Science stunner: On our current emissions path, CO2 levels in 2100 will hit levels last seen when the Earth was 29°F (16°C) hotter

Paleoclimate data suggests CO2 “may have at least twice the effect on global temperatures than currently projected by computer models”

The disinformers claim that projections of dangerous future warming from greenhouse gas emissions are based on computer models. In fact, ClimateProgress readers know that the paleoclimate data is considerably more worrisome than the models (see Hansen: ‘Long-term’ climate sensitivity of 6°C for doubled CO2). That’s mainly because the vast majority of the models largely ignore key amplifying carbon-cycle feedbacks, such as the methane emissions from melting tundra (see Are Scientists Underestimating Climate Change).

Science has just published an important review and analysis of “real world” paleoclimate data in “Lessons from Earth’s Past” (subs. req’d) by National Center for Atmospheric Research (NCAR) scientist Jeffrey Kiehl. The NCAR release is here: “Earth’s hot past could be prologue to future climate.” The study begins by noting:

Climate models are invaluable tools for understanding Earth’s climate system. But examination of the real world also provides insights into the role of greenhouse gases (carbon dioxide) in determining Earth’s climate. Not only can much be learned by looking at the observational evidence from Earth’s past, but such know ledge can provide context for future climate change.

The atmospheric CO2 concentration currently is 390 parts per million by volume (ppmv), and continuing on a business-as-usual path of energy use based on fossil fuels will raise it to ˆ¼900 to 1100 ppmv by the end of this century (see the first figure) (1). When was the last time the atmosphere contained ˆ¼1000 ppmv of CO2? Recent reconstructions (2–4) of atmospheric CO2 concentrations through history indicate that it has been ˆ¼30 to 100 million years since this concentration existed in the atmosphere (the range in time is due to uncertainty in proxy values of CO2). The data also reveal that the reduction of CO2 from this high level to the lower levels of the recent past took tens of millions of years. Through the burning of fossil fuels, the atmosphere will return to this concentration in a matter of a century. Thus, the rate of increase in atmospheric CO2 is unprecedented in Earth’s history.

I will repost the references at the end, since this is a review article (see also U.S. media largely ignores latest warning from climate scientists: “Recent observations confirm “¦ the worst-case IPCC scenario trajectories (or even worse) are being realised” “” 1000 ppm)

So now the question is — how much warmer was it back then?

What was Earth’s climate like at the time of past elevated CO2? Consider one example when CO2 was ˆ¼1000 ppmv at ˆ¼35 million years ago (Ma) (2). Temperature data (5, 6) for this time period indicate that tropical to subtropical sea surface temperatures were in the range of 35° to 40°C (versus present-day temperatures of ˆ¼30°C) and that sea surface temperatures at polar latitudes in the South Pacific were 20° to 25°C (versus modern temperatures of ˆ¼5°C). The paleogeography of this time was not radically different from present-day geography, so it is difficult to argue that this difference could explain these large differences in temperature. Also, solar physics findings show that the Sun was less luminous by ˆ¼0.4% at that time (7). Thus, an increase of CO2 from ˆ¼300 ppmv to 1000 ppmv warmed the tropics by 5° to 10°C and the polar regions by even more (i.e., 15° to 20°C).

What can we learn from Earth’s past concerning the climate’s sensitivity to greenhouse gas increases? Accounting for the increase in CO2 and the reduction in solar irradiance, the net radiative forcing””the change in the difference between the incoming and outgoing radiation energy-of the climate system at 30 to 40 Ma was 6.5 to 10 W mˆ’2 with an average of ˆ¼8 W mˆ’2. A similar magnitude of forcing existed for other past warm climate periods, such as the warm mid-Cretaceous of 100 Ma (8). Using the proxy temperature data and assuming, to first order, that latitudinal temperature can be fit with a cosine function in latitude (9), the global annual mean temperature at this time can be estimated to be ˆ¼31°C, versus 15°C during pre-industrial times (around 1750) (10). Thus, Earth was ˆ¼16°C warmer at 30 to 40 Ma. The ratio of change in surface temperature to radiative forcing is called the climate feedback factor (11). The data for 30 to 40 Ma indicate that Earth’s climate feedback factor was ˆ¼2°C Wˆ’1 mˆ’2. Estimates (1, 11) of the climate feedback factor from climate model simulations for a doubling of CO2 from the present-day climate state are ˆ¼0.5 to 1°C Wˆ’1 mˆ’2. The conclusion from this analysis””resting on data for CO2 levels, paleotemperatures, and radiative transfer knowledge””is that Earth’s sensitivity to CO2 radiative forcing may be much greater than that obtained from climate models (12–14).

