In a recent Science perspective Peter Cox and Chris Jones show why things are probably worse than we think and may become worse than we can think. A major issue for climate science is the availability of a second Earth. A duplicate would be very handy to experiment on, or at least use as a second example. This, of course means that we can study in detail what is happening today, and in the absence of a way-back machine we can use proxies to look at the past. However, in both these activities, natural variability (aka a combination of random chance, chaotic behavior and stuff we don't know enough about) provides wide ranges of various climate sensitivities and looking at only one period badly constrains them.
While there are not very many if any unknown unknowns there sure are a pile of badly known knowns (Eli plays the anti-Rumsfeld here) among which are CO2 sensitivity of carbon stores and the climate sensitivity of CO2.
The first is one of the denialists' favorite outs. It expresses how increases in CO2 will increase (or decrease) the amount of carbon stored in soils/oceans, etc. It determines changes in the flux of CO2 from the atmosphere to other reservoirs and is given in GtC/ppmv or gigatons of carbon stored elsewhere per part per million change in atmospheric CO2. The ranters from the right (mostly they come from there although some are way out in left field and others, like the Larouchies are commuting from Mars) claim that plants will flourish at high atmospheric CO2 in a slightly warmer world, net primary production of vegetative matter increase and, dear Candide Bunny this will be the best of all possible worlds. Unfortunately for us, it appears that this effect will saturate well before Peabody digs up all the coal and burns it.
The second, on the other hand (this is a Science blog at base, and there is nothing without caveats) asks the question, if you have increase (or decrease the amount of carbon in soils and the oceans, how does that affect the global temperature. It is expressed in GtC/K or the change in the amount of stored carbon per degree K. Now there are several effects here. One, of course is you can change the amount of carbon stored by burning fossil fuel. Or you could change it by decomposing leaves (something we in the Northern Hemisphere are currently doing, come to think about it we are also burning a lot of fossil fuel). Or you could change it if the temperature of the oceans changes, and so on.
Cox and Jones point out that modern measurements badly constrain both of these (pink band in the leftmost part of the image from their article) given interannual variability (light green band, the overlap being shown in brown), but that if we add information from the Little Ice Age in the Northern Hemisphere (turquoise band in the rightmost panel) we can narrow the range considerably (purple area in the right panel).
First, Eli obviously has to work on him image skills, but in the meantime you can click on the image to get a clearer view. Second, the middle panel clearly shows a case where temperature driven by other forcings drives CO2 concentrations, but as has been pointed out n+1 times, CO2 in the atmosphere can both be a driving force as is the case when we burn fossil fuel, and a response or feedback, as when, for example the sun cools or heats up.
Eli, being a sunny bunny will leave the last word to Cox and Jones (emphases added for our denialist friends)
Read the comments for further enlightenment
The perturbations of climate and CO2 during the LIA period from 1500 to 1750 are strongly correlated, with climate leading CO2 by ~50 years (11). These records indicate a tight relation between CO2 and climate, with a gradient of 40 ppmv/K. However, given the discrepancies between different temperature reconstructions, and the uncertainties associated with interpreting Northern Hemisphere climate proxies in terms of global mean temperature, we estimate a gradient of 20 to 60 ppmv of CO2 per kelvin of global warming (see the figure, middle panel).
This is a conservative estimate based on the assumption that human CO2 emissions from land-use change were not significant in the LIA, which seems consistent with the strong lead-lag relationship between climate and CO2 during this period. Even so, the estimate is at the high end of the 20th-century simulations with the IPCC C-CC models, encompassing only the model with the largest feedback over this period. When considered alongside contemporary constraints, the LIA data thus enable a much tighter constraint on the climate and CO2 dependences of the carbon cycle (see the figure, right panel).
The LIA data imply that atmospheric CO2 will increase more quickly with global warming than most models suggest. One implication is that the 20th-century CO2 rise due to anthropogenic emissions may have been amplified by 20 to 30 ppmv through the impacts of global warming on natural carbon sinks. Furthermore, the existence of a strong climate effect on the carbon cycle indicates that larger emissions cuts are required to stabilize CO2 concentrations at a given level. The LIA is just one example of a natural climatic anomaly in the past that can provide insights into the strength of the coupling between the Earth's climate and carbon cycle. Paleoclimatic data cannot tell us how to meet the challenge of managing 21st-century climate change, but they can help us to better understand the nature of this challenge.