breakdown of frozen gas-ice compounds (clathrates) in permafrost or the ocean floor is in this category. Methane release has been clearly implicated in the warm event at the end of the Paleocene (Kennett and Stott, 1991; Dickens et al., 1995) (Box 4.1), and it has been argued that some of the Pleistocene methane signal is due to clathrate decomposition (Kennett et al., 2000) rather than tropical or high-latitude hydrological circumstances.
Furthermore, it must be acknowledged that the earth’s climate system has in the more distant past exhibited major switches in mode of operation that are simply not understood. Notably, the past climate has oscillated between hothouse climates lasting tens of millions of years, when there was little or no permanent polar ice, and icehouse climates like those of the present and the Pleistocene. Both states have occurred throughout geological history. The most recent period of increased warmth continued throughout the Cretaceous (65 million years ago) into the Eocene (55 million years ago) and terminated with the onset of major ice ages about 2 million years ago. However, there were also icehouse periods earlier in the earth’s history, including times during the Carboniferous and the Neo-Proterozoic. Although it is generally believed that geochemically mediated changes in atmospheric carbon dioxide played a major role in such transitions, there has been little success in reproducing the key features of hothouse climates by increased carbon dioxide alone. Concentrations of carbon dioxide high enough to prevent permanent polar ice in models generally lead to simulation of tropical oceans warmer than suggested by available data (Manabe and Bryan, 1985); the realism of both the tropical temperatures and the very high carbon dioxide levels are still under debate (Pearson et al., 2001). It had been hoped that better understanding of dynamic ocean heat transport would solve the problem, but recent work on Cretaceous and Eocene ocean dynamics does not support this idea. Moreover, even in simulations with increased carbon dioxide, continental interiors become too cold in the winters to reconcile with the equable climate that the fossil record demands. The problem of hothouse-icehouse transitions underscores that as-yet-unidentified mechanisms for mediating radical changes, some of which could well be abrupt, are lurking in the climate system.
Compounding the mystery of initiation and maintenance of the above “hot” mode of the climate is the growing evidence that the earth has fallen into an extremely cold “snowball-earth” state, in which the entire planet became ice-covered. The most recent occurrence of a snowball state was in the Neo-Proterozoic, about 600 million years ago. The circumstances in which the Snowball can be triggered are hotly debated but almost certainly