Indeed, in the release, Kiehl notes his study “found that carbon dioxide may have at least twice the effect on global temperatures than currently projected by computer models of global climate.”


Why is the ‘real world’ warming so much greater than the models? The vast majority of the models focus on the equilibrium climate sensitivity — typically estimated at about 3°C for double CO2 (equivalent to about ¾°C per W/m2) — only includes fast feedbacks, such as water vapor and sea ice. As Hansen has explained in deriving his 6°C ‘long-term’ sensitivity:

Elsewhere (Hansen et al. 2007a) we have described evidence that slower feedbacks, such as poleward expansion of forests, darkening and shrinking of ice sheets, and release of methane from melting tundra, are likely to be significant on decade-century time scales. This realization increases the urgency of estimating the level of climate change that would have dangerous consequences for humanity and other creatures on the planet, and the urgency of defining a realistic path that could avoid these dangerous consequence.

For background on the tundra (and methane), see Science: Vast East Siberian Arctic Shelf methane stores destabilizing and venting: NSF issues world a wake-up call: “Release of even a fraction of the methane stored in the shelf could trigger abrupt climate warming.”

Methane release from the not-so-perma-frost is the most dangerous amplifying feedback in the entire carbon cycle. The permafrost permamelt contains a staggering “1.5 trillion tons of frozen carbon, about twice as much carbon as contained in the atmosphere,” much of which would be released as methane. Methane is is 25 times as potent a heat-trapping gas as CO2 over a 100 year time horizon, but 72 times as potent over 20 years! The carbon is locked in a freezer in the part of the planet warming up the fastest (see “Tundra 4: Permafrost loss linked to Arctic sea ice loss”). Half the land-based permafrost would vanish by mid-century on our current emissions path (see “Tundra, Part 2: The point of no return” and below). No climate model currently incorporates the amplifying feedback from methane released by a defrosting tundra.

Kiehl’s work is in line with other major studies, like this one:

Scientists analyzed data from a major expedition to retrieve deep marine sediments beneath the Arctic to understand the Paleocene Eocene thermal maximum, a brief period some 55 million years ago of “widespread, extreme climatic warming that was associated with massive atmospheric greenhouse gas input.” This 2006 study, published in Nature (subs. req’d), found Artic temperatures almost beyond imagination-above 23°C (74°F)-temperatures more than 18°F warmer than current climate models had predicted when applied to this period. The three dozen authors conclude that existing climate models are missing crucial feedbacks that can significantly amplify polar warming.

How long might it take for the extra warming to kick in? That isn’t known for certain, but two major studies looking at paleoclimate data that Kiehl didn’t cite suggest it’s sooner rather than later:

A study published in Geophysical Research Letters (subs. req’d) looked at temperature and atmospheric changes during the Middle Ages. This 2006 study found that the effect of amplifying feedbacks in the climate system-where global warming boosts atmospheric CO2 levels-”will promote warming by an extra 15 percent to 78 percent on a century-scale” compared to typical estimates by the U.N.’s Intergovernmental Panel on Climate Change. The study notes these results may even be “conservative” because they ignore other greenhouse gases such as methane, whose levels will likely be boosted as temperatures warm.

A second study, published in Geophysical Research Letters, “Missing feedbacks, asymmetric uncertainties, and the underestimation of future warming” (subs. req’d), looked at temperature and atmospheric changes during the past 400,000 years. This study found evidence for significant increases in both CO2 and methane (CH4) levels as temperatures rise. The conclusion: If our current climate models correctly accounted for such “missing feedbacks,” then “we would be predicting a significantly greater increase in global warming than is currently forecast over the next century and beyond”-as much as 1.5°C warmer this century alone.

