Autumn rainfall forecasting using ENSO indices by Neural Network method

The evolution of rainfall is due to climate signals changes. One of the climate signals is ENSO. This signal is one of the important signals that cause large scale climate anomaly change in many area of the world. This signal can increase rainfall in many areas and also decrease rainfall in other area in the same period. One of the rainfall forecasting methods is using climate signals. To forecasting seasonal rainfall in Urmiyeh Lake basin، relationship between autumn rainfall and climatic indices were investigated. The climatic indices are the indices that scientists use in the word. There are many rainfall stations in Urmiyeh basin، from those 18 stations with more than 35 years data were selected. In order to forecasting the SPI، data of two stations، Tabriz and Urmiyeh were used. SOI and NINO3،4 indices used as ENSO indices. This result show the impact of ENSO is more than other climatic signals on autumn rainfall. To forecast the Autumn rainfall one linear model and one nonlinear model used. Using R2 and MSE the model evaluated. The result show the nonlinear model (Neural Network) can forecast Autumn rainfall with higher accuracy than the linear model. Therefore using teleconnection pattern data and neural network as powerful tools، autumn rainfall can be forecasted one season earlier in Urmiyeh leak basin.