DeepMIP: Model intercomparison of early Eocene climatic optimum (EECO) large-scale climate features and comparison with proxy data

Abstract

Abstract. We present results from an ensemble of seven climate models, each of which has carried out simulations of the early Eocene climate optimum (EECO, ~ 50 million years ago). These simulations have been carried out in the framework of DeepMIP (www.deepmip.org), and as such all models have been configured with identical paleogeographic and vegetation boundary conditions. The results indicate that these non-CO2 boundary conditions contribute between 3 and 5 °C to Eocene warmth. Compared to results from previous studies, the DeepMIP simulations show reduced spread of global mean surface temperature response across the ensemble, for a given atmospheric CO2 concentration. In a marked departure from the results from previous simulations, at least two of the DeepMIP models (CESM and GFDL) are consistent with proxy indicators of global mean temperature, and atmospheric CO2, and meridional SST gradients. The best agreement with global SST proxies from these models occurs at CO2 concentrations of around 2400 ppmv. At a more regional scale the models lack skill in reproducing the proxy SSTs, in particular in the southwest Pacific, around New Zealand and south Australia, where the modelled anomalies are substantially less than indicated by the proxies. However, in these regions modelled continental surface air temperature anomalies are consistent with surface air temperature proxies, implying an inconsistency between marine and terrestrial temperatures in either the proxies or models in this region. Our aim is that the documentation of the large scale features and model-data comparison presented herein will pave the way to further studies that explore aspects of the model simulations in more detail, for example the ocean circulation, hydrological cycle, and modes of variability; and encourage sensitivity studies to aspects such as paleogeography and aerosols.

Surface-circulation change in the southwest Pacific Ocean across the Middle Eocene Climatic Optimum: inferences from dinoflagellate cysts and biomarker paleothermometry

“…The middle Eocene sporomorph assemblages from the Latrobe-1 borehole are generally similar to those identified in previous studies (Macphail et al, 1994;Greenwood et al, 2003;Hill, 2017), but also include a small proportion of mesothermal-megathermal ("paratropical") components. Although the small number of analyzed samples prohibits a description of pre-, syn-, and post-MECO vegetation, the assemblages from the Latrobe-1 core reveal that this middle Eocene vegetation of coastal southeast Australia consisted of a mosaic of mesothermal rainforest flora.…”

“…For this group, we follow the suggestion made in Bijl et al (2016) to use the taxonomy of Fensome and Williams (2004) instead of Williams et al ( , 2017a. Sporomorph taxonomy follows Stover and Partridge (1973), Macphail et al (1994), and Raine et al (2011).…”

Quantifying the Effect of the Drake Passage Opening on the Eocene Ocean

“…This bias is very classical in warm climate reconstructions with GCMs and could be responsible for a discrepancy between absolute SST from our experiments and proxies (see Huber & Caballero, 2011). Studies carried out with more recent versions of GCM, such as CESM and GFDL, show improvements in the representation of this gradient, in particular thanks to a better consideration of cloud physics and other greenhouse gases (Baatsen et al, 2020;Hutchinson et al, 2020;Lunt et al, 2020;Sagoo et al, 2013;Zhu et al, 2019). With close boundary conditions, they reconstruct flatter gradients thanks to~3°C lower SST in equatorial area and up to 2°C to 3°C higher SST at midlatitudes ( Figure S7; Baatsen et al, 2020;Hutchinson et al, 2018).…”

Paleobotanical proxies for early Eocene climates and ecosystems in northern North America from middle to high latitudes

“…Finally, the physiognomic characteristics of fossil, and a subset of modern, sites were compared using a hierarchical cluster analysis to determine similarity. This was achieved by calculating a Euclidean dissimilarity matrix without scaling using the "cluster" package in R (Maechler et al, 2019) and the "Ward D2" method for hierarchical clustering (R Core Team, 2019).…”