Climatic Temperature Data Forecasting in Nineveh Governorate Using the Recurrent Neutral Network Method

The forecasting of maximum climatic temperature is essential by using some statistical and intelligent techniques. Iraqi maximum temperature data collected monthly in several cities due to the Nineveh government will be studied in this paper. This study aims to forecast maximum climatic temperature as univariate time series and obtain the best results with minimum forecasting error. The non-linearity of climatic datasets is the main reason for data complexity, which needs to use some nonlinear methods for obtaining satisfactory results. In this paper, the maximum climatic temperature data will be forecasted by using traditional and intelligent methods. Single and double exponential smoothing (SES and DES) models have been used as traditional linear methods to forecast climatic temperature. The forecasting results reflected that the hybrid methods outperformed the traditional methods. The proposed hybrid methods can forecast climatic temperature in more accurate results. The hybrid methods SES-RNN and DES-RNN combine the SES and DES as a linear model with RNN as a nonlinear method to be one method that can handle any data, especially the nonlinear type. Recurrent neural network (RNN) as the nonlinear intelligent method is combined with SES and DES in hybrid SES-RNN and DES-RNN methods to forecast climatic temperature data and handle the non-linearity of datasets. The results reflect that the proposed hybrid methods outperformed the traditional methods for forecasting climatic temperature data. The proposed hybrid methods can be used to forecast climatic temperature in more accurate results.