that the H-LSST model of LEO satellite can detect the surface mass redistribution at the spatial resolution level of 900 km.In addition,the latest quality changes in Greenland were obtained using Swam satellite data over five years.