I went to a talk yesterday by Mark Pagani (Yale University), on the role of methane hydrates in the Paleocene-Eocene Thermal Maximum (PETM). The talk was focussed on how to explain the dramatic warming seen at the end of the Paleocene, 56 million years ago. During the Paleocene, the world was already much warmer than it is today (by around 5°C), and had been ice free for millions of years. But at the end of the Paleocene, the tempature shot up by at least another 5°C, over the course of a few thousand years, giving us a world with palm trees and crocodiles in the arctic, and this “thermal maximum” lasted around 100,000 years. The era brought a dramatic reduction in animal body size (although note: the dinosaurs had already been wiped out at the beginning of the Paleocene), and saw the emergence of small mammals.
But what explains the dramatic warming? The story is fascinating, involving many different lines of evidence, and I doubt I can do it justice without a lot more background reading. I’ll do a brief summary here, as I want to go on to talk about something that came up in the questions about climate sensitivity.
First, we know that the warming at the PETM coincided with a massive influx of carbon, and the fossil record shows a significant shift in carbon isotopes, so it was a new and different source of carbon. The resulting increase in CO2 warmed the planet in the way we would expect. But where did the carbon come from? The dominant hypothesis has been that it came from a sudden melting of undersea methane hydrates, triggered by tectonic shifts. But Mark explained that this hypothesis doesn’t add up, because there isn’t enough carbon to account for the observed shift in carbon isotopes, and it also requires a very high value for climate sensitivity (in the range 9-11°C), which is inconsistent with the IPCC estimates of 2-4.5ºC. Some have argued this is evidence that climate sensitivity really is much higher, or perhaps that our models are missing some significant amplifiers of warming (see for instance, the 2008 paper by Zeebe et al., which caused a ruckus in the media). But, as Mark pointed out, this really misses the key point. If the numbers are inconsistent with all the other evidence about climate sensitivity, then it’s more likely that the methane hydrates hypothesis itself is wrong. Mark’s preferred explanation is a melting of the antarctic permafrost, caused by a shift in orbital cycles, and indeed he demonstrates that the orbital pattern leads to similar spikes (of decreasing amplitude) throughout the Eocene. Prior to the PETM, Antarctica would have been ice free for so long that a substantial permafrost would have built up, and even conservative estimates based on today’s permafrost in the sub-arctic regions would have enough carbon to explain the observed changes. (Mark has a paper on this coming out soon).
That was very interesting, but for me the most interesting part was in the discussion at the end of the talk. Mark had used the term “earth system sensitivity” instead of “climate sensitivity”, and Dick Peltier suggested he should explain the distinction for the benefit of the audience.
Mark began by pointing out that the real scientific debate about climate change (after you discount the crazies) is around the actual value of climate sensitivity, which is shorthand for the relationship between changes in atmospheric concentrations of CO2 and the resulting change in global temperature:
The term climate sensitivity was popularized in 1979 by the Charney report, and refers to the eventual temperature response to a doubling of CO2 concentrations, taking into account fast feedbacks such as water vapour, but not the slow feedbacks such as geological changes. Charney sensitivity also assumes everything else about the earth system (e.g. ice sheets, vegetation, ocean biogeochemistry, atmospheric chemistry, aerosols, etc) is held constant. The reason the definition refers to warming per doubling of CO2 is because the radiative effect of CO2 is roughly logarithmic, so you get about he same warming each time you double atmospheric concentrations. Charney calculated climate sensitivity to be 3°C (±1.5), a value that was first worked out in the 1950’s, and hasn’t really changed, despite decades of research since then. Note: equilibrium climate sensitivity is also not the same as the transient response.
Earth System Sensitivity is then the expected change in global temperature in response to a doubling of CO2 when we do take into account all the other aspects of the earth system. This is much harder to estimate, because there is a lot more uncertainty around different kinds of interactions in the earth system. However, many scientists expect it to be higher than the Charney sensitivity, because, on balance, most of the known earth system feedbacks are positive (i.e. they amplify the basic greenhouse gas warming).
Mark put it this way: Earth System Sensitivity is like an accordion. It stretches out or contracts, depending on the current state of the earth system. For example, if you melt the arctic sea ice, this causes an amplifying feedback because white ice has a higher albedo than the dark sea water that replaces it. So if there’s a lot of ice to melt, it would increase earth system sensitivity. But if you’ve already melted all the sea ice, the effect is gone. Similarly, if the warming leads to a massive drying out and burning of vegetation, that’s another temporary amplification that will cease once you’ve burned off most of the forests. If you start the doubling in a warmer world, in which these feedbacks are no longer available, earth system sensitivity might be lower.
