Rolling up the circus tent: Dispatch #7

Of particular interest to me were two papers presenting the first geoengineering simulations carried out with fully coupled dynamic ocean-atmosphere models. This is especially interesting in view of the importance of sea ice response in evening out the difference between the tropics-heavy solar radiation reduction vs. the more uniform CO2 radiative forcing. At the Harvard geoengineering workshop, David Battisti stated that mixed layer ocean simulations of geoengineering were of dubious utility, because they lack the most important processes governing sea ice formation and retreat. The two fully coupled simulations were presented by D.J. Lunt and co-workers, and C.M. Amman and co-workers. Sure enough, these simulations show that geoengineering is much less effective at restoring the natural temperature pattern than was suggested to be the case in the earlier simulations. In particular, if one tunes the global mean to have the right value, one fails to save the Arctic perennial sea ice. This is not a way to save the polar bears, as it has been sold, nor is it a reliable way to save Greenland. Another concern comes from atmospheric chemistry. In a talk substituting for a cancellation, the NCAR group showed that stratospheric warming in a geoengineered world increased ozone destruction — by a factor of 2-3 in the Arctic — even if one took into account the downward trend in stratospheric chlorine coming from the gradual reduction in CFC content of the atmosphere.

I continue to think that geoengineering is a big and unfortunate distraction, but since the cat is out of the bag, it is good that some people are doing the work to head off rosy and over-optimistic projections of sulfate geoengineering as a magic bullet that could substitute for the hard but necessary work of mitigation of CO2 emissions.

What was really most exciting to me in today’s sessions (and last night’s Sagan lecture, presented by Ralph Lorenz) was all the great thinking about solar system and extrasolar planetary climate. The missions in the planning stage for Solar System exploration are really something to look forward to. Now that people think Europa has a thick ice crust (maybe 20km or more) there’s less talk of drilling through with mini-subs, but there’s an orbiter planned, and possibly a lander with a seismometer,which would settle a lot of questions about the nature of Europa’s ice crust. And Ralph Lorenz’s work on hot-air balloon missions to Titan is really cool (about 110K, to be precise).

That wraps it up for this year. I’m sitting in the airport waiting for the red-eye to Chicago. The spontaneously organized session on extrasolar planetary evolution was so productive I’m thinking of organizing a Union session on the subject for next year.

With best wishes to all for Happy Holidays!

Page 2 of 2 | Previous page

96 comments on this post.
  1. Ellis:

    Just a little housekeeping, could you please update with your comments. Also, could you please clear up a couple of points. Is it generally agreed that the Earth has warmed .8c since the 1900? What portion of that warming, whatever the number may be, is currently ascribed to CO2 forcings? Also, in your comments to EOS you state that the current forcing of CO2 is 2.6 w/m^2. Is that figure based from 1750 and thus 2.34 w/m^2 is a better estimate per the IPCC’s recommendation of using 1860 as a starting point for forcing calculations? Or do you not agree with the IPCC, and could you please provide the reason why.
    Thank you.

    [Response: 2.6 W/m2 comes from IPCC AR4 and is the forcing for all long-lived greenhouse gases since 1750. The argument is not affected in the slightest by using 1850 or whatever as a starting point. Your question about attribution is not very well posed, see here – gavin]

  2. Lawrence Brown:

    Thank you, Ray and David for taking the rest if us along for the ride. It’s the next best thing to being there.
    I’m glad that energy policy was included as a topic. It’s critically intertwined with the future of Earth’s climate. Business as usual, as a policy, is no longer a rational option, yet Under Secretary Dobriansky and her delegation in Bali, acted pretty much as if this is their preferred course.

    The BAU approach almost insures that the Greenland and the West Antarctic ice sheets will continue to melt more rapidly than predictions called for, with a severe rise in ocean mean sea levels. More efficient energy use, alternative sources, and running vehicles on renewable electricity is mandatory to avoid inevitable disaster.

    Further explorations on Europa and Titan, and the possibilies of extra Solar Earthlike planets have to stir anyone’s imagination, who isn’t comatose.

  3. Hank Roberts:

    David, you write:
    > assume it was possible to instantly convert
    > completely now.

    You mean assume a zero net increase of greenhouse gases from human activity, as of right now?

  4. Andrew:

    “Andrew, you claim there is no difference between CO2 reduction and sulfate/aerosol geoengineering.”

    I did not say that.

  5. Patrick 027:

    QUESTION – …”stratospheric warming in a geoengineered world increased ozone destruction — by a factor of 2-3 in the Arctic — even if”…

    Shouldn’t the second word in the above be ‘cooling’ or am I wrong?

