In the present study, the Standardized precipitation index (SPI) was employed to analyze the drought status of the Dapoling basin over a period of autumn from September to November, because drought events frequently occur during this period. Three time scales were used, 3-, 6- and 12-month time scale. Daily precipitation data from 13 weather stations covering a period of 31 years from 1980 to 2010 were collected, and the Tyson polygon method was used to calculate the monthly precipitation of the basin. Based on the SPI value, the classification of drought was provided. Besides, considering the fact that the length of sample used to calculate the SPI influences the accuracy on SPI estimation, in turn to lead to the uncertainty of drought classification, the bootstrap technique was employed to analyze the uncertainty of SPI estimation and drought assessment. Results showed that, for September, October or November, drought event mainly occurred in 1985, 1986, 1988, 1990, 1992, 1993, 1994, 1997, 1999, 2001, 2002 and 2007. Especially in 1999 and 2001, severe drought and extreme drought occurred. And the uncertainty analysis results indicated that in term of expected estimation, the two methods with consideration and no-consideration of impact of sample on SPI calculation has no considerable difference, while in term of confidence interval estimation of SPI, there are obviously different between the two methods. This means the impact of the sampling uncertainty on SPI calculation and drought assessment should be noted and not ignored. Copyright Springer Science+Business Media Dordrecht 2014
