Earth System Sensitivity and things

There was a recent paper about the Evolution of global temperature over the past two
million years
. The reason it is controversial is not because it produces a global temperature reconstruction for the last 2 million years, but because it attempts to determine climate sensitivity; in particular the Earth System Sensitivity (ESS). This is essentially the climate sensitivity after the all feedbacks have responded, even the ones that operate on very long timescales.

The analysis in this paper suggests the ESS is around 9oC (range – 7 to 13oC) which, as this Realclimate post points, is wrong. The one pretty obvious problem is pointed out in this post by James Annan. The ESS is almost certainly state dependent. It will probably be higher if the initial state is one with large ice sheets (such as during a glacial period) than it will be during a period with small ice sheets (such as now). Therefore using the transition between a glacial and an inter-glacial will probably over-estimate the ESS for today.

There is a more fundamental problem, though. Determining the ESS by regressing CO2 forcing against temperature essentially assumes that changes in CO2 is the primary driver. As I think this Realclimate post is pointing out, the transition from a glacial to an inter-glacial is triggered by changes in orbital forcing; in particular a large increase in solar insolation at high Northern latitudes (although the net change in global solar forcing is small).

This is thought to then trigger a retreat of the ice sheets which then produces some warming, releasing CO2, changing vegetation, and changing circulation patterns. This then produces further warming, causing the ice sheets to retreat further, releasing more CO2, etc. In other words, it is complex situation with multiple drivers of warming and, hence, assuming that the primary driver is CO2 is probably wrong.

So, using changes in CO2 and changes in temperature to suggest an ESS of around 9oC is almost certainly wrong. However, there is another factor that doesn’t seem to be considered, that I wanted to discuss here (mainly because it confuses me slightly). If you look at projections for cumulative emissions and atmospheric CO2, doubling atmospheric CO2 (280ppm to 560ppm) would require emitting about 1200 GtC (about double what we’ve emitted to date).

According to Archer et al. (2009), the long-term enhancement in atmospheric CO2 will be equivalent to somewhere between 20% and 30% of our total emissions. Therefore if we do emit a total of 1200 GtC (and double atmospheric CO22), the atmospheric CO2 concentration will then tend (over hundreds of years) towards a long-term (millenial timescales) concentration of around 400ppm – 450ppm. So, if this is correct, the long-term warming is not determined by the maximum atmospheric concentration that we reach, but by the atmospheric concentration to which we will tend (somewhere between 400ppm and 450ppm if the maximum is around 560ppm). An ECS of 3oC would then suggest a committed warming of around 2oC, and an ESS of 5 – 6oC would suggest a committed warming of somewhere between 2.5 and 3oC (all numbers approximate and there are, of course, ranges for all quantities).

However, and this is where I get confused, there is another paper that consider permafrost carbon feedback, and which Andy Skuce discusses here. This paper suggests that even if we shutdown anthropogenic emissions after reaching 560ppm, the release of CO2 because of thawing permafrost would be sufficient to maintain atmospheric CO2 at the level when anthropogenic emissions ceased (assuming an ECS of 3oC).

This has become somewhat convoluted and I’m no longer quite sure what my point was. I had thought that we would expect atmospheric CO2 to drop if we were to halt emissions, but it seems that there is a chance that permafrost feedbacks could be sufficient to fix atmospheric CO2, even if we were to halt all anthropogenic emissions. I wrote this thinking that I had some point to make, but I’m no longer sure what that is. Maybe someone who understands this better than I do can clarify our current understanding as to what would happen if we were to halt all anthropogenic emissions.

This entry was posted in Climate change, Climate sensitivity, Global warming, Research, Science and tagged , , , , , . Bookmark the permalink.

24 Responses to Earth System Sensitivity and things

  1. Since I mentioned Andy Skuce in the post, I should also mention that he has published a nice article about climate sensitivity.

  2. David B. Benson says:

    “The Long Thaw” by David Archer.

  3. -1=e^iπ says:

    “Here we quantify the median γ as 7.7 p.p.m.v. CO2 per °C warming, with a likely range of 1.7–21.4 p.p.m.v. CO2 per °C.”

    temperature CO2 feedback just isn’t that strong.

