Research On Soil Moisture Inversion Based On GNSS Multipath

Soil Moisture,also known as Soil Moisture Content(SMC),is a physical quantity to characterize the degree of Soil dryness and wetness.As an important ecosystem impact factor,SMC plays an important role in water cycle,weather forecast,and slope stability prediction.A large number of researches have shown that GNSS(Global Navigation Satellite System)satellite signals can penetrate a certain depth of soil,and their reflected signals can perceive the changes of SMC keenly,resulting in a new and efficient SMC monitoring method,namely GNSS-R(GNSS Reflectometry)microwave remote sensing technology.Compared with traditional SMC monitoring methods,GNSS-R has the advantages of all-weather,low-cost,high-precision,etc,and it is easy to realize high spatial and temporal resolution monitoring of SMC.In view of the fact that SNR(Signal-to-Noise Ratio)observations at low satellite azimuth bear the physical characteristics of the reflector around GNSS station well,and most existing SMC inversions based on GNSS-R technology take SNR as the observation value.However,GNSS-R soil moisture inversion based on SNR may have the following defects:Firstly,SNR is useless for most GNSS users,so it is not always directly recorded in the original GNSS files,so that SNR-based GNSS-R soil moisture monitoring may not be implemented;Secondly,as the system input SNR time series of GNSS-R depends largely on the performance of SNR observation qualities and the successful removal of SNR direct components(trend items),however the actual signal-to-noise ratio are easily effected by the abnormal noise of the outside environment,lead to remove the trend of the multipath signal to noise ratio obtained after reflection components might be "pure",which will inevitably reduce the inversion accuracy of the SMC.At present,all GNSS systems transmit dual-frequency and triple-frequency satellite signals,based on which a more accurate multi-path error calculation model can be constructed.Therefore,in order to enrich the data sources and methods of GNSS-R soil moisture inversion,this paper analyzes the multi-path error generation mechanism and studies the multi-path error calculation model,and proposes the three-way error based on dual-frequency pseudorange and the three-way error based on BDS.The soil moisture inversion method based on frequency-phase multi-path error,and the performance of the proposed method is verified by using the measured soil moisture data.The main research content and achievements of this paper are as follows:(1)Summarize the satellite signal situation of GNSS systems(GPS satellite system,BDS satellite system,GLONASS satellite system,GALILEO satellite system);introduce the calculation models of dual-frequency and triple-frequency multipath errors in more detail,and systematically analyze the SNR,double frequency pseudorange(DFP)multipath error,L4?IF(L4 ionosphere free)and triple frequency carrier phase(TRFCP)multipath error at low satellite altitude feature.(2)Study GNSS-R soil moisture inversion principle based on SNR observation values;In order to evaluate the accuracy of the proposed method,the calculation methods of several precision indexes are introduced.(3)A soil moisture inversion method based on dual-frequency multi-path error is proposed,and based on the GPS satellite data of the P041 site of the Plate Boundary Observatory(PBO)program in the United States,the SNR observations(SNR method),dual-frequency pseudorange observations(DFP method),and dual-frequency carrier phase observations affected by deionization(L4?IF method)were used to carry out soil moisture inversion experiments.The results show that: taking the measured soil moisture as a reference,the L4?IF method obtains a higher correlation coefficient and a smaller root mean square error,and the overall accuracy is better.(4)A soil moisture inversion method based on dual-frequency multi-path error is proposed,and based on the measured soil moisture data and the BDS tri-frequency carrier phase observation value,the BDS tri-frequency multi-path error method(TRFCP method)and the L4?IF method were used to carry out soil moisture inversion experiments.The results show that compared to the L4?IF method,the TRFCP method benefits from the high accuracy of the carrier phase observations and the three-frequency multipath error calculation model,which obtains higher correlation coefficients and better fitting accuracy.(5)Soil moisture is often affected by the combined effects of vegetation coverage,soil temperature,air humidity and other influencing factors,which lead to the conventional linear models may not be able to describe the change trend of soil moisture well.Therefore,in order to improve the accuracy of GNSS-R soil moisture retrieval,this paper constructs soil moisture prediction models based on unary linear regression(ULR),BP(Back Propagation)neural network and RBF(Radial Basis Function)neural network,and compares and analyzes their respective accuracy.