Estimation and prediction of temperature in Iraq using the multi-layered neural network model

Abstract

The forecasting using the multi-layered neural network model is one of the methods used recently in fore-casting, especially in climate forecasts for certain regions, because of its accuracy in forecasting, which sometimes reaches levels close to the real collected data. In this research, the daily temperatures in the cli-mate of Iraq were predicted, by taking data from the Iraqi Meteorological Authority by (228) observations, which represent the daily temperatures of Karbala Governorate in the year (2021), The results of the auto-correlation and partial autocorrelation showed that the daily temperature series of Karbala governorate is unstable, and this was confirmed by conducting the augmented Dickey Fuller test. The data was analyzed using the multi-layered neural network model in two stages, and it was later shown that the accuracy of estimation and prediction using the multi-layered neural network even if the time series is not stable, The results showed an indication of an rising increase in temperatures during the coming years. The researcher concluded that it is necessary to pay attention to the vegetation cover and to conduct many predictive studies of the climate using the multi-layered neural network.