Research On Precipitable Water Vapor Retrieval Based On Multi-GNSS Precise Point Positioning

Water vapor is the most active factor in the water cycle and plays a vital role in the transmission of atmospheric energy.The change of water vapor phase is closely related to the occurrence of precipitation events.The water vapor content in the atmosphere is the material basis for the formation and evolution of catastrophic weather,and is the main parameter for real-time monitoring and accurate forecasting of storm weather.Therefore,real-time monitoring and accurate prediction of atmospheric water vapor are of great significance.Atmospheric water vapor detection based on the Global Navigation Satellite System(GNSS)has the advantages of high precision,all-weather,low cost,and high temporal spatial resolution.It can overcome the shortcomings of traditional atmospheric water vapor detection methods.The Zenith Wet Delay(ZWD)and atmospheric water vapor conversion coefficient are two important parameters for ground-based GNSS retrieval of Precipitable Water Vapor(PWV).At present,ZWD based on GNSS technology mainly has two modes of relative positioning and precise point positioning.In order to effectively weaken the influence of the tropospheric delay spatial correlation,the relative positioning mode needs to additionally introduce observation data from long-distance base stations to participate in the calculation,thereby increasing the data processing complexity of ground-based GNSS retrieval PWV.Precise Point Positioning(PPP)has the advantages of flexible operation,simple model,high positioning accuracy,etc.,and can obtain high-precision ZWD valuation without introducing a reference station,and it has advantages in GNSS/PWV retrieval.However,affected by factors such as measurement error,number of satellites and space configuration,PPP based on a single system often requires a long initialization time to obtain a high-precision ZWD estimate,which will inevitably affect the efficiency of GNSS retrieval of PWV.When GNSS technology is used to obtain the ZWD estimate,the atmospheric water vapor conversion factor becomes the only parameter for GNSS retrieval of PWV.Weighted mean temperature of atmosphere(T_m),as the only variable in the calculation formula of atmospheric water vapor conversion coefficient,directly determines the accuracy of the solution of atmospheric water vapor conversion coefficient.In order to facilitate the acquisition of atmospheric weighted mean temperature,scholars have established a large number of global and regional weighted mean temperature of atmosphere models.Affected by the spatial differences of meteorological data,T_m has a strong geographic correlation,which leads to the poor suitability of the T_m model based on non-localized meteorological data.Based on this,combined with the current development pattern of GNSS multi-mode and multi-frequency,in order to improve the retrieval accuracy and efficiency of PWV,this paper mainly conducts relevant analysis and research around the following issues:(1)The mathematical model of PPP,the linear combined observations commonly used in PPP,the error sources of PPP and their impact characteristics are systematically introduced,and the processing strategies of each error source are analyzed.(2)The principle of ground-based GNSS inversion of PWV and the use of sounding data to calculate atmospheric precipitation is described.Several calculation methods of atmospheric weighted average temperature and the accuracy evaluation indexes of PWV and Tm are given.(3)Aiming at the characteristics of mountain climate in Guizhou,a local weighted mean temperature of atmosphere model suitable for Guizhou was established,and the atmospheric weighted mean temperature and PWV(RS/PWV)calculated from radiosonde data were used as references to analyze and discuss the accuracy of the local weighted mean temperature model.(4)Aiming at the problem of long ZTD convergence time for single-system PPP solutions,a GNSS/PWV retrieval method using multi-system combined PPP solution for ZTD is proposed,and the efficiency of ZTD solution and PWV retrieval accuracy are verified by setting different combination schemes analysis.The experimental results show that the ZTD convergence speed and accuracy of the GPS/BDS/GLONASS/GALILEO(GCRE)multi-system combined PPP are higher than the GPS single-system PPP,GPS/BDS(GC)and GPS/GLONASS(GR)dual-system combined PPP Both have been significantly improved;GC and GR combined PPP PWV retrieval accuracy is better than GPS unit system PPP;GCRE combined PPP further improves PWV retrieval accuracy on the basis of dual system combination.(5)According to the distribution characteristics of GNSS/PWV time series,BP neural network,RBF neural network,wavelet neural network and support vector regression model are used to predict the GNSS/PWV time series before and when the precipitation event occurs.The experimental results show that the RBF neural network model has the highest prediction accuracy and the support vector regression model has the lowest prediction accuracy before the precipitation event and the GNSS/PWV time series show a continuous upward trend.When the precipitation event occurs and the GNSS/PWV time series fluctuates greatly,the prediction accuracy of the RBF neural network model and the wavelet neural network model is equivalent,and the prediction accuracy of the support vector regression model is higher than the BP neural network model.