Theories And Methods Of GNSS-R Based Snow Depth Estimation

Snow depth is one of the important environmental parameters;snow depth observations with a high temporal and spatial resolution could provide scientific foundation for global climate change,policy making and water resources management.The GNSS signals reflected by surface of objects could produce multipath errors in GNSS observations.The technologies of GNSS-R snow depth estimation make use of the multipath error to measure the snow depth.The technologies of GNSS-R snow depth estimation could be divided into two categories: interference pattern based and reflection pattern based GNSS-R snow depth esitmation.The former use geodetic GPS receiver to measure the snow depth,and it is realistic and feasible to realize a high temporal and spatial snow depth observation.However,methods included in the former are relatively less,and researches about the feature of the multipath error are not sufficient.In addition,it is also lack of the comparison and analyzation and weighted average model for snow depth observaitons obtained by differenc GNSS system and signal frequencies.To deal with the quesions mention above,theories and methods of interferometric reflectometry based GNSS-R snow depth estimation are systematically and deeply studied in this paper.The model and features of GNSS interferometric signals are analyzed at first.Then,the methods of SNR-based,carrier phase combination-based,and pseudorange and carrier phase combination-based snow depth estimation are introduced.The snow depth estimation errors for all the methods are compared and analyzed by using GNSS observations.At last,a weighted average method of snow depth estimations is presented.The main contents and innovation points of the paper are lists as follows:(1)The model of amplitude attenuation factor which is used to measure the strenght of reflected signals is estiblished;and this model improves the modeling of GNSS multipath error.(2)Snow depth estimation errors of SNR-based method for BDS,GPS and Galileo are compared and analyzed in the paper.The results shows that,the SNR-based snow depth measurements exsit a significant nagtive error;averaging the multi-satellites snow depth measurements could improve the accurancy and reliability of snow depth.(3)The dual-frequceny carrier phase comibation methods and the triple-frequency carrier phase combination method are studied.Moving average algorithm is used to extract the multipath error for dual-frequency carrier phase combinations.By analyzes and simulation,the theoretical models are established for snow depth estimation based on dual-frequcney carrier phase combination for BDS,GPS,and Galileo.By using the GNSS observations,the estimation errors of those methods for BDS,GPS,and Galileo are compared.The result shows that,the triple-frequency carrier phase combination method is more accurancy than dual-frequency carrier phase combination method.(4)Two new GNSS-R based snow depth estimation methods are proposed: single-frequency pseudorange and carrier phase combination method,and dual-frequency pseudorange and carrier phase combination method.The features of combined multipath error for the two combination methods are analyzed and theoretical models for estimating snow depth are also estiblished.By using the GNSS observations,the estimation errors of those methods for BDS,GPS,and Galileo are compared.The results shows that,the accurancy of pesudorane has a significiant effect on the snow depth observations;and the shorter periods of pseudorange code is,the more accurancy of snow depth could obtained.(5)A new weighted average method is proposed for snow depth estimation using observations of multiple satellites.By analyzing the relationship betweeen obsevation noise and peak spectral density and that between the noise and the estimation error of peak spectral frequency,the model of weighted average is presented;it is shows that the measurement noise is proportional to the error of peak frequency estimations.The model is proved to be more effective and accurate than the normal average method.