An Improved Method of ZTD Model in Yunnan Province Based on GPT2w Model

Abstract

The GPT2w model is one of the currently disclosed high-precision empirical models of tropospheric key parameters. The model can provide a variety of meteorological parameters, and the ZTD can be calculated using the model parameters. However, when using the GPT2w model to calculate ZTD directly, its accuracy is reduced, and there is a specific regional. In order to obtain higher precision ZTD parameters, this paper proposes a ZTD model that improved GPT2w(IGPT2w). The model uses the ZTD calculated by the GPT2w as the initial value of the IGPT2w and calculates the ZTD residual for the ZTD actual value and the GPT2w. The difference was fitted to obtain a ZTD residual fitting model. Finally, the IGPT2w model consists of the initial value calculated by the GPT2w and the residual estimated by ZTD residual model. Taking Yunnan, China, as an example, this paper selects 27 GNSS stations in Yunnan from 2015–2017 to establish the IGPT2w model and verify its accuracy. Firstly, MSSA and LS methods are used to analyze the characteristics of the ZTD residual and determine the periodic signals and establish a ZTD residual model considering annual, semiannual, and seasonal period, and then establish the IGPT2w in combination with the GPT2w. The IGPT2w model was established based on GNSS data from 2015–2016 and applied to ZTD estimation in 2017; The result shows that the IGPT2w estimates that the external RMS and MAE (2.4/1.8 cm) of the ZTD are lower than the GPT2w (2.8/2.3 cm). Compared with the GPT2w, the average accuracy of ZTD was improved by 13.4%. The above results show that the IGPT2w model is superior to the GPT2w model in the Yunnan province of China.