    [Response: No. Increasing the amount of stratospheric aerosol leads to a significant stratospheric warming (as the aerosols absorb solar radiation from above and long wave radiation from below). You can see the volcanic warm spikes in the MSU4 record. – gavin]

    [Response: Something that wasn’t discussed in the talk, and which I didn’t have the opportunity to follow up on, is why the temperature increase lead to an increase in ozone destruction. Certainly, all sorts of chemical reactions proceed faster at higher temperature, and increasing stratospheric water vapor could make a difference. My previous understanding, though, was that the occurrence of polar stratospheric clouds had the big leverage, and I’d think they’d become less prevalent with a warmer stratosphere. I’ll have to try to follow this up once the paper is available, but meanwhile maybe one of the atmospheric chemistry experts can chime in. Remember, these talks are only 10 to 15 minutes each, leaving little opportunity for presenting details. –raypierre]

  6. Andrew:

    $44: #

    #43 “Andrew,

    if you strayed into a minefield, what would you do? Carefully retrace your steps? Step where there are other footprints?”

    Depends on how I got there. If you skiied downhill into a minefield, would you carefully ski backward uphill?

    And that is sort of the point. To stretch your analogy to the current situation, explain what “retracing steps” means. Does it mean getting rid of the Clean Air Act so we get sulfate aerosols back up? Or returning to a pre-industrial population? Depopulating the Americas?

    But we aren’t slowly walking into this minefield. We skiied down pretty far before we suspected it. We aren’t really going to literally retrace our steps. We are going to go foward with attempts to change various chemical species in a very big reactor that we happen to live in. The way that humans successfully control complicated large scale chemical reactions is not normally through minefield analogies. It is much more often to apply modern control theory.

    I suspect that some people here might believe that some proposed approaches to climate control are silly and others are realistic. Well, if that is true, you would expect that to arise naturally from attempting to compute the control.

    [Response: You’ve got to talk specifics. What you have in mind is some kind of hysteresis loop that we’ve crossed. Can you provide evidence of this kind of hysteresis from increasing CO2 and decreasing it back? Certainly, hysteresis is possible, and becomes more likely as we go to higher and higher CO2. The whole point of CO2 mitigation is to avoid the very high levels of CO2 over long time scales, which could get us into a high risk of hysteresis. I don’t see any basis for comparing that to the situation where you let CO2 get very high, and then try to limit the damage by sulfate-based albedo management. -raypierre]

  7. Ellis:

    Gavin your response is a bit strange to me. You link to a post that says

    the total forcing from 1750 to 2000 is about 1.7 W/m2 (it is slightly smaller for 1850 to 2000, but that difference is a minor issue).

    And now you say

    The argument is not affected in the slightest by using 1850 or whatever as a starting point.

    Whereas IPCC 4 says on page 12 of 106

    The RF calculations usually take 1750 as the pre-industrial
    index (e.g., the TAR and this report). Therefore, using 1750
    may slightly overestimate the RF, as the changes in the mixing
    ratios of CO2, CH4 and N 2O after the end of this naturally
    cooler period may not be solely attributable to anthropogenic
    emissions. Using 1860 as an alternative start date for the RF
    calculations would reduce the LLGHG RF by roughly 10%.

    Now, I am sure you are right that this does not effect your argument, but it will definately effect the numbers. With that said, I just want to know your justification for using 1750 as a jumping off point, and please, I really want to know, don’t tell me it is because the IPCC uses that date.

    As for ill-posed questions, it is no secret that I am not a scientist, I believe that every keystroke gives me away a little more. I will be blunt, because I do not know the right question to ask, how does a forcing at the tropopause effect the surface? I am sure you have a post that already answers the question, perhaps you could point me in the right direction.

  8. Andrew:

    “What you have in mind is some kind of hysteresis loop that we’ve crossed.”

    I think there are lots of changes to the biosphere which can occur which are a lot harder to reverse than the effect of CO2 on temperature.

    For example there is a possibility that climate change could essentially destroy the Amazon rain forest. Well suppose that happens. OK now drop the CO2 back down. Will that rain forest grow back the same as before or might some other ecosystem displace it?

    The real point here is what is the problem with posing the question in a way that can objectively assess what the most efficient control for the climate is?

    Next guy that thinks we just have to look at the sensitivities to figure out what to do? I got this helicopter I want to see your computer fly.

  9. dhogaza:

    Next guy that thinks we just have to look at the sensitivities to figure out what to do? I got this helicopter I want to see your computer fly.

  10. Nick Gotts:

    Re #45 [Greg] “If everyone in the country went and got their Carbon Footprint score from and then took just one pledge I think we could stop global warming.”

    You appear to make the implicit assumption, remarkably common among US residents, that nothing outside the USA could possibly make a difference. I’m sure if you had stopped to think a moment, you would have realised this is not so, but such automatic assumptions are very revealing, and can be very damaging. AGW cannot be halted without the USA, but neither can it be halted by the USA alone, nor will the USA necessarily lead the effort. It’s time to think globally.

  11. pete best:

    Re #58. How long does it take a tree to become a forest naturally, probably some 300 years anyway. Once lost, lost for a long time even if we did somehow remove CO2 down to 1750 leves.

  12. Martin Vermeer:

    And that is sort of the point. To stretch your analogy to the current situation, explain what “retracing steps” means. Does it mean getting rid of the Clean Air Act so we get sulfate aerosols back up? Or returning to a pre-industrial population? Depopulating the Americas?

