Permafrost Projections

During my summer at UVic, two PhD students at the lab (Andrew MacDougall and Chris Avis) as well as my supervisor (Andrew Weaver) wrote a paper modelling the permafrost carbon feedback, which was recently published in Nature Geoscience. I read a draft version of this paper several months ago, and am very excited to finally share it here.

Studying the permafrost carbon feedback is at once exciting (because it has been left out of climate models for so long) and terrifying (because it has the potential to be a real game-changer). There is about twice as much carbon frozen into permafrost than there is floating around in the entire atmosphere. As high CO2 levels cause the world to warm, some of the permafrost will thaw and release this carbon as more CO2 – causing more warming, and so on. Previous climate model simulations involving permafrost have measured the CO2 released during thaw, but haven’t actually applied it to the atmosphere and allowed it to change the climate. This UVic study is the first to close that feedback loop (in climate model speak we call this “fully coupled”).

The permafrost part of the land component was already in place – it was developed for Chris’s PhD thesis, and implemented in a previous paper. It involves converting the existing single-layer soil model to a multi-layer model where some layers can be frozen year-round. Also, instead of the four RCP scenarios, the authors used DEPs (Diagnosed Emission Pathways): exactly the same as RCPs, except that CO2 emissions, rather than concentrations, are given to the model as input. This was necessary so that extra emissions from permafrost thaw would be taken into account by concentration values calculated at the time.

As a result, permafrost added an extra 44, 104, 185, and 279 ppm of CO2 to the atmosphere for DEP 2.6, 4.5, 6.0, and 8.5 respectively. However, the extra warming by 2100 was about the same for each DEP, with central estimates around 0.25 °C. Interestingly, the logarithmic effect of CO2 on climate (adding 10 ppm to the atmosphere causes more warming when the background concentration is 300 ppm than when it is 400 ppm) managed to cancel out the increasing amounts of permafrost thaw. By 2300, the central estimates of extra warming were more variable, and ranged from 0.13 to 1.69 °C when full uncertainty ranges were taken into account. Altering climate sensitivity (by means of an artificial feedback), in particular, had a large effect.

As a result of the thawing permafrost, the land switched from a carbon sink (net CO2 absorber) to a carbon source (net CO2 emitter) decades earlier than it would have otherwise – before 2100 for every DEP. The ocean kept absorbing carbon, but in some scenarios the carbon source of the land outweighed the carbon sink of the ocean. That is, even without human emissions, the land was emitting more CO2 than the ocean could soak up. Concentrations kept climbing indefinitely, even if human emissions suddenly dropped to zero. This is the part of the paper that made me want to hide under my desk.

This scenario wasn’t too hard to reach, either – if climate sensitivity was greater than 3°C warming per doubling of CO2 (about a 50% chance, as 3°C is the median estimate by scientists today), and people followed DEP 8.5 to at least 2013 before stopping all emissions (a very intense scenario, but I wouldn’t underestimate our ability to dig up fossil fuels and burn them really fast), permafrost thaw ensured that CO2 concentrations kept rising on their own in a self-sustaining loop. The scenarios didn’t run past 2300, but I’m sure that if you left it long enough the ocean would eventually win and CO2 would start to fall. The ocean always wins in the end, but things can be pretty nasty until then.

As if that weren’t enough, the paper goes on to list a whole bunch of reasons why their values are likely underestimates. For example, they assumed that all emissions from permafrost were  CO2, rather than the much stronger CH4 which is easily produced in oxygen-depleted soil; the UVic model is also known to underestimate Arctic amplification of climate change (how much faster the Arctic warms than the rest of the planet). Most of the uncertainties – and there are many – are in the direction we don’t want, suggesting that the problem will be worse than what we see in the model.

