Predicting Drought Using Fuzzy Logic and Decision Tree: A Case Study of Mazandaran and Golestan Provinces, Iran

The main goal of this study is to predict drought of Mazandaran and Golestan provinces using Fuzzy Logic. In this study, the annual rainfall data for 12 stations in Mazandaran and Golestan provinces for a period of 3 years (1971-2001) were analyzed. The sufficiency and accuracy of available data and scattered data were evaluated. The drought indices used in this study were 1.Standard Index of Annual Precipitation (SIAP) 2.Rainfall Anomaly Index (RAI). According to the available indices, three classes (weak drought, medium drought and intense drought) were taken into consideration. The feature of rainfall values were extracted using descriptive statistical parameters. J48 algorithm is used as a feature selection procedure to choose pertinent feature from data set. The results of J48 algorithm was used to generate rules and membership function set. Finally, in order to evaluate the proposed J48-Fuzzy model, the data sets were used to test the model. The results showed that the total classification accuracy for SIAP and RAI indices were 72.22 %, 86.66%, respectively.