    Not exactly… no need to traverse dead ends or loops on the way back, as you wouldn’t in a real minefield either. Also no need to re-play old television programs from past eras backwards… we’re talking climatology here (is it an American specialty to equate/confuse driving SUVs with physically existing? In that case yes to the last question :-) )

    But we aren’t slowly walking into this minefield. We skiied down pretty far before we suspected it. We aren’t really going to literally retrace our steps.

    But it would be wise to try. Of course you won’t be able to precisely ski in the same tracks, due to the dynamics of skiing… and that makes for some excitement in life.

  13. Ray Ladbury:

    Andrew, climate mitigation is not simply a matter of twiddling knobs. There are many knobs where we have only a limited understanding of how sensitive they are–and aerosols are among them. On the other hand, we have quite a good understanding of CO2 forcing. Now given that we are monkeying with the only habitable planet we have yet discovered in any planetary system, does it not make sense to mitigate first where we fully understand the efficachy and consequences of what we are doing?

  14. David B. Benson:

    Hank Roberts (53) — Yes. Obviously conterfactual, just an attempt to keep the post short and simple.

  15. Hank Roberts:

    > sensitivities … I got this helicopter
    > I want to see your computer fly.

    “Airplanes want to fly. Helicopters want to crash.”

    The climate system has been remarkably resistant to crashing; I suspect it’s more like an airplane than a helicopter in that regard.

    But remember, there’s only so many rivets you can take out of an aircraft in flight and have it hold together. The important thing about tinkering is … you know.

    They say in flying you need two of three things — airspeed, altitude, and ideas.

    We have the first two, gift of nature and a longterm quiet climate. So we’ve pushed the envelope.

    Now? We’re going to put the planet into something exciting this century (hammerhead stall? flat spin?).

    Next? We step aside, put our nieces and nephews, children and grandchildren in the pilot’s seat, and they take over — to perform the recovery portion of the exercise.

    Good luck, kids. Hope you have good ideas.

    PS, want to learn to fly that helicopter, or write a program that can fly one? Here’s a start: and instructions

  16. Ed Sears:

    #48 Ralph Smythe

    We can remove Co2 from the atmosphere by increasing biomass and soil organic matter. Broadly speaking, continuous cover is the best thing for soils so woodland has the soils with the highest stored carbon, and annually cultivated arable farmland has the least. Approaches to creating sustainable human agroecosystems which preserve soils and biomass and which have been implemented around the world include organic farming, agroforestry and permaculture. Some prominent researchers are Peter Smith (Aberdeen University) for carbon sequestration in soils and J. Pretty for the benefits of organic farming methods in developing countries. Peat soils (also known as ‘organic’ but in a different sense to the farming method) deserve a special mention as they are often metres thick as opposed to centimetres in normal ‘mineral’ soils, and are correspondingly more important as a carbon store. Which brings us to the real world and politics: there are not yet specific measures within the Common Agricultural Policy to encourage carbon sequestration, although other policies on biofuels and set-aside may have an effect. So in the EU there is not a big increase in carbon stored in agricultural areas, more likely a gradual decline. Forest cover is increasing moderately. Comparable situations exist in other developed countries. However, as you are no doubt aware, some parts of the planet are losing forest cover and agricultural soils through logging, desertification and the introduction of intensive farming. In carbon terms, the draining and destruction of peat soils in Indonesia is of critical concern. In human terms, decreasing productivity around the sahara may cause big problems, as high temperatures are a threat to carbon storage in soils.

  17. Ed Sears:


    What are the likely dynamics of the carbon-climate-human system into the future, and what points of intervention and windows of opportunity exist for human societies to manage this system?

    This is the rather all-encompassing title of a UNESCO-SCOPE policy brief no.2 October 2006 (pdf available online) which I think comes at the climate change problem from the angle you would like. Sulphate geoengineering is in fact mentioned in the list of rapidly deployable technologies, although it is one of 16, and most of the others have far more definable and manageable risks. This has to be emphasised: raising vehicle fuel efficiency standards or reforestation of degraded land (the Chinese have planted 49.2 billion trees since 1980) is a whole different level of safety/risk/lunacy compared to simulating a major volcanic explosion in the atmosphere. The cost of doing sulphate geoengineering is nothing compared to the possible cleanup costs when (if?) it goes wrong. For what it’s worth, if I was in charge of a country, I would see it as a matter for national security that other countries *don’t* carry out ill-planned climate engineering.

  18. JohnLopresti:

    Somewhat offTopic: I wondered how much energy acceleration, i.e., warming plotted over time, we are witnessing in the record setting retreat of summer arctic ice in 2007; i.e., in a modelled environment with scores of variables, it might be possible to characterize the weight of discretely expressed elements in such an aggregate equation.
    Also only slightly OffTopic, appreciation to the author for the glimpses into AGU’s challenging presentations at the yearly meeting*.
    *The city where the AGU meeting just folded its tents is in an area that has received only 3/5 its yearly average rainfall so far in its currently ongoing wet season.

