Paleoclimate data constraints on climate sensitivity: The paleocalibration method

Abstract. The relationship between paleoclimates and the future climate, while not as simple as implied in the 'paleoanalog ' studies of Budyko and others, nevertheless provides ufficient constraints to broadly confirm the climate sensitivity range of theoretical models and perhaps eventually narrow the model-derived uncertainties. We use a new technique called 'paleocalibration ' to calculate the ratio of temperature response to forcing on a global mean scale for three key intervals of Earth history. By examining surface conditions reconstructed from geologic data for the Last Glacial Maximum, the middle Cretaceous and the early Eocene, we can estimate the equilibrium climate sensitivity to radiative forcing changes for different extreme climates. We find that the ratios for these three periods, within error bounds, all lie in the range obtained from general circulation models: 2-5 K global warming for doubled atmospheric carbon dioxide. Paleocalibration thus provides adata-based confirmation oftheoretically calculated climate sensitivity. However, when compared with paleodata on regional scales, the models how less agreeement with data. For example, our GCM simulation of the early Eocene fails to obtain the temperature contrasts between the Equator and the Poles (and between land and ocean areas) indicated by the data, even though it agrees with the temperature data in the global average. Similar esults have been reported by others for the Cretaceous and for the Last Glacial Maximum. 1