Application of M5 Tree Model in Forecasting Drought (Case Study, Maragheh, Iran)

Accurate prediction of droughts in arid and semi-arid countries, like Iran, have important role in water resources management and designing appropriate plans for coping with drought consequences. Since the standardized precipitation index (SPI) is known as a suitable index for drought analysis, in this study, we used the M5 rule tree model for forecasting SPI values. For this purpose, the monthly precipitation data of Maragheh synoptic station were used during a 25-year period for calculating SPI values at 6-month time scale (SPI-6). The results indicated that the Maragheh region was faced with successive and severe droughts in recent two decays. In the next step, the SPI-6 values were forecasted for next 1 to 12 months using M5 rule tree model. The results showed that the SPI-6 values in previous time steps had the most effect on forecasting the next SPI-6 values, and the forecasting accuracy decreases with increasing prediction length. So the correlation coefficient of forecasting SPI-6 for next month was obtained 0.94 which this value was decreased to about 0.40 for forecasting SPI-6 for next 12 months. However, the M5 rule tree model provides more understandable, applicable and simple linear relation in forecasting droughts and shows relatively good performance and accuracy.