  19. Bryan S:

    I would like to give some short background then pose some interesting questions to the Real Climate team.

    During the Upper Miocene, it is generally understood that NH ice sheets were not yet formed (maybe beginning as isolated small sheets). However by the beginning of the Upper Miocene, both the WAIS&EAIS were present. Even without the Greenland Ice sheet however, sea levels in the Upper Miocene (5-11 million years ago) averaged very near present levels, with four distinct lowstands, one of which (10 mya) stood 100 meters below present levels. Only one significant highstand of sea level is present in the Upper Miocene sequence strat record (approximately 6 mya, and it only stood maybe 20-30 meters above present. The d18O proxy certainly points to the global Miocene climate being much warmer than the present climate. In light of these observations:

    Questions: 1) Why were Upper Miocene sea levels averaging about what they are presently, puctuated by several significant drops well below present levels? b) How is this possible without a NH ice sheet during the Upper Miocene (a global climate much warmer than present)?
    2) Does this tell us something important about the long term stability and even possible growth of the SH ice sheets in a warmer world?
    3) Why is not more attention being given to research this? It seems particularly relevant to me, given that the paleogeography was much closer to our modern world than further back in the record such as during the PETM.

  20. David B. Benson:

    Bryan S (69) — Good questions! The only pieces of evidence I can offer is that the Panama Isthmus closed during 3–5 million years ago and the ‘Wall of Africa’, dividing east from west rose during this same time interval (approximately). Both affected climate, but I amateurishingly opine these changes cannot alone account for your observations.

  21. Patrick 027:

    Re 55 Responses – Thanks. (I hadn’t read the abstract for that presentation; I guess I was thinking more of a space-based sunshade or troposphere/surface based albedo enhancement, granted that depending on wavelengths and cloud cover, etc, the later one could also increase solar heating of the stratosphere.) (PS I thought that the stratospheric aerosols themselves would be an ozone depletion issue.)

    Re 69 – Not sure about this particular case, but I do know that over long-enough geologic timescales, geological processes contribute significantly to sea level variations. The rate of sea floor spreading can vary a bit, and faster sea floor spreading leads to wider mid-ocean ridges (elevated by heat – the crust subsides as it cools, which occurs at further distance if spreading is faster), which will displace ocean water – effectively, the average sea floor elevation rises relative to the continents, so they ‘sink’ beneath the water. Also, when continents collide and uplift mountains, that adds to the volume of crust above sea level and leaves a larger ocean basin, allowing sea level to drop. Erosion dumping sediments onto continental shelves and slopes would reverse that, as would spreading and thinning of continental crust. There may be other factors; individual continents may ride over parts the mantle with variations that cause them to sink or rise, for example.

    But those changes may not happen fast enough to explain this case, though I’m not sure – maybe. Also, the Mediterranean sea dried up significantly at some point (more than once?), due to temporary closure of the Strait of Gibralter.

    Re 57 last paragraph – I have some other posts elsewhere that go into the matter in some depth, to which I might refer you at a later time; for now:

    lapse rate = rate of temperature decrease with height

    dry adiabatic lapse rate = the rate of temperature decrease with height that occurs during dry adiabatic convection, wherein no net diabatic heating (via radiation or latent heating/cooling, or for that matter, frictional heating or chemical reaction, etc.) occurs.

    moist adiabatic lapse rate = similar to dry adiabatic lapse rate except that this is for when the water vapor is saturated and condensing during ascent. The latent heating causes the air to cool off less rapidly with height (but it still cools off).

    In a very thin layer near the surface, where convection occurs but large motions are inhibited by the surface, the lapse rate can become larger than dry adiabatic, but on larger scales the dry adiabatic lapse rate sets an upper limit on lapse rates, because any larger lapse rate is unstable and will lead to rapid overturning that would then reduce the lapse rate.

    The distribution of radiative heating and cooling in the atmosphere is such that radiation alone would tend to drive the lower atmosphere near the surface towards being unstable; this and horizontal heating/cooling variations drive the convection that define the troposphere, the lowest layer of the atmosphere (which happens to contain a large majority of the mass of the atmosphere). Because of the moisture available, this convection tends to maintain a lapse rate lower than a dry adiabatic lapse rate.

    To be precise one would have to go into the horizontal and temporal variations – and there are such variations in the above. But one can get a good basic idea of how the greenhouse effect works by considering a one-dimensional model, which is just a column of atmosphere that is representative of the Earth’s atmosphere (or the atmosphere of whatever planet one wants):

    There is a distribution of solar heating (SW radiation) – a majority is concentrated near the surface but some does heat the atmosphere directly. Convection redistributes this heat. (Some SW radiation reflected or scattered back to space and so does not participate in this heating.)

    LW radiation is the radiation the Earth’s surface and atmosphere can emit at their temperature ranges; this is a band of wavelengths longer than SW wavelengths (PS SW does extend into the infrared a bit, so it is incorrect and confusing to consider solar radiation to be UV and visible only).