This paper went in my mental “oh shit” folder, because it made me realize that we are starting to lose control over the climate system. No matter what path we follow – even if we manage slightly negative emissions, i.e. artificially removing CO2 from the atmosphere – this model suggests we’ve got an extra 0.25°C in the pipeline due to permafrost. It doesn’t sound like much, but add that to the 0.8°C we’ve already seen, and take technological inertia into account (it’s simply not feasible to stop all emissions overnight), and we’re coming perilously close to the big nonlinearity (i.e. tipping point) that many argue is between 1.5 and 2°C. Take political inertia into account (most governments are nowhere near even creating a plan to reduce emissions), and we’ve long passed it.

Just because we’re probably going to miss the the first tipping point, though, doesn’t mean we should throw up our hands and give up. 2°C is bad, but 5°C is awful, and 10°C is unthinkable. The situation can always get worse if we let it, and how irresponsible would it be if we did?


45 thoughts on “Permafrost Projections

  1. Thank you for a good summary. The expression “oh shit” folder is a really good. I found this citation that I use in order to keep my mental state from sinking even more
    “It’s far too late and things are far too bad for pessimism.”

  2. Is there a typo in “… and people followed DEP 8.5 to at least 2013 before stopping all emissions …”? That’s next year. Without reading the original, I’m guessing that’s supposed to be 2130.

      • I’ll try to read the paper – right now the link returns a “we’re sorry for technical difficulties” message. Now I’m guessing that the labels for the DEP scenarios are based on the total emissions rather than the rate. But I’m guessing in the dark right now.

      • Thanks for trying to explain this. It took the longer explanation with diagrams that Andy posted at SkS, but I finally got it. “Oh shit” indeed.

  3. This paper also went into my mental “oh shit” folder, because my thoughts while reading the description of it were, “Where’s the CH4?” You write, “the paper goes on to list a whole bunch of reasons why their values are likely underestimates.” Do the authors also go on to “estimate” the “effects” of these “underestimates” on their readers and the public? And do the authors go on to “estimate” the “impact” of these “underestimates” on TransCanada’s Keystone XL pipeline?

    For example.

    How will the underestimated findings in the authors’ incomplete paper effect the Obama/Romney debates related to the Keystone XL pipeline that’s being rammed down our throats? Keystone XL proponents (whose lawyers or their lackeys will certainly have read the incomplete paper) will argue that climatologists who can’t get their models right can’t be trusted to make climate predictions.

    Forget Romney. His decision regarding Keystone XL was made years ago, but Obama presumably hasn’t decided yet.

    “The Keystone XL tar sands pipeline could become a point of contention during the presidential debate in Denver this Wednesday. Mitt Romney has pledged that he would approve the pipeline on Day 1 of his administration, opening the spigot for 900,000 barrels of the world’s dirtiest oil to flow down through the country’s breadbasket to the Gulf every day.

    “President Obama has not ruled out approving the pipeline. Last year, he ordered a Supplemental Environmental Impact Study to be done on the risks presented by the pipeline, and the State department is again evaluating a route proposed by TransCanada, the pipeline’s owner and operator.”

    How about a poll?

    1. How many visitors to ClimateSight think that Romney or Obama or any of their staff members have read or will read the authors’ paper?

    2. How many visitors to ClimateSight think that any member of Congress or their staff have read or will read the authors’ paper?

    3. How many visitors to ClimateSight think that the majority of people who will read the authors’ paper are other climate scientists and TransCanada’s lawyers’ lackeys?

    4. How many visitors to ClimateSight think that this paper will accomplish much more than adding another “oh shit” to climate scientists’ mental “oh shit” folders?

    • And how many scientist are comfortable delivering underestimations of anything?

      An under-calculation, conservative bias, or any underestimation is bad science – or bad presentation of real data. To present a problem through rose-colored glasses serves no one well.

      And moreover, we should not expect politicians to be ruthless conveyors of reality. Few have, ever.

  4. Neither Romney nor Obama will read the paper. They’re not scientists.

    But there is a huge difference between them. When somebody who has read it discusses it with them, Obama will listen.

