Abstract: The precipitable water vapor (P_WV) data are retrieved from the ground-based Global Positioning System (GPS) observational network together with the automatic weather stations in Chengdu region during the period of August 21, 2007 to November 30, 2008. The GPS-retrieved P_WV data are analyzed for diurnal cycle characteristics as well as for different precipitation types including the summer heavy rainfall, the autumn continuous rain and the winter snow-rainfall. The result shows that storms occurred in different seasons correspond the distinct P_WV. For example, the summer storms are associated with the change of P_WV values from peaks to troughs. The opposite P_WV change characteristic to the summer heavy rainfall is found at the end of summers. The result suggests that the change ratio of P_WV values before a rainfall event may well indicate the intensity of subsequent heavy rainfalls. Significant variations in P_WV ratios during the autumn continuous rain can potentially predicate the beginning or the ending of a rainfall event. The real-time consecutive GPS-retrieved P_WV data can thus be used in operational precipitation forecasting.