Research On Performance Of Multi-GNSS PPP Assisted By ERA5

Tropospheric delay is an important error source for GNSS high-precision positioning,especially for precise single point positioning.Relevant studies show that the tropospheric delay not only has spatial difference,but also has temporal variation characteristics such as annual cycle,seasonal cycle and daily cycle.In order to improve the performance of GNSS positioning,scholars have successively constructed a large number of global and local tropospheric delay empirical models.All these regional or global tropospheric delay models have their own advantages.However,due to the limitations of the temporal resolution of the meteorological data used for modeling,most of the existing models cannot well describe the variation trend of tropospheric delay on small time scales.In addition,most of the existing models use quadratic polynomial and exponential model to represent the variation trend of tropospheric delay in the direction of elevation.However,the actual variation of tropospheric delay is a very complex physical process,and if an unreasonable model is adopted,additional human error will be introduced.The atmospheric reanalysis product ERA5 released by the European Centre for Medium-Range Weather Forecasts has been gradually applied to tropospheric delay modeling in recent years due to its high accuracy,high spatial resolution and high temporal resolution.However,in order to better describe the variation characteristics of tropospheric delay in the elevation and time domains,if the tropospheric delay model is constructed by using ERA5 meteorological data of long time and different isobars,it will bring a great computational burden to the computer.It is bound to reduce the efficiency of model construction.Compared with most machine learning algorithms,Light GBM has the advantages of high precision,high training speed and high efficiency.Therefore,since the grid model is better than the local model and the global model,in order to improve the accuracy of the tropospheric delay model,this paper uses the ERA5 reanalysis meteorological product with a spatial resolution of0.25°×0.25° and a temporal resolution of 1h covering the Guizhou region.As the data source,a tropospheric delay grid model is constructed based on Light GBM to provide a more reliable initial value of tropospheric delay for GNSS precision single-point positioning.To sum up,the main work and achievements of this paper are as follows:(1)The principle and error source of precise single-point positioning are summarized,and the main functional modules and characteristics of RTKLIB are introduced.(2)On the basis of introducing the generation mechanism of tropospheric delay and several classic tropospheric delay models,the ERA5 grid data of 5 IGS stations in China for the whole year of 2021 is used to estimate the tropospheric delay,and provided by the IGS Analysis Center For reference,the tropospheric delay product of ERA5 is analyzed and discussed for the applicability of the ERA5 data in the Chinese region.The experimental results show that,compared with GPT2w-1S and GPT3-1S,the tropospheric delay obtained by the ERA5-S model has smaller deviation and root mean square error,which are-11.11 mm and 23.08 mm,respectively,indicating that the ERA5 meteorological product has better performance.high precision.(3)Using the ERA5 reanalysis meteorological data from 2011 to 2019 as the data source,based on the Light GBM machine learning algorithm,a grid tropospheric delay model(EGRID model)that takes into account annual,seasonal,daily and elevation changes is constructed.The experimental results show that the elevation is the main factor affecting the troposphere,followed by the annual cumulative day(doy),the annual trend(year)and the daily cumulative time(hod);For reference,the ZHD model built by the three grid points in this paper has obtained high modeling accuracy and prediction accuracy.The root mean square error and mean absolute error of the model prediction values are both millimeters,and the coefficient of determination R2 is close to 1;although affected by ZWD Due to the influence of fluctuations,the ZTD obtained by the EGRID model also achieved high modeling accuracy and prediction accuracy,and the modeling determination coefficient R2 was better than 0.98.(4)Aiming at the difficulty of obtaining measured meteorological data,a PPP positioning method based on EGRID tropospheric delay assistance is proposed.The experimental results show that,compared with the ESTG model,the GPT2w-1S model,the UNB3 m model and the GGOS model,the precise single-point positioning based on the tropospheric delay-assisted EGRID model can significantly improve the positioning accuracy and convergence speed of the PPP,and the static and dynamic point positioning accuracy is improved respectively.1.9%,67.9%,61.2%,61.2% and 4.9%,4.6%,5.5%,8.0%.