Arguments
Climate inertia
Posted on 11 August 2016 by Bart Verheggen
This is a re-post from Bart Verheggen's blog
Imagine you’re on a supertanker that needs to change its direction in order to avoid a collision. What would you do? Would you continue going full steam ahead until you can see the collision object right in front of you? Or would you try to change course early, knowing that changing a supertanker’s course takes a considerable amount of time?
The supertanker’s inertia means that you have to act in time if you wish to avoid a collision.
The climate system also has a tremendous amount of inertia built in. And like with the supertanker, this means that early action is required if we want to change the climate’s course. This inertia is a crucial aspect of the climate system, both scientifically but also societally – but in the latter realm it’s a very underappreciated aspect. Just do a mental check: when did you last hear or read about the climate’s inertia in mainstream media or from politicians?
The inertia of the climate system could be compared to that of a supertanker: if we want to change its course, it’s important to start steering the wheel in the desired direction in time.
Why is it so important? Because intuitively many people might think that as soon as we have substantially decreased our CO2 emissions (which we haven’t), the problem will be solved. It won’t, not by a very long shot. Even if we reduce CO2 emissions to zero over a realistic timeframe, the CO2 concentration in the atmosphere – and thus also the global average temperature- will remain elevated for millennia, as can be seen in the figure below. The total amount of carbon we put in the atmosphere over the course of a few hundred years will affect life on this planet for hundreds of thousands of years. And if we want to reduce the amount of warming that we commit the future to, we need to reduce our carbon emissions sooner rather than later. The longer we postpone emission reductions, the stronger those emissions reductions would need to be in order to have the same mitigating effect on long-term warming.
That’s why climate inertia is so important.
Modeled response of the atmospheric CO2 concentration (panel b) and surface air temperature compared to the year 2000 (panel c) to prescribed CO2 emissions (panel a). The CO2 concentration remains elevated long after CO2 emissions have been reduced, because the long-term sinks for CO2 operate very slowly (see e.g. IPCC FAQ 6.2 for an explanation of these carbon sinks). Since CO2 impedes infrared heat loss, for millennia the globe will remain warmer than it was before CO2 concentrations rose. The temperature lags behind the CO2 concentration because of the time it takes for the oceans to warm up. Figure from Zickfeld et al (2013).
As I wrote before: Postponing meaningful mitigation action until the shit hits the fan comes with considerable risk, because many changes in climate are not reversible on human timescales. Once you notice the trouble, it’s only the beginning, because of the inertia in the various systems (energy system, carbon cycle and climate system). The conundrum is thus that those who caused the problem are in the best position to solve it, but since the full consequences will not materialize until much later, they have the least incentive to do so.
Over at Bits of Science two Dutch science journalists, Rolf Schuttenhelm and Stephan Okhuijsen, published an interesting piece that focuses on the same issue: we only see a portion of the warming that we have committed ourselves to, due to the thermal inertia provided by the oceans. Just as a pot of water doesn’t immediately boil when we turn on the stove, the oceans take time to warm up as well. And since there’s a lot of water in the oceans, it takes a lot of time.
They included the following nifty graph, with the observed surface temperature but also the eventually expected temperature at the corresponding CO2 concentration (which they dub the ’real global temperature’), based on different approaches to account for warming in the pipeline:
Observed and eventually expected (“real”) temperature at concurrent CO2 concentration, via Bits of Science
This is a nice way to visualize the warming that’s still in the pipeline due to ocean thermal inertia. From a scientific point of view the exact execution and framing could be criticized on certain aspects (e.g. ECS is linearly extrapolated instead of logarithmically; the interpretation that recent record warmth are not peaks but rather a ‘correction to the trend line’ depends strongly on the exact way the endpoints of the observed temperature are smoothed; the effect of non-CO2 greenhouse gases is excluded from the analysis and discussion), but the underlying point, that more warming is in store than we’re currently seeing, is both valid and very important.
Timescales, timescales, timescales. Why art thou missing from the public discussion about global warming?
Great article though somewhat terrifying, since it is very hard to quantify additional risks posed by positive|negative feed back loops for carbon sinks and possible short-lived methane contributions. There may be some hope in drawing down CO² levels by enhanced weathering or other efforts (at the scale of current global military spending), but social inertia stymies the kind of efforts we need to cut emissions and energy waste, develop sustainable energy alternatives, and draw down CO² at the scale required by the risks bearing down on us.
(I almost filed this one under the insert tab, mistakenly fixated on the inert response to risks we prefer to think of as tentative.)
