Using Causal Discovery Methods to Explore Subseasonal Teleconnections in a Changing Climate.

The Madden Julian Oscillation (MJO) is well known as an influencer of the global atmospheric circulation and a primary source of predictability at subseasonal-to-seasonal (S2S) time scales. In this study, we use different causal discovery methods, including nonlinear methods that bring together the ideas of Granger causality and artificial neural networks, to identify possible cause-effect relationships between the MJO and multiple climate indices based on the stratospheric polar vortex and the North Atlantic circulation. We use these methods to investigate the tropospheric and stratospheric teleconnections of the MJO in the present climate and how these interactions change in the future climate projections. Compared to the traditional approach of using targeted model studies to attribute cause-effect relationships, these causal discovery methods allow atmospheric scientists to gain additional insights into interactions while accounting for internal memory and feedbacks between variables.