Effect of sea surface temperature configuration on model simulations of “equable” climate in the Early Eocene

A major challenge in paleoclimate modeling studies is reproducing the “equable” climate conditions recorded by proxy climate data. This challenge has long been a focus in studies of Eocene paleoclimate. Climate models consistently overestimate mean annual temperature range (MATR) relative to proxy data interpretations and produce minimum temperatures that are far lower than proxy estimates. We hypothesize that the lack of accurate sea surface temperatures (SSTs) and a complete annual cycle definition is responsible for some of these model-data discrepancies in Eocene comparisons. To test this hypothesis, we developed two Eocene annual cycle SST data sets as boundary conditions for model experiments and ran full annual cycle and abbreviated cycle (perpetual) cases. When a perpetual case is replaced by a full annual cycle of SST values, the January 0°C isotherm (freezeline) over North America shifts poleward by ∼5° latitude and (MATR is reduced by ∼5°C. A change from one annual SST cycle to another, equally plausible, annual SST cycle results in a further latitudinal migration (∼5°) of the freezeline and another ∼5°C change in continental interior MATR. Overall, the global mean annual and cold month mean temperatures show little sensitivity to the forcing, while MATR shows larger sensitivity in some continental regions. Our results suggest that the annual cycle of SSTs and the actual SST values incorporated into experiments are major sources of the model-data continental temperature discrepancies reported in past paleoclimate modeling studies. We also find that the relatively coarse spatial resolution of the topography incorporated into paleoclimate models is an additional source of the discrepancy.