    I won’t be reading the paper either. It’s paywalled.

  5. Has anyone started working on how much of the permafrost will end up as methane? Particularly when the karst is underwater, or falls into water, methane is likely. As the permafrost melts more lakes are likely, depending upon topography. But we will also loose lakes, so not an easy project one well beyond me.

    • Plants sucked it up, transformed it into organic compounds like cellulose, and eventually died. Usually what happens next is the plants decompose and the carbon in their tissues turns back into CO2. But when they’re frozen into permafrost, they can’t fully decompose, so the carbon gets trapped.

      • OK. So every Arctic summer we should ship lots of vegetable matter from around the world into permafrost zones, spread it around in a very thin layer, and then during winter cause regional cooling with judicious release of aerosols, so as to form just enough new ice to trap that summer’s carbon layer. Then repeat this every year until the world climate control authority is satisfied. :-)

  6. Nice write-up, Kate, thanks.

    There is also a concern that the oceans may not keep absorbing CO2 as effectively as in the past, although I’m not sure where the science on that stands.

    Tipping point-wise, a big one that gets little attention is an increase in thermocline depth, important because it will give the T increase long-term persistence (and I suspect would be one way for ocean CO2 absorption to decrease). AFAIK there are no signs of it so far, though.

      ” The new measurements, which have not yet been published, suggest the densest waters in the world ocean are gradually disappearing and being replaced by less dense waters.”
      “The amount of dense Antarctic Bottom Water has contracted each time we’ve measured it since the 1970s,” said Dr Steve Rintoul, the voyage leader and oceanographer with the CSIRO and the Antarctic Climate and Ecosystems CRC. “There is now only about 40% as much dense water present as observed in 1970.”

      These observed changes aren’t direct observations of thermocline depth, but IMHO, if there’s less cold dense water on the bottom, there’s more warm water on top. Either the thermocline depth has increased, or the thermoincline, the gradient, has decreased. The details of how the shape and amplitude of the thermal profile are changing probably alter the impact on warming persistence.
      No doubt we will see “skeptical” arguments similar to those on climate sensitivity, that if we don’t know precisely which shade of grey we’re facing, then black must surely be white, and all uncertainty must favor benign outcomes.

  7. If we are going to talk about carbon feedbacks, we need to include clathrates and free methane capped by permafrost.

    Permafrost melt is resulting in permakarst formations leading to changes in local hydrology resulting in rapid erosion and slumping of carbon materials into rivers. These are fairly large blocks of material, so the process is faster than a layer by layer in situ melt. The result is large amounts of nutrients (including carbon) released suddenly. This tends to release methane capped by the permafrost.

    Changes in local hydrology are allowing permafrost to drain and dry. Cltahrates associated with the bottom of the permafrost then decompose. Arctic amplification means the Arctic is now warm enough to have storms with lighting, which sets the dry carbon material on fire, thereby releasing the CO2 from many layers at once.

    Slumping of carbon materials will back fill polar bear dens and reduce the back country hazard of confined spaces, so the process is not all bad.. : )

  8. Is the year in this part corect? This scenario wasn’t too hard to reach, either – if climate sensitivity was greater than 3°C warming per doubling of CO2 (about a 50% chance, as 3°C is the median estimate by scientists today), and people followed DEP 8.5 to at least 2013 before stopping all emissions (a very intense scenario, but I wouldn’t underestimate our ability to dig up fossil fuels and burn them really fast), permafrost thaw ensured that CO2 concentrations kept rising on their own in a self-sustaining loop.

