Abstract

Dansgaard-Oeschger (DO) events are large, abrupt millennial-scale climate reorganizations that occurred during the ice ages. Each event is characterized by an abrupt warming of 5 to 16.5°C during transitions from stadial (cold) to interstadial (warm) periods, with recurrence times varying from approximately 1 to nearly 9 thousand years. These events have long been associated with changes in the strength of the Atlantic Meridional Overturning Circulation, yet the causes of these changes are still under debate. A primary obstacle hindering the understanding of their mechanisms is the unclear relationship between different climate variables. In this research, I aim to employ two causal methods: Granger Causality and PCMCI , to reconstruct the causal relationships among DO event climate variables from climate model outputs. Before that, simple systems mimicking DO event dynamics have been built to study the applicability of these causal methods on the highly non-linear, non-stationary DO event data. My analyses showed that each causal method has unique advantages and limitations and requires careful consideration when utilising these methods to study DO event causalities.