Stochastic Exploration of Transition Pathways for Aviation Using Temporal Convolutional Networks

There is a growing sense of urgency when it comes to reducing aviation's environmental impact. Several comprehensive studies have been published in the recent years, setting out goals for the industry and providing recommendations on what actions key stakeholders can take to gradually transition toward a net-zero future. However, a lot of uncertainty remains in the actual impacts that the stakeholders' decisions will have on the industry. In order to analyze how this uncertainty translates to uncertainty on CO2 emissions, this study sets out to develop a methodology enabling a probabilistic evaluation of specific decision scenarios. First, a framework to construct a surrogate model linking a decision scenario timeline to its corresponding evolution of CO2 emissions over time is introduced. Then, this model is used to analyze the sensitivity of a few scenarios to variations in their respective decisions' impacts.