Refining empirical tropospheric model with meteorological stations for large height difference RTK positioning

In real-time kinematic (RTK) positioning in areas with larger height differences (over 100 m), omitting tropospheric delay is insufficient and can seriously affect positioning accuracy. This issue is known as a large height difference RTK. External correction can be employed to eliminate these residual errors. However, empirical tropospheric models, commonly used in Global Navigation Satellite System (GNSS) have limited performance due to systematic biases that are linearly related to height. Fortunately, the prior model can be refined if a more accurate tropospheric delay is available, such as one derived from observed meteorological parameters. This study proposes a height-related correction model to improve the Saastamoinen model with standard atmospheric parameters, a simple tropospheric model commonly used in GNSS. The proposed model is based on vertically distributed surface meteorological observations. We compare the performance of three strategies, namely, the Saastamoinen model with standard atmospheric parameters (STM), the Saastamoinen model with meteorological observation (MET), and the refined model (REM), using GNSS and meteorological data from the Wudongde Location Service Experiment Field. The results show that the RTK positioning accuracy of REM and MET performs 35.8% and 32.4% better in winter and 14.1 and 4.6% in summer than STM. The performance of the refined model is primarily influenced by meteorological observation errors and inhomogeneous tropospheric delay, which are affected by seasonal atmospheric variations. Despite being less effective in summer than in winter, the model provides a new approach to realizing GNSS differential services utilizing meteorological stations.