The Characteristics Investigation of Ground-Based GPS/PWV During the “7.21” Extreme Rainfall Event in Beijing

GPS/PWV data, combined with automatic station data in Beijing on 20–22 July 2012, the four partitions are analyzed, and the k-means clustering method is employed. The contract results of precipitation characteristics on the basis of four partitions show that (1) the trend of 1 h PWV has the almost same phase, during the rising and falling process corresponding to the start and end of precipitation. (2) The 1 hourly variable PWV is neither increase nor decrease before the precipitation, but 3 hourly variable PWV rises can reach more than 10 mm. The 3 hourly variable PWV is better indication of the intensity of rainfall and the precipitation area, with 3–5 h forecast precipitation for the warning instructions in the practice of forecasting. (3) The PWV of spatial distribution is consistent to the characteristics of precipitation movement from southwest to northeast and gradually increase, and precipitation has a maximum value in 3–4 h before precipitation starts. It has the certain significant instruction to predict future precipitation area and precipitation distribution in practice of forecasting.