Abstract: Atmospheric weighted mean temperature (T_m) participation is required for ground-based GNSS water vapor inversion, and saturated water vapor (E_s) is an important variable in the calculation process of T_m that effects T_m, so e_swill be indirectly affect the inversion accuracy of Precipitable Water Vapor (PWV). In view of the three saturated water vapor pressure models (Magnus-Tetens model, BUCK model, Goff-Gratch model) established by different researchers commonly used in the research of ground-based GNSS water vapor inversion, this paper will research different saturated water vapor pressure models participate in the inversion Whether cause differences in results. Taking Hong Kong as the research area, using GAMIT to solve the Zenith Wet Delay (ZWD) of the dry and rainy season (February and July) in 2016, meanwhile using the sounding data of the King's park sounding station to calculate The T_m of the dry and rainy seasons (February and July) through the way of integrate numerical, and then calculating the PWV of the dry and rainy seasons (February and July) through programing with reference to the inversion steps.Thorough comparing and analyzing to get the effects and reasons of different saturated water vapor pressure models participating in the calculation on T_m and PWV. The results show that between the PWV calculated by the three saturated water vapor pressure models and the true value (PWV calculated by the sounding station) have no statistically significant differences, so all of them can be used to provide the saturated water vapor pressure e_s in the calculation of T_m, However, through comparative analysis, it is found that some researchers use the variable T in the BUCK model as the dew point temperature instead of the atmospheric temperature to make T_m produce a larger error. The analysis in this paper will provide a certain reference for the treatment of T_m in the subsequent research of ground-based GNSS water vapor inversion.