  4. -1,
    I’m not sure that’s quite the same as was being suggested in the paper I mentioned at the end of the post. That was considering the continued release of CO2 from thawing permafrost even if we stopped emitting CO2. This doesn’t have to be very high to fix atmospheric CO2.

  5. Andrew Dodds says:

    By definition, many of the areas releasing CO2 under a +2 degree or higher scenario, will have been frozen for the past couple of million years or more; this means that results derived from areas frozen for just a few tens of thousands of years may not apply.

    And there is a rate effect; if permafrost degasses over a few thousand years, the effect on atmospheric CO2 levels will be smaller than if it degasses over a few hundred years or less.

  6. BBD says:


    You ain’t the only puzzled one. In an interview with Greg Laden, Dr Snyder said:

    [ESS] is likely state dependent, and thus I focused on comparing my new estimate to estimates from previous research on the late Quaternary. I also was able to investigate the state dependence of the metric within the last 800,000 years and found that it was lower in deep glacial states. This is an interesting finding, as some people have assumed that the ESS metric would be higher at the glacial maxima (e.g., the LGM) than during interglacials. That is not what the data shows.”

    My understanding too is that ESS is higher when the cryosphere is large. Possibly I have entirely misunderstood what Dr Snyder said.

  7. angech says:

    “Determining the ESS by regressing CO2 forcing against temperature essentially assumes that changes in CO2 is the primary driver.”
    I’m confused too.
    There are two types of climate sensitivity that seem to be mixed up here. The definition of ECS normally used is that of the response to a doubling of atmospheric CO2 as the forcing agent.
    But there is a general climate sensitivity ESS that is the response to any forcing increase or decrease.
    the Earth System Sensitivity (ESS) is essentially the overall response to the changes with solar variation both due to the sun,CO2, aerosols, inclination changes and milankovitch cycles etc.
    The ECS seems to be limited to the CO2 range near our current range in that some people suggest that increased concentrations may have little extra effect and lower levels presumably show a sharper response to doubling.
    What I am trying to say is that if the response is logarithmic then at 9 degree different temperature ranges [multiple CO2 doubling or halving] the general sensitivity may well be quite different to that at the current range.
    As you say when CO2 is not the primary driver ESS should not be determined by regressing CO2 forcing against the temperature.

  8. angech,
    No, the ECS is the climate sensitivity when only fast feedbacks are considered (i.e., the feedbacks that respond quickly). These are water vapour, clouds, lapse rate, etc. The ESS is the overall climate sensitivity, when all feedbacks have operated. This includes changes to the cryosphere, heating the very deep ocean, changes in vegetation and, possibly, changes in global circulation. So, the ESS is almost certainly larger than the ECS, but an estimate of around 9oC is almost certainly too large. There is a Hansen paper that suggested around 6oC, but this is – I think – also regarded as too high for the current climate state.

  9. izen says:

    ALL the metrics of climate sensitivity are arbitrary definitions. They focus on global surface temperature as another indicative metric. TCR, ECS and ESS all specifically include and exclude other feedbacks and factors according to timescales and a priori assumptions.

    Values for all three metrics can be estimated from models, from the irredeemably simple to the infinitely complex and observations. Diurnal, seasonal and volcanic impacts can provide evidence of the short term effects. Paleoclimate gives indications on longer timescales.
    That there is a wide range of results is a feature not a flaw. Climate sensitivity metrics are an instant snapshot, trying to capture the evolving behaviour of the climate in a numerical result. Like a police photo-fit, they are a limited description.

    I think the quibbles over whether 2 million years of paleoclimate can indicate a long-term sensitivity of 9degC may be missing the point. The issue of attribution to CO2 is mainly a matter of definitions.
    The 9deg C figure is a Bikini statistic. It reveals the envelope of change that the climate can undergo during a period when the basic geography was unchanged. Solar input varied little except for small effects on distribution as a result of orbital shifts,
    But conceals the inhospitable conditions that the spread of temperature covers. For most of that 2 million years, local conditions and the large variations would have made maintaining a stable civilisation, never mind developing one, extremely difficult.