I'm a bit confused, and need help understanding. Hansen's and other articles say that there is one equilibrium average global surface temperature based on CO2 concentration, and that it is ~3.0C (2.5 - 4.0) rise for every doubling of CO2 above a base concentration. This can be expressed as Teq = 3.0 x (Ln(CO2_new/CO2_base) / Ln(2)). If I apply this math to 400ppm (new) and 290ppm (base), I get 1.39C rise, which is the brown line in the lower chart. I understand that this is the average global surface temp increase, and that it would be more in the higher latitude regions, etc. And, I understand that this is the equilibrium temperature and that it will take ~135+/- years to reach full equilibrium based on a SkS chart I saw several months ago (from that I surmised that the 1st-order process lag is about 45 years); although another SkS article suggested this lag time is much less.
If I do the calculus, I get a the following crude model expression: Temp = Teq (above) x [1 - exp (- time_yrs/45 years)]. If I apply this expression to actual temps since 1955 (LINK), this fits very well, in terms of hitting the actual annualized or 30-year avg temps (lines in above chart). One could go on & find the best fit 'gain' multipler and 'lag time constant' using SOLVER for the minimum sum of squares of error; a project for another day.
All the above concepts are simple, straight forward and easy to understand. But, this article introduces two new temperatures (much higher than the Tequil that I've learned about so far). I don't understand what the RED and gray region temps represent, and how they are are applicable to benchmark temperatures like UNFCCC's agreed +2.0C limit; or Hansen's chart in 'Storms' that shows that the Antarctica ice formation started to develop at +4.5C. Are these RED and gray region temperatures on a different scale (apples-to-oranges) vs these latter benchmark temperatures? Are these RED and gray temps actually applicable to apples-to-apples comparison with the temperatures used in the bulk of scientific documents.
Bottom-line, what exactly do the RED and gray region temps represent in a tangible sense and in comparison to the normally discussed brown curve temps? Thank you!
sauerj @2, there are three climate sensitivity values it is helpful to know about.
First, the Transient Climate Response (TCR) is the temperature in the 70th year after when the only change in forcing consists of a 1% increase in CO2 levels per annum. The 70th year is when CO2 levels double under that condition. TCR is approximately 1.5 C per doubling of CO2, or 0.4 C/(W/m^2). It is important because it closely approximates to the immediate temperature response to forcing when forcing is increased more or less steadilly. It corresponds, more or less, to the red line in the graph above.
Second, the Equilibrium Climate Sensitivity (ECS) or Charney Climate Sensitivity is the temperature increase following a doubling of CO2 that is experienced once radiative/convective equilibrium is reestablished, but with no changes to land cover, or ice sheet extent as a result of feedbacks. It is approximately 2.8 C per doubling of CO2, or 0.76 C/(W/m^2). It is also the temperature we will achieve in 50-200 years if we increase CO2 concentration to a given level and hold it constant. The above graph is confused in that it purports that that is the brown line, but at ECS thermal equilibrium in the ocean has been established, so it should also be the yellow line.
Finally, if you allow forests to become deserts (or vice versa), and ice sheets to melt, you get the Earth System Sensitivity (ESS). That is not well constrained, but is likely 33% or more greater than the ECS. The grey shaded area is the ESS as determined from a particular period. Again, it is only relevant if we hold CO2 concentrations constant at their peak value.
In practise, if we should bring net anthropogenic emissions to zero, the ocean and chemical weathering will absorb CO2 at a rate that approximately balances the rate at which temperatures approach, first the ECS and then the ESS. The result is an approximately constant temperature near the TCR. Therefore with rational climate policy, ECS and ESS are more or less irrelevant. Even a continuation of 10% of emissions, however, will hold CO2 levels constant in the short term (up to 200 years) and rising in the long term so that temperatures will increase to ECS and then ESS values. Indeed, greater than ESS values because of the rising CO2 levels. That is why zero net emissions within the next 50 years is a must for any sensible climate policy.
Tom @3. That helps; I think I get it.
TCR would be like the real temp rise (I assume that TCR could also be based on %increases CO2 other than 1%/annum as we are now increasing in the 0.5%/annum range). The TCR would be like my crude model mentioned above in @2 [Tactual = fx(CO2)], which, yes, with the right constants, would hit the RED line.
ECS is the Tequil temp. The point at which global temperature has reached the point where Qin = Qout.
And ESS is the fully integrated model with all the real-life biosphere feedbacks played out. ... But, I had always assumed that these feedbacks (all of them ... stuff like warming oceans releasing more CO2, higher humidities adding to GHG's, ice sheet albedo changes, land cover impacts, etc.) were already packed into the ECS sensitivity curve. Was that assumption wrong? Note, many SkS articles clearly indicate a sensitivity # in the range of 2.5 - 4.5C. Is this range of sensitivity #'s only referring to ECS #'s and ESS #'s are even higher?
