Enhancing \({\mathbf{C}\mathbf{O}}_{2}\) emissions predictions through historical events-aware artificial intelligence models

Abstract

Accurate prediction of Carbon dioxide (\({\text{CO}}_{2}\)) emissions is crucial for informed decision-making and proactive measures to combat climate change. Anticipating future emissions trends empowers policymakers, businesses, and environmental agencies to devise strategies for emission reduction and adaptation to evolving environmental conditions. This paper explores first the intricate relationship between historical events and their impact on \({\text{CO}}_{2}\) emissions through advanced time series analysis models and introduces then a methodology that integrates historical events into multivariate forecasting models to enhance the prediction of future \({\text{CO}}_{2}\) emissions. Using time series analysis trained on extensive historical data, the results reveal distinct emissions patterns tied to these events, showcasing the necessity of considering multifaceted historical factors in \({\text{CO}}_{2}\) emissions predictions. The paper results demonstrate as well that the proposed methodology can outperform traditional forecasting methods, underscoring its robustness and predictive accuracy. The paper results not only emphasize the importance of integrating historical context into emissions forecasts but also provides valuable insights for policymakers and researchers aiming to devise more effective strategies for emission reduction and climate adaptation.