Abstract: Along with the development of automatic measurement and forecast technology, more and more automatic rainfall stations are put into operation. Analyzing the quality of data measured by automatic rainfall station is of great significance to the operation of water project. The analysis on the hourly rainfall series from telemetry rainfall stations in the lower reaches of the Jinsha River from 2008 to 2015 was carried out by using the Grubbs criterion and k-medoids clustering algorithm. Based on the statistical theory, the results showed that the methods have a remarkable effect in determining the abnormal value of annual accumulated rainfall and the singular points of hourly rainfall data. The method provides some references for hydrological data quality analysis that is changed by increasingly fine particles.