عنوان مقاله [English]
نویسندگان [English]چکیده [English]
Drought forecasting plays an important role in the planning and management of natural and water resources. In this study, multiplicative seasonal autoregressive integrated moving average (SARIMA) models as the linear stochastic models were used to forecast droughts. The models were applied to forecast droughts using standardized precipitation index (SPI) time series. The SPI values for time scales of 3, 6, 12 and 24 months for Abadeh, Shiraz and Fasa synoptic stations in Fars province were calculated. The SARIMA model with the minimum of Akaike's information criterion bias corrected (AICc) was selected as the best model. The auto-correlation function plots of the residuals for the selected models indicated that the residuals were uncorrelated. The SPI values from January 2004 to December 2005 as the test data have forecasted using fitted models. For example, model was identified for 12 months SPI time series in Shiraz station. The results of this model indicated that the correlation coefficient between the observed and predicted values of SPI is 0.74 which is significant at 1 % level and 22 months of drought category was correctly forecasted.