Abstract
Drought monitoring by the Standardized Runoff Index (SRI) presents some uncertainties, mainly dependent on the choice of the probability distribution used to describe the cumulative precipitation and on the characteristics of the dataset. In this study, uncertainty analysis for estimation of the hydrologic drought characteristics (intensity, duration and frequency) was performed. Four distribution functions, two time period (30 and 49 years), six time scales (3, 6, 9, 12, 24 and 48 months) and Latin hyper cube sampling (LHS) method ware used. For each event at per year and month, was generated 50000 random sampling.Then, lower and upper bands of certainty was calculated for confidence level of 95% . In addition to the drought characteristics (intensity, duration and frequency) were calculated for six time scales, four distribution functions and two length of time series . Investigation of the longest duration and highest intensities showed that an increase time scale led to decrease the frequencies of drought classes and as a result increase drought intensity and duration . Further, no significant difference in the assessment of intensity and duration was between various distribution functions, meanwhile significant difference was between normal compared to weibull and gamma for the estimation of drought frequency in short time scales (3 and 6 months). Results of this study emphasized that considering drought intensity and duration, the normal distribution function, 24-month time scale and 30-years’ time series had the largest uncertainty for hydrologic drought estimation.
)