In the longer term, past 2100, if we were to get anywhere near the kind of warming that Kiehl’s analysis of the paleoclimate data suggests we are headed to, that could render large tracts of the planet uninhabitable. That was the conclusion of a recent PNAS paper coauthored by Matthew Huber, professor of earth and atmospheric sciences at Purdue (release here). I haven’t blogged on it, but I guess I will have to now. The bottom line:

“We found that a warming of 12 degrees Fahrenheit would cause some areas of the world to surpass the wet-bulb temperature limit, and a 21-degree warming would put half of the world’s population in an uninhabitable environment,” Huber said. “When it comes to evaluating the risk of carbon emissions, such worst-case scenarios need to be taken into account. It’s the difference between a game of roulette and playing Russian roulette with a pistol. Sometimes the stakes are too high, even if there is only a small chance of losing.”

So don’t even think about what 29°F warming could mean.

Kiehl concludes:

The above arguments weave together a number of threads in the discussion of climate that have appeared over the past few years. They rest on observations and geochemical modeling studies. Of course, uncertainties still exist in deduced CO2 and surface temperatures, but some basic conclusions can be drawn. Earth’s CO2 concentration is rapidly rising to a level not seen in ˆ¼30 to 100 million years, and Earth’s climate was extremely warm at these levels of CO2. If the world reaches such concentrations of atmospheric CO2, positive feedback processes can amplify global warming beyond current modeling estimates. The human species and global ecosystems will be placed in a climate state never before experienced in their evolutionary history and at an unprecedented rate. Note that these conclusions arise from observations from Earth’s past and not specifically from climate models. Will we, as a species, listen to these messages from the past in order to avoid repeating history?

Will we?

Related Posts:


  1. S. Solomon et al Climate Change 2007: The Physical Science Basis. Contribution of Working Group I to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change, S. Solomon et al., Eds. (Cambridge Univ. Press, Cambridge, UK, 2007).
  2. M. Pagani, J. C. Zachos, K. H. Freeman, B. Tipple, S. Bohaty , Science 309, 600 (2005); 10.1126/science.1110063. doi:10.1126/science.1110063Abstract/FREE Full Text
  3. B. J. Fletcher, S. J. Brentnall, C. W. Anderson, R. A. Berner, D. J. Beerling , Nat. Geosci. 1, 43 (2008). CrossRefWeb of Science
  4. D. O. Breecker, Z. D. Sharp, L. D. McFaddenn , Proc. Natl. Acad. Sci. U.S.A. 107, 576 (2010). Abstract/FREE Full Text
  5. P. K. Bijl, S. Schouten, A. Sluijs, G. J. Reichart, J. C. Zachos, H. Brinkhuis , Nature 461, 776 (2009). CrossRefMedlineWeb of Science
  6. P. N. Pearson et al ., Geology 35, 211 (2007). Abstract/FREE Full Text
  7. D. O. Gough, Sol. Phys. 74, 21 (1981). CrossRef
  8. D. L. Royer, Geochim. Cosmochim. Acta 70, 5665 (2006). CrossRefWeb of Science
  9. G. R. North, J. Atmos. Sci. 32, 2033 (1975). CrossRef
  10. The cosine temperature expression can be integrated analytically to obtain the global annual mean temperature. Paleotemperatures from (5) for a subtropical location and a high southern latitude location were used to determine the two coefficients in the analytical expression for global mean temperature.
  11. S. E. Schwartz, Clim. Change; 10.1007/s10584–010–9903–9 (2010). doi:10.1007/s10584–010–9903–9 CrossRef
  12. J. Hansen et al., Open Atmos. Sci. 2, 217 (2008).
  13. P. K. Bijl, A. J. Houben, S. Schouten, S. M. Bohaty, A. Sluijs, G. J. Reichart, J. S. Sinninghe Damst©, H. Brinkhuis , Science 330, 819 (2010). Abstract/FREE Full Text
  14. D. J. Lunt et al., Nat. Geosci. 3, 60 (2010). CrossRefWeb of Science