Drought forecasting and monitoring play a significant role in reducing the negative effects of global meteorological droughts caused by different intensities at different temporal and spatial scales in different regions, especially in regions with high dependency on rainwater. The present study tries to compare the accuracy of stationary time series (ST) models including autoregressive moving average (ARMA), moving average (MA) and autoregressive (AR) and cyclostationary time series (CT) models including periodic autoregressive moving average (PARMA), periodic moving average (PMA) and periodic autoregressive (PAR) to predict drought index (i.e. monthly reconnaissance drought index (RDI)) in periodic data series considering that CT models are more powerful and efficient than ST models by using data series of 8 synoptic stations with different climate conditions in Iran from 1967 to 2017. According to the results the monthly RDI was significantly periodic in all selected stations. The PAR (25) model was the best fitted CT model in data series at all stations and on the other hand, the following models were the best-fitted ST models in data series: the AR models at Babolsar and Rasht AR (25) and at Gorgan AR (24) and ARMA models at Tehran ARMA (2, 3), at Zahedan and Shiraz ARMA (2, 4) and at Esfahan and Shahre Kord ARMA (2, 5). Based on the best fitted CT and ST models, the results showed that the correlation coefficients (R) between observed and simulated RDI vary from 0.882 to 0.946 and from 0.693 to 0.874, respectively from January 1967 to December 2017. According to the best fitted CT and ST models, the validation test of the best fitted models indicated that the R between observed and simulated RDI vary from 0.634 to 0.883 and 0.585 to 0.847, respectively from January 2012 to December 2017. In total, it can be concluded that that the accuracy and capability of CT models in predicting the RDI were more than those of the ST models at all stations and the hypothesis of the study was confirmed.

干旱的预测和监测在减少不同地区在不同时空尺度上不同强度造成的全球气象干旱的负面影响方面发挥着重要作用，特别是在高度依赖雨水的地区。本研究试图比较包括自回归移动平均值（ARMA），移动平均值（MA）和自回归（AR）的平稳时间序列（ST）模型以及包括周期性自回归移动平均值（PARMA）的循环平稳时间序列（CT）模型的准确性，周期性移动平均值（PMA）和周期性自回归（PAR）来预测干旱指数（即 考虑到使用1967年至2017年伊朗不同气候条件的8个天气观测站的数据系列，CT模型比ST模型更强大和高效，因此定期数据系列中的每月侦察干旱指数（RDI）。根据结果，每月RDI在所有选定的电台中都有明显的周期性。PAR（25）模型是所有站数据系列中最拟合的CT模型，另一方面，以下模型是数据系列中最拟合的ST模型：Babolsar和Rasht AR（25）的AR模型和在德黑兰ARMA（2、3），Zahedan和Shiraz ARMA（2、4）以及Esfahan和Shahre Kord ARMA（2、5）的Gorgan AR（24）和ARMA模型上。基于最佳拟合的CT和ST模型，结果表明，观察到的RDI与模拟RDI之间的相关系数（R）在0.882至0之间变化。