Progress in paleoclimate modeling

This paper briefly surveys areas of paleoclimate modeling notable for recent progress. New ideas, including hypotheses giving a pivotal role to sea ice, have revitalized the low-order models used to simulate the time evolution of glacial cycles through the Pleistocene, a prohibitive length of time for comprehensive general circulation models (GCMs). In a recent breakthrough, however, GCMs have succeeded in simulating the onset of glaciations. This occurs at times (most recently, 115 kyr B.P.) when high northern latitudes are cold enough to maintain a snow cover and tropical latitudes are warm, enhancing the moisture source. More generally, the improvement in models has allowed simulations of key periods such as the Last Glacial Maximum and the mid-Holocene that compare more favorably and in more detail with paleoproxy data. These models now simulate ENSO cycles, and some of them have been shown to reproduce the reduction of ENSO activity observed in the early to middle Holocene. Modeling studies have demonstrated that the reduction is a response to the altered orbital configuration at that time. An urgent challenge for paleoclimate modeling is to explain and to simulate the abrupt changes observed during glacial epochs (i.e., Dansgaard-Oescher cycles, Heinrich events, and the Younger Dryas). Efforts have begun to simulate the last millennium. Over this time the forcing due to orbital variations is less important than the radiance changes due to volcanic eruptions and variations in solar output. Simulations of these natural variations test the models relied on for future climate change projections. They provide better estimates of the internal and naturally forced variability at centennial time scales, elucidating how unusual the recent global temperature trends are.