Mid-term Prediction of Meteorological Drought Using Fuzzy Inference Systems

Forecasting and monitoring droughts are important elements of optimum water resources management specifically in the metropolitan areas. Tehran as the biggest city of Iran and its five dams (Amirkabir, Lar, Latyan, Mamloo and Taleghan) are also exposed to drought hazards. In the current article, monthly meteorological data in the geographic area covering [0˚, 60˚] Northern latitudes and [0˚, 90˚] Eastern longitudes with 10×10 degree resolution including air temperature and geopotential height at 1000, 850, 700, 500 and 300 mbar levels are used as the model predictors. These data recorded in the period of 1948 to 2008 have been used to develop a model for forecasting SPI (Standardized Precipitation Index) values in Winter and Winter-Spring seasons with 2.5 and 4.5 months leadtime. This model has been calibrated using 31 years of data. Mutual Information (MI) index has been used to select the inputs (predictors) for each basin in each season. Fuzzy Inference System (FIS) has been used to formulate the model. The fuzzy membership functions have been selected based on sensitivity analysis and engineering judgment. The results of the study have shown that geopotential height in 850 and 300 mbar levels are the best predictors for forecasting SPI values in the selected seasons. The model results have had enough accuracy to be used for forecasting SPI values in Winter and Spring seasons in Karaj and Taleghan basins and SPI values in the Winter season in Mamloo, Latyan, and Lar basins.