Identifying Physical Interactions from Climate Data: Challenges and Opportunities

Recent research has shown the potential of using structure learning from probabilistic graphical models to identify and visualize interactions in the Earth´s climate system by training them on observed climate data. The resulting models indicate pathways of physical interactions occurring within a subsystem of the climate (such as the atmosphere) or between different subsystems (such as from ocean to atmosphere). Studying these pathways is of great interest to climate scientists because it allows them to learn subtle details about the underlying dynamical mechanisms governing our planet´s climate. Here, the authors focus on interactions in one specific subsystem, the atmosphere. The details of this research have been discussed previously, so they focus on providing a general overview and discussing challenges and opportunities for this emerging area.