    In that 2 million years of climate variation the CO2 level is a key feedback and forcing in the whole system. Climate change over 2 million years may seem like far too long a timescale to be relevant to present human concerns. But it would usually take such geological time to shift CO2 levels as human technology has done.
    9deg C and massive shifts in the climate may not be a ‘kosher’ climate sensitivity estimate, but it probably does represent the scale of change that human activity is having on the Earth system.

  10. angechangech says:

    “There is a Hansen paper that suggested around 6 C, but this is – I think – also regarded as too high for the current climate state.”
    Schmittner et al use a spatially more complete data set than Snyder and finds LGM-Holocene warming = 2.3C. Snyder finds LGM-Holocene warming = 6.2C. We have estimated that the LGM-Holocene warming is 4-5C, which led to our estimate of ESS sensitivity ~6C in our 2008 paper, which she references.
    “Snyder also uses the inferred record of global temperature to estimate equilibrium climate sensitivity including slow feedbacks, sometimes called ESS, suggesting that doubled CO2 (4 W/m2 forcing) would eventually cause global warming of 7-13C.”
    “the long term committed warming from today’s CO2 levels is a further 3-7ºC.”
    Hansen does not condemn her findings.
    I find it odd that the well defined Equilibrium Climate Sensitivity that we have been talking about is not even a true equilibrium after all, just an early best guess.
    The ESS range which would seem to be from 4 to 6 to 9 degrees from the 3 papers done on this seems to suffer from the same problem mentioned previously. There is no unanimity on the figure and the range of values is so great as to make predictions of future temperature rise so variable in time as to be most unuseful.
    Dare I ask BBD or others if there are graphs of sea levels anywhere for the corresponding time frames and if there is any correlation?
    Sea level could be a partial proxy for temperature on long scales such as this despite adjustments for glacier melting rebound there should be underlying trends.

  11. I find it odd that the well defined Equilibrium Climate Sensitivity that we have been talking about is not even a true equilibrium after all, just an early best guess.

    No, this isn’t really true. It’s defined in terms of fast feedbacks only. Also, if you consider past climate changes and use a forcing that is a sum all of the CO2 plus all the slow feedbacks (ice sheet retreat, vegetation, etc) then you get a decent estimate of the ECS. The ESS is harder because it is probably more state dependent and it’s difficult to decouple all the different drivers of warming. However, it is probably the case that the ESS is greater than the ECS.

  12. Dikran Marsupial says:

    “Dare I ask BBD or others if there are graphs of sea levels anywhere for the corresponding time frames and if there is any correlation?”

    Yes, I had mm resolution data, but I left them in my time machine, and forgot when I parked it.

  13. BBD says:

    Dare I ask BBD or others if there are graphs of sea levels anywhere for the corresponding time frames and if there is any correlation?

    Panel b:

    Source: Hansen & Sato (2012) figure 2.

  14. verytallguy says:

    I find it odd that the well defined Equilibrium Climate Sensitivity that we have been talking about is not even a true equilibrium after all, just an early best guess.

    I find it odd that someone who doesn’t understand the basic definitions of the science is so active in questioning the conclusions, rather than trying to learn more, but each to their own.

  15. -1=e^iπ says:

    @ ATTP – current rate of CO2 uptake is ~2 ppm per year. Even if global temperature increase were to accelerate to 0.04 C/year, this means you would need a CO2-temperature feedback of 50 ppm/C to offset the uptake, which is well above what is suggested by empirical estimates, by models, etc.

  16. -1,
    But you’re talking about uptake in the oceans and biosphere. What’s discussed in the paper at the end of the post is the release of CO2 via thawing permafrost, something that probably wasn’t relevant between glacials and inter-glacials, may not yet be doing much, but may do so as we warm.

    If we halt all emissions and there is no permafrost feedback, then CO2 levels would be expected to drop to a long-term level in which the enhancement is about 20-30% of our total emissions. However, continued emission at a relatively low level (~1 GtC per year) could fix atmospheric concentrations; that’s around 0.5 ppm/year.

  17. Jutland says:

    First time commenter here.

    Firstly, it strikes me that whether the ESS is 9 or so degrees or 6 or so degrees hides the scale of likely impacts. Six degrees now could be much worse than nine degrees back in the past.

