Extended Data Fig. 1: Schematic of causal discovery using CCM model, illustrated with the example “ALT causes VPD”. | Nature Geoscience

Extended Data Fig. 1: Schematic of causal discovery using CCM model, illustrated with the example “ALT causes VPD”.

From: Amplified Arctic–boreal fire regimes from permafrost thaw feedbacks

Extended Data Fig. 1: Schematic of causal discovery using CCM model, illustrated with the example “ALT causes VPD”.The alternative text for this image may have been generated using AI.

a, Year-to-year variations of ALT and VPD serve as inputs to the CCM model. b, The CCM model detects causality based on non-linear state space reconstruction. It first constructs shadow manifolds for ALT (MALT,t) and VPD (MVPD,t) in a three-dimensional state space using the original (ALT(t); VPD(t)) and two time-lagged series (ALT(t-τ), ALT(t-2τ); VPD(t-τ), VPD(t-2τ)). Predict skill (ρ) is then used to assess whether MVPD,t can reliably predict the state of ALT. c, If ρ converges as the increase of time series length (L), it suggests that ALT causally influences VPD. This convergence is assessed by calculating Spearman’s rank correlation between ρ and L, identifying causality only when the correlation is significantly positive (Methods). Maps created in ArcGIS (v.10.8) using a base map derived from the IPCC-WG1 Atlas GitHub repository80.

Back to article page