Application Of Semiparametric Model To Troposphere Fitting In Hongkong

The atmosphere can be divided into troposphere,stratosphere,middle layer and thermosphere according to the thermal state.The atmospheric layer below the ground height of 60 km will affect the propagation of electromagnetic waves.the atmosphere can be divided into two parts of the troposphere and the stratosphere,because the delayed 80% occurs in the troposphere,so is called the tropospheric delay error.Tropospheric delay can be divided into two parts: dry delay and wet delay.There are many parameters that influence tropospheric delay,including meteorological parameters and geographical location.geographical location of the station and altitude have great influence on the troposphere,and can we know the tropospheric delay of the station only according to the geographical location of the station and the altitude fitting region tropospheric model.At present,most of the models for tropospheric delay are based on the global average atmosphere.It's difficult to satisfaction of wants for the precision of the baseline solution and positioning,but with the establishment of regional CORS network in china,some achievements have been obtained in the establishment of an accurate regional tropospheric delay model.The tropospheric delay is not only related to the geographic location of the station,but also to the meteorological parameters of the station.Because the conventional measuring instrument doesn't determine the function of the meteorological parameters,the tropospheric fitting model is built by the location of the station and the altitude information,and the error is unavoidable.Therefore,eliminating errors is particularly important for improving the accuracy of tropospheric fitting models.In 1977,Stone put forward a semiparametric statistical model.This model adds nonparametric components,which provides a new idea for estimating errors.The semiparametric model contains parameter components and nonparametric components,which can simultaneously have the advantages of nonparametric and parametric models.Therefore,the software used in this study is the high precision GNSS data processing software GAMIT,and the station data in Hongkong satellite positioning reference network are used.The GAMIT software is used to calculate the tropospheric delay of station in the reference network,and the tropospheric fitting model of Hongkong area is established with the surveying station coordinates provided by the Hongkong General Administration.The tropospheric delay values obtained by different fitting models are compared with the tropospheric delay values obtained by the GAMIT software,and the tropospheric fitting model which is the most suitable for Hongkong area is selected.The semiparametric model is applied to the tropospheric fitting model of Hongkong area.The main contents are as follows:(1)The commonly used tropospheric delay fitting model is introduced,and the two fitting methods of least square and the whole least square are used to fit the Hongkong area,and the tropospheric fitting model suitable for Hongkong area is selected by comparison.The result shows that the total least squares method is more accurate than the least square method in fitting the troposphere model.Through different fitting models,different fitting methods are used to fit the tropospheric delay in the experimental area.By comparison,the most suitable tropospheric fitting model for the experimental area is H1QM4.(2)The error is unavoidably contained in the regional tropospheric delay fitting model based on the observation data of CORS network and the geographical location and elevation of the station.The model accuracy is often not high,so it is necessary to eliminate the error.In this paper,the method of eliminating the error is selected as a semiparametric model,so the theory and method of the semiparametric model are studied,the characteristics of the semiparametric compensation least square estimation solution are discussed,and the method for determining the regular matrix and the smoothing factor is introduced in detail.(3)In this paper,semiparametric model is applied to eliminate the error of H1QM4 selected in the experimental area,and experiment with MATLAB software.Compare and choose different methods to determine smoothing factor and regular matrix,In order to eliminate the error in troposphere fitting model,we choose the most suitable smoothing factor and regular matrix in semiparametric model.The tropospheric delay values obtained by the H1QM4 model,the least square fitting H1QM4 model and the whole least square fitting H1QM4 model are compared with the GAMIT software to obtain the tropospheric delay.The error of the parameters calculated from the mobile station shows that the accuracy of the least squares fitting model is increased by about 1mm.The fitting accuracy of semiparametric H1QM4 model is increased by 2mm compared with least square fitting.