    • Magnus, the question of whether we followed DEP 8.5 seems to be addressed by a recent post at RC and the answer was, sadly, in the affirmative. This confirms with what I recall, but I can’t back up my own vague recollections directly right now. Here’s the text of the RC post (#269 on the open thread) with link:

      “I don’t like the fact that the pre-release paper about ICPP RCPs happens to show RCP8.5 as the worst case emission scenario in that chart and that this scenario also happens to be the apparent reality in the International Energy Outlook from 2011. If the IPCC RCPs don’t include anything worse and if we are, in reality, on track for this worst case, then the other RCPs are wishful proposals more than anything else. That’s very bad. But that seems to be reality, too. I think I will use RCP8.5 as the ONLY scenario I use for thinking purposes until AFTER I see serious political action AND YEARS AFTER I see significant implementation already taking place. Until then, RCP8.5 is reality.”

      Comment by Jon Kirwan— 9 Oct 2012 @ 3:56 AM

  9. Kate – many thanks for this overview.
    The paper addresses the clear failure of previous efforts by incorporating warming due to permafrost CO2 outputs, so I’m wondering why it didn’t also incorporate the effects of the CH4 fraction of total carbon released ?

    In hopes that a subsequent paper may remedy this shortfall, one further seminal shift would seem worth including, being a case for permafrost GHG contribution on top of the warming from the best case of emissions control, specifically:
    – present realized warming,
    – plus pipeline warming,
    – plus warming from phase-out emissions reaching near-zero by 2050,
    – plus a multiplier for the consequent loss of the fossil sulphate parasol.

    While this would still of course exclude interactions with five of the other mega-feedbacks now accelerating, it would set the precedent of basing projections on the best case of the current strategy for mitigation, which would be highly informative of the gravity of our position.



  10. I’d like to know what was used for the degradation of the land sinks due to drought fires and bark beetles. Is it down 20% or more What is the total degradation used for the total land/ocean sinks? All for the year 2300 in your modelling. Great blog.

  11. “this model suggests we’ve got an extra 0.25°C in the pipeline due to permafrost. It doesn’t sound like much, but add that to the 0.8°C we’ve already seen, and take technological inertia into account (it’s simply not feasible to stop all emissions overnight), and we’re coming perilously close to the big nonlinearity (i.e. tipping point) that many argue is between 1.5 and 2°C.”

    Not wanting to add to the gloom but you didn’t mention the committed warming already in the pipeline due to Earth’s energy imbalance. Is there a reason for not factoring that in – most estimates I have seen put that at 0.6C but I am aware of higher values being suggested notably Ramanathan and Feng’s suggested 1.6C.

    As Schellnhuber stated in his commentary on Ramanathan and Feng,

    “However, the quintessential challenges remain, namely bending down the global Kyoto-GHG output curve in the 2015–2020 window (further procrastination would render necessary reduction gradients too steep thereafter) and phasing out carbon dioxide emissions completely by 2100. This requires an industrial revolution for sustainability starting now.”

  12. Kate – please could you clarify whether the authors have, as the RCP curves appear to show, adopted the UK MoD/MET Office notion of the carbon sinks increasing their efficiency to more than 100% of peak annual anthro-CO2 output post emissions control? The implied rates of decline of CO2e appear otherwise inexplicable. Yet in view of the entirely predictable erosion of sink capacity (by forest loss & combustion, soils’ desiccation, permafrost melt, and oceans’ warming and acidification) that notion appears to be an outstanding example of optimism bias.

    If the model were run with say Archer’s 2.7% methane output (giving a precise doubling of CO2e output with CH4 GWP set at 100 for the crucial 20yr horizon)
    and with the rational assumption of the carbon sinks’ efficiency declining throughout this century,
    what multiple of the authors’ finding of 0.25C to 1.0C warming from permafrost melt would you expect ?

    This is not to knock the authors’ efforts, but to try to put their explicit warnings of understatement in context. It is the first paper I’ve heard of that postulates that we are now committed to permafrost melt (as just one of the seven mega-feedbacks) under a minimalist assessment, generating a self-reinforcing CO2e increase regardless of an ‘overnight’ success at emissions control. I wonder if you would agree that its core implication is thus that sufficient geo-engineering (in both its albedo restoration & carbon recovery modes) mandated under globally accountable scientific supervision, is thus inevitably required as the necessary complement to emissions control for the resolution of global warming ?

