In order to predict water requirement in the medium-to-long period,we studied dynamics features of refer-ence crop evapotranspiration.Multifractal analysis was applied to the time series of reference crop evapotranspiration of three stations (Zhongxiang,Tianmen and Wuhan),located in Hanjiang basin from 1978 to 2007.Results show that,the daily reference crop evapotranspiration contains characteristics of irregular high-frequency fluctuation and exhibits the strongest multifractal characteristics among different time intervals such as one day,ten days and one month.Further analysis by partition function shows that,most part of multifractality in the time seriesdata was due to correlations caused by fluctuations and the fat-tailed probability distributions caused by extreme events.Moreo-ver,the multifractal features vary within the different parts of the time series and the changes can be explained by the methods of piecewise linear fitting and trend changing points (PLFIM).Finally,both of multifractal spetrum and local piecewise width are influenced by extreme climate events in different time intervals,and the effects of extreme climate events are relevant to different locations in basin.