Predicting monthly precipitation with signals and climatic elements, Case study: Dezful city

In this study, in order to Dezful precipitation modeling were Used from the monthly precipitation data of Dezful Joint Station in the statistical period (2014-1961) for 53 years as a dependent variable and climatic indicators and climatic elements as independent variable. Factor analysis was used to apply the most important climatic elements affecting the study area, and varieties of regression analysis methods were used to identify the most important climatic signals affecting the dependent variable.
Due to the nonlinear behavior of precipitation, artificial neural networks were used for modeling. To enter the neural network, precipitation data were applied to regression analysis. The results of the study revealed that the prediction of climatic signals had a higher correlation coefficient. For example, the correlation coefficient with the gradual data removal method was 100% and the step-by-step method was 99.88%. Between of 144 units (climatic signal per month), 16 units or indicators were selected in 18 months with 18 networks for precipitation with a correlation coefficient of 99.88% for forecasting. To predict precipitation with climatic elements, among the 7 effective components, the second component of precipitation-temperature was 100% probably successful in regression and 99.7% probably in nervous system.