Sensitivity of the Eocene climate to CO2 and orbital variability

Keery, John S.; Holden, Philip B. and Edwards, Neil R. (2018). Sensitivity of the Eocene climate to CO2 and orbital variability. Climate of the Past, 14 pp. 215–238.



The early Eocene, from about 56 Ma, with high atmospheric CO2 levels, offers an analogue for the response of the Earth’s climate system to anthropogenic fossil fuel burning. In this study, we present an ensemble of 50 Earth system model runs with an early Eocene palaeogeography and variation in the forcing values of atmospheric CO2 and the Earth’s orbital parameters. Relationships between simple summary metrics of model outputs and the forcing parameters are identified by linear modelling, providing estimates of the relative magnitudes of the effects of atmospheric CO2 and each of the orbital parameters on important climatic features, including tropical–polar temperature difference, ocean–land temperature contrast, Asian, African and South (S.) American monsoon rains, and climate sensitivity. Our results indicate that although CO2 exerts a dominant control on most of the climatic features examined in this study, the orbital parameters also strongly influence important components of the ocean–atmosphere system in a greenhouse Earth. In our ensemble, atmospheric CO2 spans the range 280–3000 ppm, and this variation accounts for over 90 % of the effects on mean air temperature, southern winter high-latitude ocean– land temperature contrast and northern winter tropical–polar temperature difference. However, the variation of precession accounts for over 80 % of the influence of the forcing parameters on the Asian and African monsoon rainfall, and obliquity variation accounts for over 65 % of the effects on winter ocean–land temperature contrast in high northern latitudes and northern summer tropical–polar temperature difference. Our results indicate a bimodal climate sensitivity, with values of 4.36 and 2.54 ◦C, dependent on low or high states of atmospheric CO2 concentration, respectively, with a threshold at approximately 1000 ppm in this model, and due to a saturated vegetation–albedo feedback. Our method gives a quantitative ranking of the influence of each of the forcing parameters on key climatic model outputs, with additional spatial information from singular value decomposition providing insights into likely physical mechanisms. The results demonstrate the importance of orbital variation as an agent of change in climates of the past, and we demonstrate that emulators derived from our modelling output can be used as rapid and efficient surrogates of the full complexity model to provide estimates of climate conditions from any set of forcing parameters.

Viewing alternatives

Download history


Public Attention

Altmetrics from Altmetric

Number of Citations

Citations from Dimensions

Item Actions