    The atmosphere can be divided arbitrarily into a number of layers; each layer emits and absorbs some LW radiation, as does the surface. Some LW radiation escapse to space. Proportionally more LW radiation can escape to space from upper layers because the layers can absorb what is emitted by other layers and so radiation from lower layers is more easily absorbed before escaping. The surface can also absorb LW radiation from the atmosphere.

    In an equilibrium climate, LW radiation escaping to space must balance the SW radiation that is absorbed. SW radiation is absorbed, that heat energy is redistributed by convection and by the LW radiation that is emitted and absorbed, and ultimately leaves by that portion of LW radiation that escapes.

    Optical properties vary by wavelength, and it is helpful first to consider what would happen at just one wavelength. At any wavelength, for fixed optical properties, emission of LW radiation increases with increasing temperature. Thus, the net flow of LW radiation tends to be from hot to cold, as with conduction of heat… But I have to go now; stay tuned.

  22. Patrick 027:

    Re 57 part II: With that introduction, I refer you now to my comments 184 and 189 in

    And then, note that radiative forcing at the tropopause is defined for after equilibrium is restored in the stratosphere (and above, presumably), but before any climatic response at the surface or within the troposphere. Of course, as the troposphere and surface respond, the stratosphere will again adjust, and the tropopause will also tend to move.

    PS other comments I made in the same thread (some but not all applying to this subject):

    171 – 176

    and at


    and I also have some comments pertaining to the topic at: (as in a geologic time column, the comments there go from the bottom up.)

  23. David B. Benson:

    Patrick 027 (71) and Bryan S (69) — It seems that the Mediterranean dryed up ten times in the Late Miocene! Here is a line to an abstract:

  24. David B. Benson:

    Bryan S (69) — The sea high stand appears to be at (approximately) the same time as the Mediterranean desiccations. The paper linked below suggests the possibility of NH ice sheets during the sub-epoch in question, being titled Reconstructing the late Miocene climate and oceanic heat transport flux…:

  25. David B. Benson:

    The abstract by Tripati et al. contained in

    on page 22 concludes with The … storage of ice in both the Northern and Southern Hemisphere[sic] … after 34 Ma.

  26. Andrew:

    63: “Andrew, climate mitigation is not simply a matter of twiddling knobs.”

    From the point of view of control theory, it is. You can substitute the word “control input” for “knob” but that’s basically all it amounts to. Want to call it a manifold? Fine. But it’s still really just knobs.

  27. Andrew:

    “The climate system has been remarkably resistant to crashing”

    Is that why it has done it so often in history?

  28. Andrew:

    59: “”

    No not that helicopter. This helicopter:

    The reason to point at this paper is because it is a case where exact linearization leads to an unstable controller.

    So it is a relatively precise demonstration of my point – that it is NOT sufficient to know sensitivities of a nonlinear system in order to control it.

    In the climate control case, the analogy would be unnecessarily large seasonal (or perhaps decadal) oscillations.

    Sure you might get the overall average level where you want it but it is worth asking whether the road we travel to get there is the smooth one.

  29. Andrew:

    “What are the likely dynamics of the carbon-climate-human system into the future, and what points of intervention and windows of opportunity exist for human societies to manage this system?”

    It’s worse than that. I am not convinced that we know the likely trajectories for the future. This is pretty much what gets you to the realization that we have to deal with robust control – the question of how to control something when you do not have good knowledge of the dynamics, (and in this case, you do not have good knowledge of the control inputs either). If we take into account the skeptics of the world, we will actually be arguing over the knowledge of the plant outputs as well.

    The huge advantage of control theory is that it is not really a new problem of how to control a system where there is imperfect knowledge of the plant. And it is possible in these methods to be able to know how much of the uncertainties can be tolerated, and which cannot. There are many aspects of climate where the “error bars” are a pain in the butt to nail down. Well, it is entirely possible that not all of them need to be nailed down.

  30. Ray Ladbury:

    Andrew, if you apply control theory, you will find that either:
    1)We have no control over many parameters (e.g. insolation)
    2)That we do not have sufficient understanding of the parameters to predict how they will affect climate.
    3)That the duration of the effect is too short to really help us much (e.g. aerosols)
    4)That we cannot rule out adverse consequences for the manipulation on the scale that would be needed to have an effect (e.g. sulfate injection).

    In fact, the level of assurance we require that the technique will be effective and that we will not make things worse pretty much precludes anything but reducing CO2 in the atmosphere.

  31. Andrew:

    81. “Andrew, if you apply control theory, you will find that either:
    1)We have no control over many parameters (e.g. insolation)
    2)That we do not have sufficient understanding of the parameters to predict how they will affect climate.
    3)That the duration of the effect is too short to really help us much (e.g. aerosols)
    4)That we cannot rule out adverse consequences for the manipulation on the scale that would be needed to have an effect (e.g. sulfate injection).”

    1) Parameters which cannot be controlled are no problem. In fact, there are some parameters which are too costly to be part of efficient control. What is more important (and somehow continues to elude you) is that this is why you don’t want to just start pulling levers – you really do want to look at this from the point of view of control theory.