Note: To hit the lowest point of the grey region (2.1C on a 400/290 ppm rise), the gain# would be 4.5C; and to hit the midpoint (2.6C), the gain # would be 5.6C. Maybe the range indicated on the SkS articles did go up to 6.0C (?); but I do remember that the most probable value was ~3.0C.
So I'm a little confused on the discrepancy on past SkS articles on this sensitivity # compared to this ESS curve. Note: If it's too difficult to explain this, don't worry about it; its not that important that I understand this detail. A sensitivity # of 3.0C at our current trend of carbon emissions is plenty bad enough!!!
sauerj @4, I have responded to your comments by subject rather than by where you located the discussion. In particular, some of the points you raise under ESS are more correctly related to ECS, and so I have discussed them under that heading:
TCR: The TCR is strictly defined only for the 1% increase per annum experiment. It approximates to the real world temperature increase because most of the initial response to a forcing occurs in the first few years and the net forcing is variable in a short to medium time span (<10 years) due to the solar cycle, volcanic activity, and the effects of ENSO and other oceanic oscillations on albedo and water vapour concentrations in the atmosphere. If, however, you had a steady increase of CO2 concentration of 0.5% per annum, the temperature achieved at the 70 year mark would be higher than that predicted from an expectation based on the TCR.
Going into more detail, I compared the BEST LOTI temperature data to a prediction based on the sum of forcings from Kevin Cowtan's two box model of global temperatures (default settings). The correlation was 0.855, the Root Mean Squared Error was 0.176, and the linear trend was 0.914 of the observed values. In a monotonic 0.5% increase we would expect the linear trend to be greater than the observed values, so the perturbations more than compensate for the difference. For what it was worth, weighting the TCR prediction based on the difference in trend (ie, eliminating that difference) increased the RMSE by 0.001, decreased the mean error (which was still negative) but increased the standard deviation of the errors.
Given that, using TCR plus ENSO as a predictor of temperatures gives a reasonable approximation, but an approximation only.
ECS: Correct, with the provisio that in the approximately 200 year time frame to reach the ECS, there will have been some albedo change from changes of ice sheets and vegetation cover - so the balance will not not be perfect. The assumption is that the change small over that time frame, so the approximation is good enough on that time scale.
As an aside, the 1.5-4.5 C range for ECS is the likely (66.6%) certainty range. According the the IPCC AR5, the probability that ECS is 1 C is less than or equal to 5%. The probability that it is less than 6 C is greater than or equal to 10%. And assuming the probabilities are symmetrically distributed, the probability that it is less than 1.5 C is less than or equal to 16.7%, while the probability that it is less thanr than 4.5% is greater than or equal to 83.3%. Here are three Probability Density Functions (PDF) that more or less satisfy those conditions:
Strictly Roe and Baker (2007) is a PDF from the time of AR4 (although it better satisfies the AR5 constraints than the AR4 constraints). However, its high probability of low values comes at the expense of substantial probabilties of very high values of ECS. Rogelj et al (2014) is a representative PDF pubished in the scientific literature, and represents a best fit to the above constraints using a log normal curve. In accomplishing the good fit, however, it violates the requirement that the probability that ECS is less than than 4.5 C be greater than or equal to 83.3%. The alternate has adjusted values to ensure compliance with the above conditions, but as a result its threshold values diverge more from the stated limits than does Rogelj et al. For what it is worth, here are the 95% confidence range, mean, median and modes for the three distributions:
Value Alternate Rogelj Roe & Baker
2.5% 1.8 0.9 1.1
97.5% 7.8 7.8 29
Mean 3.03 3.07 2.66
Median 2.56 2.53 2.20
Mode 1.85 1.74 1.65
You will note that the median value (the 50:50 split) is around 2.5 C for all three distributions, and the mode below 2 C. The more reasonable climate skeptics are not wrong in expecting these lower values as the reasonable expectations (although they tend to artificially deflate the probability of higher values). In risk assessment, however, the mean value is more relevant. To exclude reasonable and dangerous possibilities from consideration, just because their probability of occuring is only 6.5% (6+ C using the alternate PDF) is unreasonable, but that is what concentrating on the median (or still worse, the mode) does. On the other hand, in popular discussion of climate risks, many people who want to take action forget that most probably (61.6% chance on alternate), ECS will be below 3 C. Of course, all of these PDFs give an artificial precession to the probability estimates.
