Abstract: The precipitable water vapor(PWV) are estimated from GPS data of a ground-based GPS network in Chengdu plain with help of the corresponding meteorological data from automatic weather stations.The variations of the derived GPS-PWV are analyzed for the process of convection and stratiform rainfalls.It shows that both in the convection and stratiform rainfall processes,precipitations always happened during the time where water vapor is rather high.In the process of convection rainfall,the local convergence of water vapor is the source of the growth of GPS-PWV.The GPS-PWV increased before the occurrence of precipitation,and condensed after heavy rains,for the value of GPS-PWV is lower than before.If the water vapor accumulated to a high-value afterwards,it usually indicates a new precipitation.During the whole process,fluctuation in GPS-PWV is significantly and frequently,and the intensity of rainfalls is highly correlated with the GPS-PWV.In the process of stratiform rainfalls,the growth of GPS-PWV comes from large-scale water vapor transport.The variation of GPS-PWV is slow comparatively.GPS-PWV remained at high level during participation and until a certain time afterwards,and then reduced gradually.Precipitation is systematic;there is no obvious relation between the intensity of rainfalls and the GPS-PWV.This contribution can optimize the use of the PWV derived from ground-based GPS networks for weather forecasts of different precipitation processes.