Abstract: The weighted mean temperature (WMT) is an important parameter to calculate precipitable water vapor in ground based GNSS meteorology. Four local linear statistical models of the WMT were derived from their historical radiosonde data in 2019. The comparison with other WMT models and the precipitable water vapor calculated by them show that Mao model and Mendes model have higher precision in the four cities, and it is necessary to establish local model for high-precision applications. The findings provide valuable reference for engineering fields concerned with the weighted mean temperature.