In recent decades, the climate change phenomenon has caused changes in the hydrological cycle of the earth. Therefore, studying the effects of this phenomenon on different parts of the hydrological cycle and predicting its effects on occurrence of extreme phenomena such as drought is necessary. In the present study, the performance of 24 GCM models was analyzed using a multivariable statistical approach including one-variable mean, standard deviation, coefficient of variation, Mann-Kendall test, Kolmogorov-Smirnov test, and multivariate test of the main component analysis and the superior models were selected for the study area. Then, the large-scale monthly time series of precipitation and temperature were extracted for the future period of 2020-2049 from the selected models under three scenarios RCP2.6, RCP4.5, and RCP8.5. The precipitation change scenarios and the temperature change are introduced into the LARS-WG downscaling model after the interpolation to the meteorological stations coordinates. The daily time series of precipitation and temperature are obtained for the future period of the considered stations. The results indicate that rainfall will not change significantly, and the temperature of the whole basin will increase by an average of about 1 °C. The simulated precipitation and temperature values are considered as inputs of the IHACRES hydrologic model. Then, the monthly time series of discharge are simulated for the future. The simulated runoff values represent a 15% reduction in average annual runoff and the highest runoff variations were reported in April. In the following, the SPI and SRI index, respectively for meteorological droughts and hydrological droughts were calculated using simulated precipitation and discharge time series for each model and scenario. The Bayesian approach is used to model uncertainty in drought simulation. A binomial distribution is considered for each Keywords : Climate change, Multi-criteria approach, Drought, Uncertainty, Bayesian approach
