Accuracy Analysis And Improvement Of Zenith Tropospheric Delay Model

Tropospheric delays are usually difficult to eliminate completely among the sources of GNSS positioning errors.At present,the commonly used model correction methods are used to correct tropospheric delay.Tropospheric delay models can be roughly divided into two categories: the observed meteorological data model and the non-observed meteorological data model.The essence of these models is the complex non-linear relationship between the observed meteorological data or the estimated meteorological data from the empirical model and the tropospheric delay.With the increasing requirement of GNSS positioning accuracy,especially in elevation direction,the correction accuracy of tropospheric delay model needs to be further improved.In this paper,the accuracy of several tropospheric delay models which are commonly used at present is compared and analyzed by using the reference value of zenith tropospheric delay which is better than 4 mm provided by IGS analysis center.At the same time,the existing models are improved to further improve the accuracy of tropospheric delay estimation.The specific research contents are as follows:?1?The applicability of three observed meteorological data models in some areas of China is analyzed.The experimental results show that the accuracy of Hopfield model in eight IGS stations is low,and the applicability in high altitude areas is poor.The accuracy of Black model and Saastamoinen model is comparable and not affected by the altitude of the stations.?2?A BP neural network model is established at a single station based on different weather data combinations as input data.The experimental results show that the precision of BP neural network model based on radio sounding precipitation is better than that of Saastamoinen model.At the same time,the regional zenith tropospheric delayed interpolation model is constructed by using this model,and the average of the selected stations is obtained.The absolute deviation is 1² cm,which is better than Saastamoinen model.?3?The accuracy of four tropospheric delay models without observed meteorological data is compared and analyzed.The results show that the accuracy of GT2w₁ model issecond to that of GPT2w₅ model,while the accuracy of EGNOS model and UNB3 m model is relatively poor.?4?The ARIMA model is established for a single IGS station,and the model error of GPT2w₁ model is compensated.Experiments show that the precision of GPT 2w₁ model improved by ARIMA model is obviously higher than that of the original model.