Prediction of Monthly Rainfall for Selected Meteorological Stations in Iraq using Back Propagation Algorithms

Artificial neural network was used for predicting monthly mean rainfall. In order to train the neural network back propagation algorithms had been employed. Rainfall data along the years (1970-2000) measured in four cities (Mosul, Baghdad, Rutba and Basra) in Iraq were used as training and ten years (2001-2010) used for testing. The logistic sigmoid activation function was used for both hidden and output layers. To estimate difference between measured and estimated rainfall values, Mean Absolute Error (MAE), Root Mean Square Error (RMSE) and correlation coefficient (R) were determined. According to three statistical indices were calculated to examine the performance of the optimum ANN model, It was found that the optimum model according three among the four considered statistical indices was in Rutba station during December month where the correlation coefficient (R), Root Mean Square Error (RMSE), Mean Absolute Error (MAE) and Mean Bias Error (MBE) was 0.9998, 0.59, 0.56, -0.56 mm, respectively, these statistical results have shown the ability of the artificial neural network for rainfall prediction.