The key point is that, unlike Charney sensitivity, earth system sensitivity depends on where you start from. In the case of the PETM, the starting point for the sudden warming was a world that was already ice free. So we shouldn’t expect the earth system sensitivity to be the same as it is in the 21st century. Which certainly complicates the job of comparing climate changes in the distant past with those of today.
But, more relevantly for current thinking about climate policy, thinking in terms of Charney sensitivity is likely to be misleading. If earth system sensitivity is significantly bigger in today’s earth system, which seems likely, then calculations of expected warming based on Charney sensitivity will underestimate the warming, and hence the underestimate the size of the necessary policy responses.
Nice post, but I don’t think you’ve stated the distinction between climate sensitivity and earth system sensitivity quite right. The example you give of state-dependence (whether you start from a high or low sea-ice cover) actually applies to climate sensitivity: sea ice cover can vary pretty quickly, and is therefore included in Charney climate sensitivity. The distinction is a little arbitrary, but has to do with both time scale, and the state of modeling in the 70’s-90’s, when stuff like surface vegetation and atmospheric composition changes weren’t explicitly handled in climate models.
I doubt that the (equilibrium) earth system sensitivity to atmospheric CO2 concentration can be formulated unambiguously. If the mass of continental ice sheet is considered an internal variable of the climate variable, CO2 concentration should be another internal variable rather than external forcing. Sensitivity to internal variables may be formulated in sophisticated frameworks of systems analysis, but the formulation is likely to dependent on the framework of analysis.
You raised two problems: not enough C in the clathrates, and implies-too-high-sensitivity. Switching to Antarctic permafrost doesn’t help the second point.
@William M. Connolley : Yeah, I said I needed to do more background reading. I think I must have missed something in the talk about this. I think it has to do with the fact that the clathrates theory involves a short term methane boost followed by a longer term CO2 boost, and some evidence that a lot of the carbon from the clathrates doesn’t end up as CO2, so when you add the methane and CO2 effects together you still need a big sensitivity to make the numbers add up. But I’m still not entirely clear how the permafrost hypothesis handles this. I need to read Pagani’s paper!
@William M. Connolley : Okay, I think I have a better handle on the story now, thanks to Jim Prall, who was paying more attention than I was. It’s largely explained by figure 2 in this paper, which admittedly is ridiculously complex. Basically, to close the loop, you have to deal with three quantities: the amount of carbon released, the dramatic shift in carbon isotopes, and the temperature change. Only the isotopic excursion is known with any degree of accuracy, although the IPCC estimates give us a plausible range for climate sensitivity. Methane hydrates have a very different isotopic signature to terrestrial organic carbon. To get the right isotopic excursion, you need either a small amount of Methane (1.5-3PgC) or a large amount of terrestrial organic carbon (say 10-20PgC). The former requires a correspondingly larger value for sensitivity (which takes us well outside the IPCC range), while the latter does not.
Yes, but. IPCC estimates of sensitivity do not apply to the PETM without reservation. Many things were different then (no polar ice, and different positions of the continents allowing a tropical ocean circulation, for starters). Also our estimates of the baseline P/E global temperature vary by 5+ degrees. The modellers can and do run models with approximate (and best-guess) PETM boundary conditions, but they understandably don’t get as much CPU time as the IPCC projections, which limits the constraints they are able to establish. Also, of course, the time resolution of our PETM observations is in the ky to 10s of ky at best, so we can only look at the equilibrium (hah!) sensitivity, not anything like Charney (so all the uncertainties about feedbacks apply in full).
Several presentations at the Royal Society’s “Warm Climates of the Past” meeting in October went through this and came up with some pretty extreme sensitivity numbers (depending, as you observe, on the carbon source): up to 9C/doubling. The most convincing presentation (I’m away from my notes, but it might well have been Pagani) had an initial bolus of maybe 1Pg carbon followed by a gradual release of maybe 4Pg over several 10s of ky. The initial bolus has one isotope signature and the rest has something else. I’m going to go out on a limb, without notes, and say permafrost followed by hydrates.