ESS: In order to avoid overwhelming complexity, climate models do not typically vary ice sheet extent, and land cover (other then anthropogenic changes introduceds as forcings). That is justified because of the long time spans required for the melting of ice sheets. Using ESS for short term estimates is merely an artificial, and IMO misleading way of inflating percieved risk.
Tom @5. Thank you for taking the time to explain all of this to me. I understand everything you wrote. I checked out the Dr. Kevin Cowtan model and was thoroughly impressed (very easy to use and understand). Of course, I am ignorant on the details of the equational differences between the 1-box, 2-box and 3-box model variations, but that's OK. Someday, if I keep reading, I might understand what this means.
It was interesting how I could set the weighting of, say, 'Solar' to zero, just to see the difference to the 'fit'. One interesting tid-bit, the 'H2O(strato)' line is hidden exactly behind the 'BlackCarbon(snow)' line, in the forcing charts (I noted that there was 10 lines in the legend, but only 9 in the charts). But, then, when I changed the weighting of the 'H2O(strato)' line, then it appeared. One question: Is it correct to assume that rising humidities ['H2O(troposhere)'] (which would occur as a natural consequence of rising temperatures and a significant positive feedback component) is packed in the 'GHG(mixed)' line?
I was about to ask about RCP, but then found the SkS 3-part post (HERE) on that subject. I skimmed thru the whole thing; and need to return and read it in detail. This in-depth article looks amazing, and something that a person like myself should read in order to take the next appropriate step in learning.
I am a chemical engineer & very active in the local CCL chapter, which I think is the best thing out there for remediation vision, spirit and policy. ... Thanks again for your time!
sauerj @6:
1) The difference between the 1 box, 2 box, and 3 box models comes down to the number of time constants used. If you want more details, I believe the model is discussed in the SkS online course, which started Aug 9th but which you can probably still enroll in. Alternatively, Kevin Cowtan has in my experience always been helpful to those interested in learning more. Finally, And Then There's Physics gives the equations for a two box model here. Others have done similarly. No doubt there will be small changes in the exact form of the equations from model to model.
2) n-box models such as Kevin Cowtan's do not model feedbacks directly, and certainly not as a component of any of the forcings. Rather, the feedbacks along with thermal inertia are handled by a feedback constant (see ATTP's first equation). As a result such models are useful for giving emperical estimates of TCR and ECS, but do not demonstrate the physics.
3) I agree that the tunability of the forcings is one of the best features of Kevin Cowtan's model. I think anybody trying to argue that "it's the sun" or some other such theory should really adjust the weightings of that model to match their theory and show us why the resulting, poor fit is preferable to the good fit from the default settings.
Glad I could help.
Tom @7: I understand everything you have written and all the math on the ATTP site, which would ultimately work down to Ts = fx(F) and he explained that a singular net 'F' comes from the RCP11 dataset. All very interesting.
Thanks for explaining the Ffeed term (feedback component). I jumped to the ATTP site right away & read it and was puzzled by the Ffeed term. Then, I read your 2nd paragraph; your explanation of Ffeed was timely. And, thanks for explaining how the Ffeed term is developed here; thus explaining that delta-forcing due to delta-humidities are not modeled directly. That's likely OK and doesn't hurt accuracy as Fd-humidity is likely near linear (w/ respect to Ts) in the temperature ranges that we are talking about.
Also, thanks for explaining the n-box detail. I assume the different time constants on Cowtan's 'model tool' correspond to the surface layer for box-1 (constant=1/yr) and the ocean layer for box-2 (constant=30/yr). And, I assume the 3-box model breaks the ocean layer into two separate boxes: possibly into 1) an upper ocean layer for its box-2 (30/yr) and 2) a deep ocean layer for its box-3 (100/yr). The 1:100 ratio here (box-1/box-3) was the same ratio used by ATTP for the heat capacity ratio.
And, thanks for heads-up on the SkS course. I did enroll for it and will start watching the course videos. You've been very helpful. Thank you!
Hi friends, I'm a noob when it comes to climate science, but I know enough to challenge "deniers" 99% of the time. Recently I was talking with someone and they mentioned ECS, which I never heard of until now. This is the only page I can find with ECS. Based on talks with them, the overall argument is that "there are credible scientists who debate the ECS values. The IPCC best guess is around 3.2, but 'Deniers' see the ECS as being significantly lower than 3.2. Likely 2.0 or under."
It seems this person is claiming that whether AGW is dangerous or not, depends on this value, and this value is what credible skeptics are debating against. I received this message from him and I cannot seem to find info to verify/reject what it asserts:
(begin message)
ECS science from what I know comes from 3 sources. Paleoclimatology, Modelling and observations. Observations seems to provide the lower end results, the others higher.