    2) How do you know it is insufficient if you do not have an estimate of an efficient controller? There are going to be many things which are irrelevant to an efficient controller; and it will be really helpful to know which things those are. I would guess that geoengineering would be in that box, but it is a real bad mistake not to come by that conclusion honestly. If you think about trying to seriously show that geoengineering is a waste of time, you will have to demonstrate something more or less along the lines that control theory would make precise.

    3) This is not at all obvious; my suspicion is that it is dead wrong. If you end up having to carefully sculpt seasonal or decadal oscillations it’s nice having some things which act on a fast time scale.

    4) Adverse consequences are included, either in the cost of control, or in the deviation of the trajectory from the specification. For the adverse consequences you are talking about here, I think it more likely that having them in the specification of the controlled trajectories would be more useful.

  32. Hank Roberts:

    Andrew, those knobs evolve under selection pressure from that twiddling, and with very short generation times. You don’t know what they’re changing into yet. Look at the paleoecology for natural rates of change in past excursions; the current rate of change is likely to be as fast as the current excursion a couple of orders of magnitude faster, assuming nature can keep up with us, and much faster and less predictable if not.

    I’m just an amateur reader, but I suggest you try reading this:

    The problem with NPZD models is that their representation of biological fluxes is entirely dependent on physical processes. These models do not include many of the ecological processes that are known to be sensitive to, for instance, changes in temperature or pH, such as bacterial remineralization (Rivkin and Legendre, 2001Go), zooplankton grazing rates (Buitenhuis et al., 2006Go), the aggregation role of mucus secreted by some phytoplankton (Engel et al., 2004Go), the ballasting of organic particles by plankton shells (Klaas and Archer, 2002Go) and pH sensitivity of calcifying phytoplankton (Riebesell et al., 2000Go) and zooplankton (Orr et al., 2005Go).

    … One of the great values of large-scale modelling is that it enables us to examine the consequences of physiological differences between PFTs for large-scale phenomena such as spatial distributions and seasonal successions. We will not understand ecology until we have built models that include the necessary processes.

    ——–end excerpt——-

    Found here:

  33. Ray Ladbury:

    Andrew, I recommend to you the counsel of H. L. Mencken:
    “Explanations exist: they have existed for all times, for there is always an easy solution to every problem — neat, plausible and wrong.”

  34. Andrew:

    “Andrew, those knobs evolve under selection pressure from that twiddling, and with very short generation times. You don’t know what they’re changing into yet.”

    Yes, that is one of the things that makes the modern robust control theory approach attractive. Remember I’m the one that pointed out the possibility of fast and possibly irreversible responses (like loss of the Amazon rain forest) being a potential outcome.

    What seems to be completely ignored though, is my repeated explanation that the less you know about the system, the more you really ought to take advantage of the tools which have recently become available for controlling systems which have large uncertainties. In particular, the main reason that you want to think about robust control for climate is that you do not completely know the plant, you do not completely know the control inputs, and it is even possible that people will not be able to agree on the measured system outputs. When you have a high dimensional nonlinear system with feedbacks as one of the happier things you can say about the problem (and this is not really that happy a thing to have to say) then you really don’t want to screw around with guessing which knobs to turn and which to ignore.

  35. Andrew:

    “Andrew, I recommend to you the counsel of H. L. Mencken:
    “Explanations exist: they have existed for all times, for there is always an easy solution to every problem — neat, plausible and wrong.””

    Well Ray, you could crack the odd book yourself. In particular there is probably a proceedings of the 2004 AGU 2004 Fall Meeting where there might be a paper from this invited talk:

    A13C-07 INVITED 15:20h

    “Control Theory and Analysis of Feedback Systems

    * Murray, R M ( , California Institute of Technology, 1200 E. California Blvd, Pasadena, CA 91125 United States

    Control theory has developed a collection of tools that can be used to better understand the behavior of complex, interconnected systems. This talk provides an introduction to some of the control theory tools relevant to climate dynamics. Specific concepts that will be described include input/output modeling, modeling reduction, and analysis in the presence of unmodeled dynamics.”

    Now I doubt that Murray would have exactly the same take on control synthesis that I would, since I use a lot of stuff that isn’t in the open literature, but he’s definitely in the business. You probably know guys at JPL who know him. I am pretty sure that what he means by “analysis in the presence of unmodeled dynamics” is a version of what I am trying to get across here.

    So it’s not so bad as I thought. At least one guy in climate figured out they should get hip to control theory.

  36. Ray Ladbury:

    Andrew, you misunderstand my criticism. I am not challenging the use of control theory. Rather I am pointing out that control theory dictates that you must take into account the uncertainties and potential adverse outcomes in deciding which knobs you can twiddle. The fact of the matter is that we cannot change most of the variables that affect climate. Of those we can change, weunderstand well precisely one–greenhouse gasses. So what your little control theory analysis will wind up telling you is “Hey, maybe better back off on the CO2, huh?”
    Since we have precisely one habitable planet, any control-theoretical analysis will have to adopt only those solutions where there is extremely high confidence we will not irreparably screw things up. Since we cannot eliminate the possibility of severe consequences or establish the efficacy of, say, using sulfate aerosols, that is not an option.
    Iron fertilization to cause algal blooms has been shown not to be viable.
    Got any other tricks up your sleve?
    Andrew, the reason control theory is not used more in climate remediation is not because climate scientists are all a bunch of stodgy, conservative, ignorant twits. Rather it is because the only viable solution is obvious.

