Confronting drought and reducing its impacts requires modeling and forecasting of this phenomenon. In this research, the ability of different time series models (the ARIMA models with different structures) were evaluated to model and predict seasonal drought based on the RDI drought index in the south of Iran. For this purpose, the climatic data of 16 synoptic stations from 1980 to 2010 were used. Evaluation of time series models was based on trial and error. Results showed drought classes varied between ‘very wet’ to ‘severely dry’. The more occurrence frequency of ‘severely dry’ class compared to other drought classes represent the necessity of drought assessment and the importance of managing the effects of this phenomenon in the study area. Results showed that the highest severity of drought occurred at Abadeh, Shiraz, Fasa, Sirjan, Kerman, Shahre Babak and Saravan stations. According to selecting the best model fitted to the computed three-month RDI time series, results indicated that the MA model based on the Innovations method resulted in maximum cases with the best performance (37.5% of cases). The AR model based on the Yule–Walker method resulted in minimum cases with the best performance (6.3% of cases) in seasonal drought forecasting.

面对干旱并减少其影响，需要对该现象进行建模和预测。在这项研究中，评估了基于伊朗南部的RDI干旱指数，不同时间序列模型（具有不同结构的ARIMA模型）建模和预测季节性干旱的能力。为此，使用了1980年至2010年的16个天气观测站的气候数据。时间序列模型的评估基于反复试验。结果表明，干旱等级在“非常潮湿”到“非常干燥”之间变化。与其他干旱类别相比，“严重干旱”类别的出现频率更高，这说明了进行干旱评估的必要性以及在研究区域中应对这种现象的影响的重要性。结果表明，干旱的严重程度最高的地区是Abadeh，Shiraz，Fasa，Sirjan，Kerman，Shahre Babak和Saravan车站。根据选择适合计算的三个月RDI时间序列的最佳模型，结果表明，基于Innovations方法的MA模型导致了最大的案例，具有最佳的性能（案例的37.5％）。基于Yule-Walker方法的AR模型在季节性干旱预报中表现出最少的案例，表现最佳（占案例的6.3％）。