theTime Series Prediction of Meteorological Parameters in the Arid and Semi-Arid Region

Time series prediction of meteorological parameters plays an important role in making decision to decrease the effects of drought and climate change. Temperature is an important factor in planning and making decision of water resources and water balancing. Therefore, precise estimation of temperature is indispensable for every computation in hydrology and other disciplines. There are a lot of methods for estimating time series climate data and all of them can be grouped in (A) statistical methods and (B) intelligent methods. In this study, Artificial Neural Network (ANN) and ANFIS were applied as intelligent methods to estimate the maximum and minimum temperature. The data was spited into two parts (A) 90% data that was used as training and (B) the rest of the data set which was applied as the test set to validate the constructed model. The performance of Multi Layer Perceptron and Neuro-Fuzzy Inference System with the fuzzy c-means clustering (FCM-ANFIS) were investigated using different numbers of neurons in hidden layers and different number of clustering , respectively. Accuracies of the models were evaluated using indices such as R2, RMSE and MAE.