Farklı troposfer modellerinin gerçek zamanlı çoklu-GNSS PPP performansına etkisi

Abstract

With the initialization of IGS (International GNSS Service) real-time service products, real-time Precise Point Positioning (PPP) applications have been a popular topic within the GNSS (Global Navigation Satellite Systems) users. The impact of the troposphere on GNSS signals is one of the most crucial error sources regarding the real-time PPP solution. In the PPP technique, the dry component of tropospheric delay is usually corrected by means of empirical models, while its wet component is estimated as an unknown parameter in the adjustment process. Hence, the troposphere model employed in the PPP solution has a considerable impact on the performance of the obtained solution. Therefore, the main objective of this study is to investigate the impact of different troposphere models on the performance of real-time multi-GNSS PPP. As a part of this study, four different troposphere models, where Saastamoinen and Hopfield models that are most frequently used in GNSS solutions are used together with GPT (Global Pressure and Temperature) 2 and 3 models separately, were constituted. In this context, the observation dataset acquired from a total of 16 different IGS stations over a ten-day period of December 19-28, 2021, were processed utilizing four different troposphere models. In addition to the positioning performance, PPP solutions were also analyzed in terms of zenith total delay (ZTD) estimation. The results show that the best positioning performance can be obtained when the Saastamoinen model is used in combination with GPT3. The three-dimensional positioning accuracy acquired from the corresponding solution is 2.72 cm, which is better than the closest solution by a ratio of 9.2%. Besides, in terms of the ZTD estimation, the best performance is achieved again in the case when the Saastamoinen model is used in combination with GPT3. For the related solution, the accuracy of ZTD estimation is calculated as 1.24 cm and this value indicates a better performance by a ratio of 10.2% compared with the closest solution.