Long- range precipitation prediction using artificial neural networks

In this paper, the effects of large scales climate signals on the low and high precipitation Spells in the southwestern part of Iran are investigated. Large scales climate signals are parameters that can play the important role on analysis variations of seasonal and annual precipitation. In this study monthly southern oscillation index (SOI), North Atlantic Oscillation (NAO) and ENSO index were applied in NINO4, NINO3, NINO1+2, NINo3.4 were used respectively. All data of above signals received from center analyzed data (NCEP) during 1960 to 2003. In order to determine the rate of importance of these parameters on quantity of precipitation was used multivariate regression method. Results of regression analysis show that ENSO index in zone of NINO1+2, NINO3 and NINo3.4 strong correlations with the variations of precipitation. In this study long- Range precipitation prediction for the time period, 3 and 6 months was done. Analysis of artificial neural network model results in comparisons with observations show that the warm phases of ENSO are accompanied with more rainy periods and, cold phases of ENSO with less rainy periods.