Research On Snow Depth Retrieval Based On GNSS-IR

GNSS-IR,as a new means of surface environmental monitoring,has become a research hotspot in snow depth inversion with the advantages of high precision,low cost and high time resolution.In this paper,the basic principle of GNSS-IR snow depth inversion is described in detail,the relevant research and improvement are carried out to solve the problems existing in the existing GNSS-IR snow depth inversion method.The main research work includes the following aspects:1.Take the observation data of P351 tracking station during three consecutive snowfall cycles as an example.According to the principle of GNSS-IR snow depth inversion based on signal-to-noise ratio,Fast Fourier Transform and Nonlinear Least Squares Fitting are introduced on the basis of the existing Lomb-Scargle Periodogram spectrum analysis.These three algorithms are used to perform GNSS-IR snow depth inversion.Through comparative analysis of the inversion accuracy of the three algorithms,a better GNSS-IR snow depth inversion combined algorithm is proposed.2.Multi system and multi frequency GNSS-IR snow depth inversion experiments are carried out using multi system observation data.By analyzing the inversion results of different systems and frequencies,the best inversion frequency of different systems is selected.The best inversion frequencies are combined,the results show that the accuracy of GNSS-IR snow depth inversion based on multi system and multi frequency combination is higher than that of single system,and the time resolution is significantly improved.3.Based on the GNSS-IR snow depth inversion results of P351 tracking station in three consecutive snowfall periods,the law of the error caused by surface tilt on GNSS-IR snow depth inversion is analyzed.The relationship between surface tilt error and surface tilt angle and satellite altitude angle is analyzed,and the corresponding improvement method of surface tilt error is proposed,and the accuracy of GNSS-IR snow depth inversion is improved by 11.6%.4.Through the long-term snow depth inversion results,it is found that there is a significant correlation between the snow depth inversion error and the snow depth,and machine learning is introduced to train the existing inversion snow depth and inversion error to construct the empirical model of the inversion snow depth and error.Then through inversion snow depth,the inversion error is predicted and corrected.