I’ve tried to stay away from discussing GCMs where my knowledge is limited, but everyone seems to want to discuss them, so, against my better judgement, I’m posting up some thoughts. asked some people at AGU about whether GCMs could model getting into and getting out of ice ages. In some treatments, the presence or absence of ice sheets appears as an independent forcing, which seems to be peeking at the hand a little when you’re trying to explain ice ages. CO2 fluctuations are integral to current glacial modeling, but there doesn’t seem to be any agreed explanations for glacial/interglacial CO2 cycles.
Caspar Ammann said that GCMs took about 1 day of machine time to cover 25 years. On this basis, it is obviously impossible to model the Pliocene-Pleistocene transition (say the last 2 million years) using a GCM as this would take about 219 years of computer time. So I think we can safely conclude that any models of this period have not been done using a GCM. Even a “short” period of (say) 40,000 years, about one obliquity cycle, would require about 4 years of computer time and, again, I think that we can assume that this exercise has not been done using a GCM. Presumably models of such time intervals are done with “intermediate-complexity” models or some other method, but I’ve not attempted to canvass the literature.
With respect to ice age models, Peter Huybers said that some models could plausibly get into ice ages, but, once in, couldn’t get out; conversely, other models could plausibly get out of ice ages, but couldn’t get in. If I recollect correctly, he thought that no intermediate-complexity model did both. Also if I recollect correctly, he thought that no intermediate-complexity model accurately modeled the Pliocene-Pleistocene transition. Some models include ice sheets as a “forcing” variable, which seems unreasonable if you’re trying to explain ice ages. Other models include CO2 levels as a “forcing” variable, which again seems unreasonable since it seems to be endogenous to the system being modeled.
I’ve been browsing some of the literature on CO2 cycles. Obviously the covariance in Antarctic cores of CO2 and àÅ½àⳏ18 (held to be a temperature proxy) is one of the remarkable aspects of these cores. Changes in CO2 levels are popularly held to be important in explaining the amplitude of major glacial changes, but there is no agreement on what causes the changes in CO2 levels. I’ll discuss this on another occasion, but, for now, I’ll mention that most explanations involve biological feedback. I suspect that the handling of biological feedback issues is not strongly done in GCMs and/or intermediate-complexity models (but haven’t checked this and could be wrong.) There seems to be a general view that lowering CO2 levels is essential to getting SH glaciation.
As to GCMs providing a base case for assessing natural variability: one of my starting points in reflecting about climate change is that one wanted to be confident as to an understanding of big changes in order to assess the significance of “little” changes such as the LIA or MWP. If we do not have GCM models of the entry into and out of ice ages, then I don’t see how GCMs can serve as a benchmark for “natural climate variability. What am I missing?