Research On Optimization Of Tropospheric Delay Model In GNSS Positioning And Timing

With the accuracy improvement of the satellite orbit and clock products,the tropospheric delay has become one of the most important error sources in the dualfrequency and multi-frequency data processing for Global Navigation Satellite System(GNSS).Due to the strong correlation between the tropospheric delay,the station clock,and the station height,the high-precision tropospheric delay model can significantly improve the performance of GNSS positioning and timing.Besides,it can reduce the convergence time in real-time GNSS applications such as Precise Point Positioning(PPP).However,it is difficult to predict the variation and distribution of water vapor in the atmosphere,which resulting in low accuracy of the tropospheric wet delay model.In addition,the accuracy of externally input meteorological values affects the accuracy of the tropospheric hydrostatic delay model.To achieve better performance of GNSS positioning and timing,it is necessary to optimize the tropospheric delay model.This paper focuses on the characteristic analysis,the estimation strategy,and the forecasting algorithm of GNSS tropospheric delay model.In further,the models are verifies in the PPP time transfer,the precipitable water vapor retrieval,and the precise point positioning.The main research contents and conclusions are as follows:1)The research on the correlation between the vertical components in GNSS were sorted out,and the verifications were carried out from the geometric and mathematical perspectives,which laid the theoretical foundation of this article.Then,more comprehensive analysis of the influence of the tropospheric delay estimation strategy in PPP time transfer was completed from the aspects of the priori model,the estimation interval,the relative constraint,and the absolute constraint.The above analysis provides a solid reference to further improve the accuracy of PPP time transfer.2)The widely used Berg atmospheric pressure extrapolation model was refined based on 17 years of data.The modified models improve the accuracy of atmospheric pressure extrapolation by 50-60%,thereby the accuracy of tropospheric hydrostatic delay is improved.The stations height and the precipitable water vapor retrieval results from the PPP verification test show that the modified models have the global applicability.Compared with the huge amount of post multi-layer meteorological data,the modified models trade off the accuracy and the amount of data effectively,which can be applied to real-time scenarios.3)To better study the wet delay for its modeling,the factors that affecting wet delay and the characteristics of GNSS wet delay were analyzed with 11 years data of the IGS second reprocessing products.And the periodic terms related to the sidereal days and the semi-lunar tide were detected.In addition,the precipitable water vapor retrieval products from GNSS,the European Centre for Medium-Range Weather Forecasts(ECMWF)numerical weather data set,the Moderate Resolution Imaging Spectroradiometer(MODIS)on the Earth observation satellite Aqua and Terra,and the Atmospheric infrared sounder(AIRS)on Aqua were evaluated for intra-and interconsistency.The results show that GNSS water vapor retrieval products have good intra-consistency,and the GPS products have good inter-consistency with ERA5 which is the latest generation reanalysis of ECMWF.4)For the GNSS high-precision real-time positioning and timing,a tropospheric zenith delay prediction model was established based on the deep learning method called N-BEATS was applied to training on 18.5 years of data.The kinematic PPP results show our model combined with the GMF mapping function achieve superior performance of positioning and timing compared to the other commonly used models.