The application of Artificial Neural Networks in the rainfall prediction

The aim of this research is the rainfall prediction of Khorasan Razavi province using artificial neural network. At the first step, the time series of average regional rainfall using Kriging method in the desired time period was calculated. In the next step, time series of climatic predictors including Sea Level Pressure (SLP), Sea Surface Temperature (SST), Sea Surface Pressure gradient (DSLP), the difference between sea surface temperature and 1000 hpa level, Sea Surface Temperature gradient (DSST), Air Temperature At 700 hpa, thickness between 500 and 1000 hpa level, Relative Humidity at 300 hpa, outgoing long wave radiation, Precipitable water, meridional wind and zonal wind in the different time steps were obtained. In continue the relation between average regional rainfall and climatic predictors using Pearson’s correlation coefficient were calculated. After identifying of the appropriate predictors, artificial neural network model was calibrated from 1970 to 1997. Finally, model was tested in the period between 1998-2007. The model that used in this research has an input layer, one hidden Layer and an output layer. Results reveal those artificial neural networks are promising and efficient. Root mean square error for this model was obtained 6.9 millimeters.