ECS is basic physics of 1.1 and feedbacks which are estimated to be 0.4 to 3.4. The most important feedback is water vapor.
This basic information has been the same for 60 years. So a low ECS has been around from the start and still persists.
One link about low ECS is Link they provided to Judith Curry
Your point about time is correct. Projections are based on 2 inputs. ECS and RCPs (CO2 estimations) Much of the new articles assume high for RCP (8.5), but that is extremely improbable.
Correct about IPCC's estimation and the rationale for supporting it. However, observations are significantly lower than model means. It is hard to justify this after so long, and the divergence is getting worse not better (depending on data sets). This, plus failed basic tests of the hypothesis are the evidence deniers have. This evidence suggests that nature is playing a bigger role than most people think.
Failed tests
- lack of tropical hotspot, expected from warming
- no warming in the antarctic (both poles should warm a lot more than other places)
(You can find all sorts of rebuttal material, but the facts are pretty basic, these are traits that will happen is a warming world..... unless nature can overcome it).
(end message)
Any help on analyzing this, or sources to help a noob like me understand what it's all about and how to verify its claimes is much appreciated.
During our discussions, I couldn't shake the feeling that there was something deceptive about this, so I want to learn more to make sure I'm not being taken for a ride.
[TD] Regarding the "tropic hotspot," see "There's no tropospheric hotspot." After reading the Intermediate tabbed pane, you might want to read the Advanced one.
Regarding Antarctic warming, see "Natural forces overpowering Antarctic Peninsula warming."
There is a lot on SkS about ECS, but here is a recent one: "Climate sensitivity is unlikely to be less than 2C, say scientists."
[PS] You might also like to consider that to prove science wrong, you have to show observations contradict a projection that the science actually made. You might think Antarctica is "basic" but science doesnt. Please cite published predictions of equal warming. Ditto for your expectations of surface warming. Dont trust what deniers tell you about what science predicts - check it yourself.
Thanks TD, I'll take a look at those sources and see how his claims hold up.
A question about posts: Is there a web location to go to so I can keep track of comments I post in comment sections? It took me a while to find this page again. I've bookmarked it now, but is there something that directs me to my comments which have received replies? I'm new to the site and I just don't know.
Thanks again.
Climate Noob:
In the middle of the menu bar, just below the Skeptical Science masthead, is a link to "Comments". That will load a page that has all new comments on all threads/topics. Easy to see what people have been saying regardless of where it is. The number of new comments here each day is fairly small, so you should find what you want on the first page (or two, at most).
Climate Noob @9, @5 above I discuss the differences between the Transient Climate Response (TCR), Equilibrium Climate Sensitivity (ECS), and Earth System Response (ESS). I also have a detailed discussion of the likelihood estimates of the values of the ECS in the IPCC AR5. Looking at it in detail is very interesting, for it shows that according to the IPCC the most likely result is an ECS below 3 C, and indeed that the most likely single value (mode) may be below 2 C. That is important if you are prone to give in to councils of despair given the plethora of bad climate news around.
Importantly, however, the mean values of the ECS estimate are almost certainly just above 3 C. In cost benefit analysis, it is typically the mean value that is the best indicator of the likely cost of the scenario. If costs increase more than linearly with increasing temperature, the best indicator in a cost benefit analysis woud be from values above the mean. It is somewhat like Russian Roulette. The most likely value from a single round of Russian Roulette is no bullet in the head, but that one in six chance of blowing your brains out pretty much always makes Russian Roulette a round of Russian Roulette a stupid play. Likewise with global warming, the most likely result has a moderate ECS and a reasonable probability of moderate harm; but the significant (if small) probability of truly catastrophic harms means if we are smart, we don't play the game, ie, we cease GHG emissions as quickly as is technically and economically feasible.
What distinguishs the ECS estimates of the more reasonable climate skeptics is that they have a PDF which cuts of near 1 C as sharply as does the estimated IPCC PDFs, but also cuts of almost as sharply at or near 3 C. That it, what distinguishes these skeptics is that they are far more certain of their result than are the IPCC. That extra certainty is obtained by excluding a large number of emperical methods of determining the ECS, including all comparisons with past times, all assessments of modern temperatures except those by energy balance models; and of course, by excluding the non-emperical estimates based on the climate models. In addition, they seem to use a number of controversial assumptions, all of which reduce the estimated ECS. In short, what distinguishes them is a biased dogmatism most noteworthy for the quantity of information excluded from their estimates.
@Tom Curtis
Thanks for the info. That was a good explanation for a layman like me.
This community is great.