  37. TheWonderer:

    Another interesting discussion on control systems, however, I just can’t imagine building such a large “control system” without a reasonably accurate simulation, and a physically representative prototype, and I can’t fathom how we’d do that. We wouldn’t allow that leap of faith for a nuclear power plant control system, a much smaller gambit, would we?

  38. Hank Roberts:

    > Now I doubt that Murray would have exactly the same take
    > on control synthesis that I would, since I use a lot of
    > stuff that isn’t in the open literature

    What justification is used for keeping tools secret that could help, in this situation?

  39. Andrew:

    “Rather I am pointing out that control theory dictates that you must take into account the uncertainties and potential adverse outcomes in deciding which knobs you can twiddle”

    Well it doesn’t, actually. This may be the reason I don’t understand where you are coming from.

  40. Andrew:

    “So what your little control theory analysis will wind up telling you is “Hey, maybe better back off on the CO2, huh?””

    It might end up with that result. But the advantage of getting it via control synthesis is that you would be explain to explain not only why that would achieve your goals, but why it would be likely to cost less than other approaches. So if you don’t like geoengineering, this would at least be an objective way to do that.

    In actual fact, I am pretty sure that the schedule of action that would result from a serious application of control theory would have a few interesting twists; sure you will probably want to deal with CO2, but the quantitative details of how and when are probably not something that would be obvious.

    The certification of minimum cost would help with skeptics who think it is going to be expensive – it would present them with essentially a verifiable claim that in order for the climate to be within some acceptable bounds, then the least we must do is X.

    There is something very suspicious with the idea that we can solve the most complex control problem ever attempted with the simplest imaginable controller.

    And frankly, there are quite obvious experiments which could be done. There are what, a dozen or so GCMs out there? For one thing, you could synthesize a minimum cost controller that can handle them all. If you can’t even do that, what reason is there to believe that we will control the real climate with anything approaching minimum cost?

    And if you don’t try to determine the minimum cost, that leaves you open to all sorts of charges of ruining the world economy. (Charges which already are pretty familiar from some quarters).

    Another very important aspect of a control theory approach would be that you would also be able to evaluate things like a market-based control. Different parties may have different desired trajectories for climate. This would provide a way to put that into economic terms – so you would know how much to charge for deviation. I believe that there are a lot of parties in the picture who seem to think entirely in economic terms.

  41. Hank Roberts:

    > There is something very suspicious with the idea that we can
    > solve the most complex control problem ever attempted with the
    > simplest imaginable controller.

    No, you’re confusing the leverage point with the controller.

    Getting people to reduce fossil fuel use is inordinately complicated, and if your control theory is being applied by the political science folks it’d be interesting.

  42. Ray Ladbury:

    Andrew, any application of control theory would have to take into account what we know about climate. That is contained in the models. If control theory does not take into account uncertainties, potential adverse outcomes and likely efficacy of the remediation at some required confidence and probability of success, then it is not suited to the task. If it does, it will tell us that reducing carbon is the only way to move at present given our state of knowledge about the climate.
    One of the reasons for this is because we would have to require a high degree of certainty that we would not be making things worse. Go ahead and do the analysis–I’m pretty sure what your outcome will be.

  43. Bryan S:

    Comment on Late Cenozoic Ice Sheet Variability and Control on Global Sea Levels

    Thank you David B. Benson for your interest in my comments, and your links.
    A powerful way to gain insight into the role of ice sheet dynamics and their control on global sea levels is to superimpose a smoothed d18O proxy (or composite of several individual cores) over a composite eustatic sea level curve. As a greater number of deep ocean cores are recovered, and the sequence stratigraphic record across stable platform areas is further refined, a fascinating picture is emerging. Although there is broad correlation between the records, there are also periods when the records are out of phase. As the record improves, these may come into closer agreement.

    As an example of correlation, a large sea level lowstand (50-100 meters below present) near the beginning of the late Miocene (10 ma) corresponds to a dramatic shift in the d18O proxy. Even so, the d18O proxy provides compelling evidence that the overall global climate throughout the Late Miocene (5-11 ma) was certainly much warmer than our modern climate, but also similar in warmth to that of the Early Pliocene (3.5-5 ma), when the most significant sea level highstand of the last 10 million years occurred. During the Early Pliocene highstand, sea levels were perhaps 100 meters higher than present. The Pliocene high sea level event contrasts to average sea levels throughout the Late Miocene which averaged near present levels, but were punctuated by four distinct lowstands 20-100 meters below present levels, and only one significant highstand near the Miocene/Pliocene boundary (Haq et. Al. 1987; Mitchum et al. 1994).

