Time Series Decomposition and Prediction of the Monthly Temperature Some Stations in Iraq

Predicting temperature is a difficult problem due to the complexity and chaos of climate. Accurate temperature prediction is crucial for industries, agriculture, energy management, and environmental sustainability. The current study examined the time series components and developed a suitable model to predict the temperature of the Mosul, Baghdad, Rutba, and Basrah stations in Iraq using the monthly mean temperature data from January 1980 to December 2021. The study aimed to use the Auto-Regressive Integrated Moving Average to select the best formula for predicting the monthly temperature values based on some statistics errors between estimated and observed values. For evaluating the model, criteria like the Bayesian Information Criterion, Akaike Information Criterion, corrected Akaike Information Criterion, and residual variance (σ2) were used. The results showed that the best formula for the model ARIMA for the Baghdad station was (2,0,2) (1,1,0) 12, for the Rutba station was (2,0,2) (2,1,0) 12, for the Mosul station was (1,0,0) (1,1,0) 12 and for the Basrah station was (0,0,1) (0,1,1) 12. Model analysis for Basrah identified the SARIMA (0,0,1)(0,1,1) 12, configuration as the optimal model based on diagnostic tests, stability, and reliability assessments. Additionally, trend decomposition and five-year forecasts of the monthly air temperature time series indicate a significant upward trend in temperature across all stations. The model has been used to examine the precise influence of time series variables on the regional warming scenario's scale. As a result, the developed model can assist in formulating plans regarding Iraqi weather conditions. The average temperature appears to be trending upward based on the anticipated outcome. The weather and economic activity in every region of Iraq could be disrupted by such an increasingly evident trend.