Daniel Kirk-Davidoff

Understanding Eocene Polar Warmth

My work on paleoclimate modeling has focused on two periods in the s past. In the Eocene epoch, starting 55.8 million years ago (Ma), temperatures at high latitudes were shockingly high: annual Arctic ocean temperatures averaged at least , and were over during an especially warm period at the beginning of the Eocene (the Paleocene-Eocene Thermal maximum, or PETM). This compares to present annual mean temperatures of about -. This poses very difficult problems for climate dynamics. Eocene warmth is generally understood to have been a result of high greenhouse gas concentrations, especially carbon dioxide. Since the radiative forcing by carbon dioxide depends in a logarithmic fashion on its concentration in the atmosphere, high temperature variability, in the context of already high temperatures would seem to require enormous fluxes of carbon into and out of the atmosphere. Independent estimates of these fluxes make concentrations more than twice present levels for the mid-Eocene unlikely. In short, the record of Eocene climate suggests that real overall sensitivity of the s climate to forcing by carbon dioxide is high compared to the mid-range of general circulation model (GCM) estimates, and that the polar regions are especially sensitive.

To solve this problem, we need to find mechanisms that can explain the extraordinary warmth of the Eocene Arctic, and are currently neglected or mischaracterized in computer models of climate. My work in this area focuses on two candidate mechanisms. The first is polar stratospheric clouds. My papers (Kirk-Davidoff et al. 2002; Kirk-Davidoff and Lamarque, 2008) established the conditions under which an initial warming due to increasing greenhouse gas concentrations could have warmed the tropical tropopause sufficiently to allow enough extra moisture into the stratosphere to form clouds that could substantially warm the polar regions. This mechanism is difficult to capture in GCMs, because the models of cloud formation typically used are not designed for stratospheric conditions. My work (and much other work) encouraged inclusion of more sophisticated cloud physics schemes in GCMs, which should improve treatment of both past and future climate. More recently I have investigated an alternative hypothesis, that deep convective clouds in the Eocene warm polar night might tend to produce more high cirrus cloud than is predicted in GCMs, again providing a positive feedback to surface warming. To investigate this claim, Amy Solomon of NOAA CIRES and I are running simulations of Eocene conditions using a high-resolution weather forecast model. Our results tend to refute the hypothesis, showing that a sophisticated treatment of convection yields only a small increase in high cloud when surface temperature warm, while low clouds are reduced. These results tend to emphasize the potential importance of the stratospheric cloud mechanism. The figure below shows a series of model simulated low cloud fraction using Polar WRF and CAM, forced with fixed sea surface temperatures increasing from 10°C to 12°C to 15°C. Note that while CAM shows an increase in low clouds as surface temperatures increase, three difference convection schemes in WRF all result in a decrease in low cloud fraction with increasing sea surface temperature.


Kirk-Davidoff, D.B., J.-F. Lamarque, 2008: Maintenance of polar stratospheric clouds in a moist stratosphere. Climate of the Past,4:69-78. PDF. Kirk-Davidoff, D.B., D.P. Schrag, and J.G. Anderson, 2002: On the Feedback of Stratospheric Clouds on Polar Climate. Geophys. Res. Let.. 29(11), 10.1029/2002GL014659. PDF.