Application of Artificial Neural Network in Climatic Elements Simulation and Drought Cycle Predication (Case Study: Isfahan Province)



  Abstract   In this research, Artificial Neural Networks (ANNs) were used as strong tool in simulation of nonlinear processes to predict drought cycle in twenty synoptic, climatic and hydrometric stations in Isfahan province. where had daily statistics for twenty years. Neural network of MATLAB-7 was used for predicting and analyzing climatic elements. Input of ANN models including: monthly rainfall mean, minimum yield and maximum temperature which were related to the period between "1984-2004". Twenty year of this period was devoted for training and the remainder four years were spent on testing. The used network was Multi-Layer Preceptron(MLP) with Back Propagation Logarithm(BP) and Levenberg-Marquardt technique(LM). Different structures of neural network were created by changing input layers (6 models), the number of tines in hidden layers and output layers (2-20). The results show that among the analyzed patterns max temperature, yield and rainfall have predict significant role to drought in Isfahan and by application of ANN can be predicted the drought cycle by the confidence interval of %95.