Abstract: It is very important to research the inconsistency (or nonstationarity) of hydrological time series under changing circumstance. Classical regression models, which are widely used in traditional hydrological trend analysis, are based on the assumption that the hydrological variables follow the normal distribution. However, this assumption does not consist with the practical findings in China that most hydrological random variables follow the Pearson type III (PⅢ) distribution. In this paper, the regression model based on the assumption of the PⅢ distribution is introduced to analyze the trend of annual minimum monthly runoff series of the Yangtze River at Hankou hydrological station observed over the period of 1952-2010. It is found that the regression model based on the PⅢ distribution is more effective than the regression model based on the normal distribution in fitting the trend of hydrological series. Furthermore, the more flexible polynomial is used to analyze the trend of the hydrological series.

Key words: trend analysis, regression model, normal distribution, PⅢ distribution, annual minimum monthly runoff