Abstract:In view of the complexity and poor performance during the real-time conversion process GNSS precipitable water vapor (PWV) in the southeast coastal area of China, based on the data of 18 GNSS stations in the area from 2017 to 2018, we analyze the linear relationship between GNSS-PWV and zenith tropospheric delay (ZTD), ground temperature (Ts) and ground atmospheric pressure (Ps). We apply the multiple linear fitting method to establish the direct conversion model of PWV,which provides a simple and effective method for predicting GNSS-PWV in the area. Experimental results show that GNSS-PWV has good correlation with ZTD, Ps and Ts, their related coefficients are 0.98, -0.65 and 0.78. The multi-factor GNSS-PWV model based on ZTD, Ps and Ts has the highest precision, and its RMS is 0.33 mm. It is much better than the single-factor PWV model based on the ZTD (RMS=4.66 mm). For the double-factor GNSS-PWV model based on ZTD, Ps has the second highest precision, and its RMS is 0.50 mm.