    The ESS is higher when the cryosphere is large, but because the cryosphere is larger, it can also absorb consequences more easily. So, the consequence of a 9 degree ESS one a large cryosphere is that it simply becomes a smaller cryosphere. Species in the biosphere can redistribute themselves as the ice retreats. However, the consequence of a 6 degree ESS on an already-small cryosphere – such as the present day – will be that the cryosphere almost entirely disappears. Species cannot redistribute themselves so easily, extinctions will be more likely, and more ecosystems will be wrecked.

    Secondly, I’m unsure whether what we are doing today is really any different from what happened in the past. I understand that changes in paleoclimate were due to orbital changes rather than direct input of CO2 into the atmosphere. But surely these are both external drivers of change to an otherwise stable carbon cycle, in other words just versions of the same thing?

    In the past, the carbon cycle was in equilibrium, then orbital changes heated the S hemisphere ocean, prompting natural sinks to give up CO2 into the atmosphere, and hence causing temperature rise. Today, the carbon cycle was also (until 1750) in equilibrium, then humans burned a lot of million-year old carbon thereby introducing entirely new CO2 into the carbon cycle; this is prompting or will prompt natural sinks to give up CO2 into the atmosphere, and hence causing temperature rise. So, the situation is pretty much the same. Have I understood that correctly?

  18. angech says:

    Thanks BBD
    Verytallguy., only learn by asking, thanks for your concern.
    ATTP thanks any time span for said fast feedbacks. Are we talking decades or millenia?

  19. Dikran Marsupial says:

    Angech “Verytallguy., only learn by asking”

    but you don’t just ask, you denigrate.

    “I find it odd that the well defined Equilibrium Climate Sensitivity that we have been talking about is not even a true equilibrium after all, just an early best guess.”

    is not a question. It is a shame you can’t take VTG’s (rather moderately expressed) comment more seriously.

    Of course it isn’t “just an early best guess”. People have been estimating CS for over 100 years and practically all science is about “best guesses” (“proof is for mathematics and alcohol”) and as ATTP explained it is “true equilibrium” (as far as that plausibly applies to the real world).

  20. Jutland,
    Welcome. The point, I think, is that the change from a glacial to an inter-glacial has multiple drivers of warming (small global, but large local, changes in insolation, changes in ice sheet, changes in atmospheric CO2, changes in circulation) so it’s hard to say that one of those is the prime driver (i.e., if it is all that changes, would the outcome be the same?).

    One thing I was trying to get at in the post was that it’s not clear what our long-term atmospheric concentration is likely to be (and hence, what the long-term warming will be). The standard carbon cycle models suggest that it would start reducing if we stopped emitting, eventually settling to a concentration enhanced by an amount equivalent to 20-30% of our total emissions. On the other hand, there are suggestions that release from thawing permafrost could stablise concentrations at the level when we stopped emitting. These two scenarios produce quite different long-term warming.

    Fast feedbacks operates on timescales of years. The figure below illustrates all the timescales.

  21. Here’s an article about thawing permafrost.

  22. Jutland says:

    ATTP, thank you, yes I think I see that now, having mulled it over a bit more. It’s not the simple first-one-step-then-another-step process that I had imagined.

  23. verytallguy says:

    On timescales and definitions, it’s worth pointing out that whilst things like ice sheets take millenia to come to equilibrium, we can already measure the start of that process now.

    Equally, water vapour is almost instantaneous, days or weeks perhaps?


    I was amazed when I first leaned that, GRACE also “weighs” aquifers. It’s incredible what we can measure, but even more amazing that we choose not to act on it.

  24. raypierre says:

    The linear relation between cumulative carbon and climate — which has as a corrollary that the temperature you have when you reach zero CO2 emissions is what you’re stuck with for the next 1000 years — is based just on ocean inorganic CO2 uptake. Land CO2 uptake can mess this up in many ways. In particular, if warming causes land ecosystems to become a net source rather than net sink of carbon, then the carbon added to the system from outgassing from land will cause warming to increase even after anthropogenic CO2 emissions cease. This is one of the biggest of the known unknowns in the climate system, and also one of the reasons it’s so important to keep the cum carbon down as low as possible, since the lower our cum C, the less the chance that land carbon will start outgassing.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.