Abstract: Atmospheric weighted mean temperature is a key parameter in CORS water vapor retrieval, and its variations are obvious in both time and space. In order to improve the accuracy and reliability of the atmospheric CORS, the surface temperature Ts and the weighted mean temperature Tm acquired by radiosonde station during 2009 to 2011 in Qingdao are analyzed. The correlation coefficient R of Ts and Tm is 0.877 6, which is strong linear correlation. The weighted mean temperature model of Qingdao area is established by regression analysis, and the weighted mean temperature in 2012, Qingdao is calculated. Compared to the weighted mean temperature calculated by radiosonde data, the average deviation, standard deviation and root mean square error are respectively 0.307 K, 3.359 K and 3.384 K. The model is applied to CORS water vapor retrieval in Qingdao, and the average deviation, standard deviation and root mean square error are respectively 0.70 mm、3.48 mm and 3.53 mm. The results show that it is feasible to apply the regional radiosonde data to establish the weighted mean temperature model and to improve the accuracy and reliability of the regional CORS water vapor retrieval to a certain extent.