Causal Discovery for Climate Research Using Graphical Models

TLDR

Causal discovery uses graphical models to recover cause–effect relationships from statistical data, offering an alternative to correlation‑based climate networks by focusing on information flow rather than node similarity. The paper aims to introduce causal discovery methods to climate scientists and demonstrate constraint‑based structure learning through a case study of four boreal winter atmospheric modes, proposing its use for climate networks. The authors apply constraint‑based structure learning to derive causal hypotheses, presenting results as static and temporal independence graphs (Bayesian Networks). The study finds strong simultaneous coupling between WPO and EPO, shows that changes in EPO or NAO can cause subsequent changes in NAO or PNA after 3–18 days, and aligns with prior dynamical process findings while suggesting new hypotheses on timing and sequencing.

Abstract

Abstract Causal discovery seeks to recover cause–effect relationships from statistical data using graphical models. One goal of this paper is to provide an accessible introduction to causal discovery methods for climate scientists, with a focus on constraint-based structure learning. Second, in a detailed case study constraint-based structure learning is applied to derive hypotheses of causal relationships between four prominent modes of atmospheric low-frequency variability in boreal winter including the Western Pacific Oscillation (WPO), Eastern Pacific Oscillation (EPO), Pacific–North America (PNA) pattern, and North Atlantic Oscillation (NAO). The results are shown in the form of static and temporal independence graphs also known as Bayesian Networks. It is found that WPO and EPO are nearly indistinguishable from the cause–effect perspective as strong simultaneous coupling is identified between the two. In addition, changes in the state of EPO (NAO) may cause changes in the state of NAO (PNA) approximately 18 (3–6) days later. These results are not only consistent with previous findings on dynamical processes connecting different low-frequency modes (e.g., interaction between synoptic and low-frequency eddies) but also provide the basis for formulating new hypotheses regarding the time scale and temporal sequencing of dynamical processes responsible for these connections. Last, the authors propose to use structure learning for climate networks, which are currently based primarily on correlation analysis. While correlation-based climate networks focus on similarity between nodes, independence graphs would provide an alternative viewpoint by focusing on information flow in the network.