The Modeling And Research Of China GNSS Tropospheric Delay

In the domain of GNSS,tropospheric delay is one of the main sources of location errors.In short base line observations,difference method can be used to weaken the positioning error that caused by tropospheric delay.However,the correction effect of this method is limited in long base line observations.In practical applications,model correction methods are also used to eliminate tropospheric delay errors,however,the traditional tropospheric delay model is difficult to meet the needs of today's high-precision positioning.Most of traditional tropospheric delay models need to input the measured meteorological data in the calculation process,and the accuracy of calculation often decreases when there is no measured meteorological data.In this paper,the author analyzed the data of tropospheric delay in CORS network and IGS station in China and its surrounding areas,the distribution law of tropospheric delay data,summarized the variation trend,and established a new high-precision tropospheric delay correction model.In order to meet the needs of high precision GNSS positioning,the contents and achievements of this paper are as follows.1.The regularity of tropospheric delay variation at 22 IGS stations in China and its surrounding areas was analyzed,including their annual variation,seasonal variation,longitude and latitude variation and altitude variation.It was found that tropospheric delay has obvious annual variation law,tropospheric delay variation in the corresponding latitudes of the northern and southern hemispheres shows strong negative correlation,and there is a certain inverse exponential relationship between tropospheric delay and tropospheric delay.2.A detailed analysis of the Saastamoinen model of single station bias distribution.The research showed that bias sequence of Saastamoinen model and EEMD filter after each period has a strong positive correlation,and theirs' change in annual cycle in July of each year to reach a maximum value,based on the method of sine function fitting on average bias sequence filter by linear fitting,and obtained on the day of a year for the function of the independent variable.3.In this paper,the Saastamoinen model had been improved,and the precision distribution before and after the improvement is calculated.The average 5-year RMSE and MAE of the original Saastamoinen model was 31.22 mm and 0.96 mm,the corresponding value was reduced to 23.68 mm and 0.72 mm,and the average daily bias sequence of the improved model was close to zero,which indicated that the calculated value of the improved model for tropospheric delay was more stable.The experiment proved that the improved method of Saastamoinen model is stable and effective,and has certain practical significance.4.In this paper,an empirical tropospheric delay model based on EEMD-SARIMA was proposed,the original ZTD sequence was decomposed by EEMD and n IMF sequences and was generated,and the SARIMA parameters were constructed for each IMF sequence.In the courses of the study,the ZTD values of different regions and different seasons in China was forecasted and analyzed,and the applicability of the EEMD-SARIMA model was summarized.The results showed that the hourly average RMSE and MAR of the EEMD-SARIMA model are 9.71 mm and 0.383 respectively.The experimental results showed that the estimation accuracy of the empirical tropospheric delay correction model is high and has certain practical significance.