    With regard to the discussion of ‘tipping point’ I don’t follow the idea of multiple events – once tipped, we’re tipped, surely ? As I’ve understood it from McGribben’s explanation back in ’89, the tipping point occurs when the combined CO2e output of all feedbacks exceeds the carbon sinks’ capacity, after which their outputs are inevitably adding to airborne GHG stocks and warming and are thus effectively self-reinforcing. Given that we must be near or past that point, with current albedo loss plus at least five other mega-feedbacks now accelerating, the nuanced definition of ‘tipping point’ being that at which we are committed to the feedbacks swamping the sinks in the future, now appears redundant. But perhaps I’m missing something ?



  13. Thanks for a great, if very depressing, post, and a wonderful site.

    If you look at the left-hand graph of figure 3 on the Skeptical Science post about this article, the dotted line shows that, with climate sensitivity of 3 degrees C (which seems to be the widely accepted value), CO2 concentrations in the atmosphere would remain at present levels indefinitely, even if we stopped all anthropogenic CO2 emissions next year.

    Since, as you and others point out, there are a number of other major reinforcing feedbacks that will be kicking in (if they haven’t already), can’t we conclude that, even if we stop emissions completely in 2013, CO2 levels, and so global temperatures, will continue to rise into the indefinite future (though not, of course, forever)?

    Isn’t that what is generally referred to as ‘runaway global warming’?

    Does this paper not prove, in other words, that we have already crossed the tipping point into runaway global warming, that warming will continue even if we remove all of the initial ‘forcing’ of all future human releases of carbon into the atmosphere? That this has happened well before global temps have reached 1.5 C above pre-industrial levels?

    Please tell me that I have misunderstood something.

  14. Lewis C, you mention “mega-feedbacks” in the last paragraph of your last post.When I tried to search the term, I kept getting other posts of yours. Is there a list of these somewhere? I am very interested in trying to get a thorough list of possible and probable feedbacks, ideally with their relative likely force and timing. Thanks ahead of time for any light you can throw my way.

    • John – I confess I coined the term ‘mega-feedbacks’ as a means of grouping the seven main dynamics and distinguishing them from the myriad internal ones and the interactions between the megas. I have to say that in decades of tracking public discussion and such papers as I can run to, I’ve seen scarcely anything looking at the potentials of the feedbacks as a whole. In terms of the science their impacts and interactions appear to be a black hole which, apart from a few brave types like the authors above, most would rather keep away from.

      In order of seniority, the seven feedbacks that seem outstanding are:
      Water vapour – rising by ~7% per 1.0C of warming;
      Albedo loss – due mostly to cryosphere decline;
      Microbial peat-bog decay – due to rising CO2 affecting ecological dynamics;
      Desiccation of tropical and temperate soils – due to SAT rise and droughts;
      Permafrost melt – due to SAT rise plus loss of snow cover, etc;
      Forest combustion – due to SAT rise, droughts, pest responses, etc;
      Methyl clathrates [aka methane hydrates] now threatened by rising sea- temperatures, increased water column mixing, etc.

      The first six were all already accelerating by the mid-’80s, with some now very well established – albedo loss was reported last year to be already imposing a warming equivalent to 30% of anthro CO2 outputs. As you’ll likely have seen, the clathrates issue is still controversial, with the report on last year’s joint expedition to the ESAS due out last May, and no explanation given of its delay, with nothing out of US participants and only terse vague statements from Russian media.

      One aspect that gets hardly any recognition of its relevance is the northward migration of increasing global precipitation – which is best depicted in Aiguo Dai’s NCAR plots of PDSI progression from 1950 to 2099 [PDSI: Palmer Drought Severity Index]. (As context in viewing these plots it is worth noting that the US dust-bowl was mostly at 3 to 4 on PDSI and only occasionally spiked to 6). That northward migration directly drives cryosphere decline & permafrost melt (by increased rainfall) and forest combustion and soils desiccation (by decreased rainfall) but as far as I know its progress is not yet incorporated into the modelling of their development.

