Drought forecasting by SPI and EDI indices using ANFIS method based on C-mean and SC clustering (Case study: Kohgiluyeh and Boyer Ahmad Province)

Drought is one of the most and oldest natural disaster that cause significant environmental impacts. Despite Kohgilouyeh and Boyerahmad is in the third place in terms of rainfall but the drought has affect the province intermittently and causes many heavy losses. In other to drought crisis management, finding the index measurement of the drought to predict and evaluate the spatial and temporal of this phenomenon, seems essential. In this research, using Artificial Neural Networks (ANN) and Adaptive Neuro Fuzzy System (ANFIS) model with phase clustering analysis with standardized precipitation index (SPI) and effective drought index (EDI) were used to predict drought. The results of study indicate that the SPI index by validation coefficient 0.87 has more capability and accuracy than EDI index by validation coefficient 0.73 in predicting of drought. On the other hand according to C-mean and SC clustering in modeling for predicting the drought, ANFIS approach has more efficacy the result show that, clustering causes the increasing of model accuracy in verification and calibration stages. C-mean clustering by calibration coefficient 0.93 and validation coefficient 0.87 is the best model.