A predictive model for regional zenith tropospheric delay correction

Abstract

Aiming at the problem that the conventional zenith tropospheric delay (ZTD) model does not consider the seasonal variations of delay and has low accuracy and stability, a method is proposed to introduce annual and semi-annual variations into the Saastamoinen model. This method uses the ZTD grid products from 2015 to 2017 provided by the Global Geodetic Observing System (GGOS) to analyze the seasonal variations of the bias of the Saastamoinen model over Asia, and then constructs the Saastamoinen model with seasonal variation corrections, denoted as SSA, which is applicable to the Asian region. In order to overcome the dependence of the model on in-suit meteorological parameters, the combined SSA+GPT3 model is formed by combining the SSA and GPT3 models. Through the high-precision meteorological parameters provided by GPT3 model, the predictive correction of ZTD over Asia is realized. The experimental comparison results show that the introduction of annual and semi-annual variations can substantially improve the Saastamoinen model. The variations of bias and RMS of the SSA are small and stable with time. In summer and autumn, the bias and RMS are noticeably smaller those from the Saastamoinen model. In addition, It is found that the SSA model performs better in low latitude and low altitude areas, and bias and RMS decease with the increase of latitude or elevation. Using the ZTD data of 66 Asian stations released by the international GNSS service (IGS) in 2018, the prediction accuracy of the SSA model is evaluated for external consistency. The statistical results show that the prediction accuracy of the SSA model (bias: −0.38cm, RMS: 4.43cm) is better than that of the Saastamoinen model (bias: 1.45cm, RMS: 5.16cm). The proposed method has strong applicability and therefore can be used for predictive ZTD correction in navigation and positioning over Asia.