Drive-Response Analysis of Global Ice Volume, CO2, and Insolation using Information Transfer

Abstract:

The processes and interactions that drive global ice volume variability and deglaciations are a topic of considerable debate. Here we analyze the drive-response relationships between data sets representing global ice volume, CO₂ and insolation over the past 800 000 years using an information theoretic approach. Specifically, we use a non-parametric measure of directional information transfer (IT) based on the construct of transfer entropy to detect the relative strength and directionality of interactions in the potentially chaotic and non-linear glacial-interglacial climate system. Analyses of unfiltered data suggest a tight coupling between CO₂ and ice volume, detected as strong, symmetric information flow consistent with a two-way interaction. In contrast, IT from Northern Hemisphere (NH) summer insolation to CO₂ is highly asymmetric, suggesting that insolation is an important driver of CO₂. Conditional analysis further suggests that CO₂ is a dominant influence on ice volume, with the effect of insolation also being significant but limited to smaller-scale variability. However, the strong correlation between CO₂ and ice volume renders them information redundant with respect to insolation, confounding further drive-response attribution. We expect this information redundancy to be partly explained by the shared glacial-interglacial "sawtooth" pattern and its overwhelming influence on the transition probability distributions over the target interval. To test this, we filtered out the abrupt glacial terminations from the ice volume and CO₂ records to focus on the residual variability. Preliminary results from this analysis confirm insolation as a driver of CO₂ and two-way interactions between CO₂ and ice volume. However, insolation is reduced to a weak influence on ice volume. Conditional analyses support CO₂ as a dominant driver of ice volume, while ice volume and insolation both have a strong influence on CO₂. These findings suggest that the effect of orbital variability on global ice volume may work primarily through its influence on CO₂. Our preliminary results are consistent with the idea that the coupling between CO₂ and ice volume likely occurs via a feedback loop that involves meltwater-induced shifts in oceanic circulation and associated changes in the carbon cycle.