Abstract:A spatio-temporal ZTD modeling method based on principal component analysis (PCA) is proposed. Based on this modeling method, three regional spatio-temporal ZTD models of Hong Kong, Yunnan province and China mainland are established using daily ZTD observation data obtained from the GNSS continuous operation reference stations. The test results show that the accuracy of the regional spatio-temporal ZTD models is significantly higher than that of the Saastamoinen, EGNOS and UNB3m models. Moreover, the proposed PCA spatio-temporal ZTD modeling method has advantages such as simple modeling process, less model parameters and greater convenience.