      The minor feedbacks seem ubiquitous, and in including even microbial interactions seem unlikely to ever be fully charted. Some are predictable, such as a widespread growth of shrubs across vast former tundra areas shedding their leaves to be windblown into the increasingly prevalent themokarst melt pools and land-slip dam lakes, where they’ll rot anaerobically to release additional methane. But it seems predictable that many such feedbacks will appear as surprises, and some will have highly subtle mechanisms – identifying that for the global peat-bog decay took about 40years of study, but it yielded the information that if the trend of rising CO2 held until the 2060s, peat-bog decay would emit CO2 annually equal to the entire anthro-output of 2000. (That scenario is now defunct since the CO2 trend has risen and additional peat bog is thawing out).

      As far as I understand it, the central message of the feedbacks’ acceleration is that a strategy of mitigation via emissions control alone is not remotely commensurate with the scope of the problem we face. Since geoengineering by unilateral decision is profoundly objectionable internationally, the paramount need of an equitable and efficient global climate treaty is thus strongly affirmed.



      • Thanks tons, Lewis. I may want to get in touch with you, since I am thinking of writing a popular press book on feedbacks. I have also noticed a lack of attention to them, especially the carbon feedbacks.

        One more question for now: What do you mean by “in order of seniority”?

        Does that mean the ones that were first understood are first in the list, or the ones that you have judged to be the most powerful feedbacks?

        Thanks again,


  15. John – thanks for your kind response to my rather sketchy outline above.

    ‘Order of seniority’ referred only to the sequence in which the feedbacks began to accelerate – albeit imperceptibly at the time – with water vapour predictably having started its rise back in the C19.

    I’d be delighted to discuss a book with you, but its worth considering quite how much contextual info is needed to give a cogent picture of the feedbacks’ relevance. On top of which, there’s the issue of the need to discuss credible solutions if people are going to assimilate fresh problems. But it is definitely needed, no question.

    Perhaps Kate would be so kind as to let you have my email ? I’m a bit chary of posting it, as you can imagine.

    All the best,


    • Thanks, Lewis.

      I’m a bit more reckless ‘-)

      harkness at augsburg dot edu

      It will only be good for a few more months, though.

  16. In response to the comment in the article body:

    “However, the extra warming by 2100 was about the same for each DEP, with central estimates around 0.25 °C. Interestingly, the logarithmic effect of CO2 on climate (adding 10 ppm to the atmosphere causes more warming when the background concentration is 300 ppm than when it is 400 ppm) managed to cancel out the increasing amounts of permafrost thaw.”

    The plant response could be quite large to such a rise in CO2 levels, lowering the threshold of light needed for photosynthesis and boosting the max output of many species will lead to significant physical changes in the structural relationship of plants to themselves and their substrates on land and in water. Will small streams become blocked by vegetation and cause an increase in soil flooding and thus methane production? Will a complex of feedbacks lead to a jump in erosion rates, leading to much larger nutrient levels in freshwater and some areas of the ocean, to the point of releasing more CO2 and CH4 while blocking CO2 uptake by the ocean? Will there be more wildfire with the potentially higher fuel load, or perhaps a change in quantity of flammable resinous content? Will rapid transitions and species invasions lead to sudden changes in plant community balance, killing large numbers of trees or otherwise upsetting the local carbon cycle, and will those disruptions put more CO2 and methane into the air or remove it? And, will there be a large change in evapotranspiration and albedo? Will the changed physical structure of plants at local scales lead to more heat-trapping ability and less air-flow at the ground surface, possibly with an increase in humidity, and will that in turn lead to changes in methane production in submerged or even wet soils?

    Those are my thoughts at the moment.

Leave a Reply to pendantry Cancel reply

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

You are commenting using your 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.