Assessment of GPT2 Empirical Troposphere Model and Application Analysis in Precise Point Positioning

Abstract

Precise Point Positioning (PPP) has been demonstrated to be a powerful tool in geodetic applications, such as deformation monitoring. Troposphere delay is an important error source which directly affects positioning accuracy in height direction. At the end of 2012, an improved model named Global Pressure and Temperature 2 (GPT2) was proposed. Compared with early empirical models, this new model mainly eliminates the weakness of limited spatial and temporal variability. In this study, we assess the precision of GPT2 model and apply it in PPP analysis. The analysis data of VMF1, which is produced using European Centre for Medium-Range Weather Forecasts (ECMWF), provides the nearly true value of zenith delays and mapping function for International GNSS Service (IGS) stations. Therefore a globally distributed set of 11 IGS stations is chosen to validate GPT2 model. In the case of using GPT2 as a priori model while the residual zenith delay still estimated with other unknown parameters, it would improve the zenith troposphere delay adjustments to nearly zero-mean. Therefore GPT2 is helpful to improve the efficiency in PPP data processing. However, GPT2 model is only resting upon a global 5° grid and sufficient global troposphere models are not yet available. Due to the complexity of wet zenith delay, PPP height solutions would be unsatisfactory when residual troposphere delay is not parameterized. We conclude that GPT2 is capable of predicting troposphere delay worldwide with an acceptable uncertainty. And GPT2-based PPP solution performs well only in the case of regarding residual model error as an additional unknown parameter.