| Soil moisture is the water content in the soil.As an important indicator to measure the dryness and wetness of the surface,it plays a key role in the fields of hydrogeography and meteorological science.The traditional soil moisture detection methods seriously limit the collection of large-scale soil moisture data in long time series due to many problems such as small measurement range,complicated operation and low spatial and temporal resolution.With the rapid development of GNSS-IR technology,the use of multipath errors generated by the interference effect between satellite signals to monitor surface environmental factors has become a global research hotspot.It has the advantages of high precision,strong continuity,cheap and easy to obtain.In view of the high proportion of uncorrelated noise in the signal and abnormal jumps in the inversion of a single satellite in the current inversion,a refined model combining VMD and SVRM algorithms was established,the model realizes the prediction of soil moisture while denoising the satellite signal.The main research contents of this paper are as follows:1.Based on the current development status of GNSS system,the basic principles of soil moisture retrieval by ground-based GNSS-IR are systematically introduced.Including the principle of multipath interference,the characteristics of the Fresnel reflection area and the distribution of satellite reflection trajectories,the characteristic changes between the multipath effect and the satellite signal-to-noise ratio are analyzed.The characteristic change between the signal-to-noise ratio and the satellite signal-to-noise ratio,the correlation between the phase and other characteristic changes of the reflected signal and the properties of the ground reflector is expounded.2.Studied the basic mathematical principles of VMD and SVRM,explained the basic theories and methods for the establishment of refined models.The mathematical model related to it is introduced systematically,including the process of signal denoising and accurate reconstruction,the accurate selection of the SVRM kernel function,and the use of the features of the kernel function to better realize the function of the SVRM.3.In this paper,the observation files of P041 station in 2017 are downloaded and processed,the effective satellites are selected based on the satellite reflection point trajectory map and the characteristics of the fresnel reflection area.After accurate screening,the overall trend of the inversion results of each satellite is consistent with the actual value of soil moisture.However,under extreme weather conditions are prone to defects such as large systematic errors and abnormal jumps.In this paper,the VMD method is introduced to complete the acquisition of the reflected signal,and to deal with the phenomenon that the local annual accumulation and daily inversion error is relatively large.The results show that the VMD is better in signal noise reduction.4.Aiming at the problem that the satellite signal is mixed with too much noise caused by ground environmental factors,a multi-satellite fusion refined model based on VMD-SVRM is proposed to predict soil moisture.The experimental results show that the determination coefficient of the VMD-SVRM model reaches 0.95,and the deviation is also controlled within±0.02 cm3/cm3.The single-star unimproved model with the highest comparison accuracy has increased by 23.38%,and the inversion accuracy has been significantly improved.It is proved that the VMD-SVRM refined model is effective and feasible for soil moisture inversion,and it also provides a new method for realizing high-precision soil moisture monitoring. |
PDF Full Text Request