    That being said, Denton et al. (1991) argued that the ice volume on Antarctica was roughly half of the present volume by 15 ma, and perhaps exceeding the present volume by roughly 12 ma. The EAIS has now been traced back into the Early Cenozoic (Barron et al., 1991; Hambrey et al., 1991). There is also evidence that the WAIS was present by the Late Miocene (Abreu and Anderson, 1998). Both the EAIS and WAIS were certainly present during the Pliocene (Denton et al., 1991; Webb and Harwood, 1991; Barrett et al., 1992). Surprisingly, the beginning of the NHIS is now placed by several workers in the Early Pliocene, during an extremely warm global climatic event, and may have been intermittently present in the Late Miocene. During the pronounced Late Miocene low sea levels, there is little evidence of a significant ice sheet present on Greenland.

    From a contemporary perspective, the combined records summarized above should give considerable confidence that a catastrophic decrease of the Southern Hemisphere Ice Sheet volume is not likely even in a severe modern global warming event. In fact, it may be argued that SH ice volumes significantly greater than present have occurred during global climates much warmer than we are likely to see even with a high climate sensitivity to added greenhouse gases. The highlighted record shows that the ice sheets respond in a very non-intuitive (non-linear) fashion to global warming, and in large part may be controlled by regional changes in oceanic heat transport and weather patterns which are especially challenging to accurately model and predict. The Greenland Ice sheet has not been around nearly as long, so its response is less certain. It should be pointed out however, that decreased ice volumes in the NH might seemingly be balanced by growth in the SH ice sheets.
    The Early Pliocene event was not likely to have been driven by significant NH melting, but rather a possible regional response to ocean circulation and weather patterns allowing changes to the SH ice volumes. In summary, the ancient record of ice sheet dynamics and correlation to eustatic sea level paints a complex picture, and illustrates a consistent theme of how the earth system is very non-linear and difficult to predict. It does show however, that the SH ice sheets are likely the key driver of global non-thermosteric sea level variation.

    It is my hope that these comments will provoke interest among the readers and contributors of Real Climate, and hopefully motivate all to recognize the significance of the paleo and geologic record in helping gain a better perspective in the issues involved with global sea level variation its response to climate change.

  44. Jim Redden:

    RE #41 Alexander Harvey

    Just a short note to thank Alexander Harvey for the expansion of the MEP principles in context.

    Seeing the climate system as a heat engine in process, and noting that increasing CO2 increases the retained energy of the system, might have more than academic implications.

    Bets are on any results will not violate any well established laws of thermodynamics, nor classical physics.

  45. Bryan S:

    Gavin, I read with some interest the article in the December 2007 EOS summaring the January 2007 NOAA workshop on ice sheet modeling. In the second paragraph, your group cites the IPCC Fourth Assessment Report, stating [the understanding of rapid dynamical changes in ice flow “is too limited to assess their likelihood or provide a best estimate of an upper bound for sea level rise”]. Yet, in the first paragraph, your group states flatly “poorly represented physical processes in the ice sheet component *likely* lead to an underestimation of sea level rise forced by a warming climate”. It seems curious that no reference of anyone’s specific work was cited, but only this rather sweeping proclamation. Would you please clarify in more detail why such an important hypothesis is presented without any reference to supporting data. If the reference was inadvertantly left out,could you please supplement the paper here so that we may become better informed on this hypothesis.

    In the section Underlying Problems, it is pointed out that key processes should be incorporated into the models in order to make reliable prediction of future ice sheet change. The first included iteraction of ice sheets with the ocean, requiring models of regional oceanic circulation. Presumably such would require that multi-decadal predictive skill of ocean circulation changes on a very fine scale, and in just the right areas to be able skillfully model their effect on ice discharge. It seems to me that such is a daunting challenge. Could you give some further insight.

    Thanks in advance for the reply.

    [Response: The reason why the uncertainty is predominantly on the up-side is related to the dynamical changes that have been seen in Greenland, West Antarctica and the Antarctica peninsula – none of which are well captured in current ice sheet models. The surface mass balance changes (where you conceivably have counteracting forces) are much easier to model and capture and so the uncertainty is less. At the workshop, the ice sheet people there – Tony Payne, Shawn Marshall, Christina Hulbe etc. – all made it clear that their models were not up to the task – they didn’t have the rheology that would allow them to predict accelerations of ice streams due to lubrication effects or collapse of the floating ice sheet, they didn’t have the water balance to allow them to keep track of the surface melt (the moulins), and they didn’t have the ocean/sub-ice sheet interaction and so on. So they are in a very uncomfortable position of having models that clearly don’t show what is already happening. That, let me tell you, is not a good place for modellers to be. There are however a lot of good ideas and improvements that could be made relatively quickly that don’t require huge advances in regional modelling (though some kind of downscaling is necessary). – gavin]

  46. Jason:

    What is an extra-solar Large Earth? Is that a planet like Saturn or Jupiter only Earth-like?