332 thoughts on “Observation vs Model – Bringing Heavy Armour into the War”

1. Here is something that has always worried me; a location on the equator has 365 days/nights per year and the poles only one, somehow you have to merge the two.

2. The reason for slower response at the poles is simply that it takes a very long time to melt an icecap. Albedo feedback is huge at those latitudes. It also takes a long time for a forest to migrate northward in response to changing climate. At the equator, there is no such migration.

3. Yes, It’s possible that the GCM’s get this bit right. But there are so many other problems, it may be chance. Upper tropospheric hotspot anyone? If the fundamental mechanisms are wrong, predictive power is questionable. Another issue is that the models do seem to be overpredicting short term temperature response as pointed out by Lucia a couple of posts ago. Even the mechanism proposed seems to me not really supported by the data. Antarctic temps are not responding anywhere nearly as fast as Northern Hemisphere temps despite low aerosols in the antarctic.

4. Paul_K,

   Could I ask for a clarification of your opening statement.

   “most of the AOGCMs exhibit a curvilinear response in outgoing global flux with respect to average temperature change. One of the consequences of this is that there is a well-reported apparent increase in the effective climate sensitivity with time and temperature in the models.”

   “Sensitivity with temperature” presumably means that most GCM’s predict a higher climate sensitivity for a warmer earth, all else being equal.

   What about “sensitivity with time”? Does this mean that most GCMs indicate that climate sensitivity was lower in the past, and will be higher in the future? Or that as a GCM run progresses (through time), that its apparent effective climate sensitivity increases? Or that the GCMs of today generally report higher effective climate sensitivities than the ones of “some time ago” (say, those published in 1990)?

5. Paul_K,

   However, there are a number of reasons to believe that some curvature is likely for simple physical reasons,

   Exactly what I’ve been trying to tell you since you posted your first article about this.

6. While the Arctic may not be a large factor in the overall energy balance or sensitivity determination, it always seemed to me that it was improper to apply equation (1) there. The area of sea ice is not time-invariant, so why should the Arctic behave as a linear time-invariant system? At a guess, albedo effects may vary more like the integral of forcing, rather than the forcing itself, and also have a built-in limit.

7. David Young:
   > Upper tropospheric hotspot anyone?
   Predicted by models, observed in data (see Tichner 2009 among many others). This is a problem?
   > If the fundamental mechanisms are wrong, predictive power is questionable.
   What fundamental mechanism do models get wrong?
   > Another issue is that the models do seem to be overpredicting short term temperature response as pointed out by Lucia a couple of posts ago.
   Though I’m not certain which post you’re referring to, what Lucia usually is interested in are the linear regression rates, not temperature response per se. After correcting for the short-term effects of ENSO, solar, and volcanoes, observed rates of global temperature rise fit IPCC projections quite well (Rahmstorf et. al. 2012).
   >Antarctic temps are not responding anywhere nearly as fast as Northern Hemisphere temps despite low aerosols in the antarctic.
   Yes, and once again as predicted by models (see, e.g., AR4 fig. 11.21). So I don’t know what your complaint really is.

8. If in the virtual world of models the earth in 2100 is different from year 2000 with respect to the relationship between forcings, feedbacks and climate sensitivity then what is the hope that there is a linear relation between now and the LGM or any other epoch climate scientists have used to estimate climate sensitivity? With respect to climate sensitivity us and our ancestors have lived on many different earths.

   Paul_K it seems like paleo estimates of climate sensitivity are also thrown into question by this work, another point to add to your inferences.

   http://www.skepticalscience.com/graphics/Climate_Sensitivity_500.jpg
   The above image shows the different methodologies used to estimate climate sensitivity. Can anybody with more knowledge than me point to the methods that aren’t affect by this or other recent critical papers on the subject.

9. AMac:

   What about “sensitivity with time”

   If you prefer, use “frequency”.

   Imagine driving a system with a (small enough) sinusoidal forcing at a fixed frequency. The amplitude of the response in general is frequency dependent. If you drive it with high enough a frequency you’ll essentially get no response. The maximum response for a “small enough” forcing will in general be for a zero-frequency aka constant forcing.

   Strictly, these statements apply only to responses of a linear system to an external forcing. For low frequency systems at higher amplitude forcings, you can get “adaptation”, and the effective response of the system may actually be lower for a DC forcing than for a band of non-zero frequencies.

   Hence I think PaulK is warranted in his cautionary: I would emphasize that the fact of there being a good explanation for the curvilinear relationship in outgoing flux exhibited by the GCMs does not per se mean that such a relationship must hold in the real world.

10. Hi Paul,
    I had in a discussion with SteveF once earlier here tried to reconcile higher sensitivity values from your earlier demonstration that higher order models yield higher sensitivity coupled with Isaac Held’s post on time-dependent sensitivity. I had paraphrased Isaac Held as saying that the energy balance at the TOA bears a non-linear relationship to the the global mean surface temperature; and that the underlying dynamics can still be studied by linear analysis but only if one chooses a surface temperature field instead of a mean surface temperature.
    Where Armour seems to diverge from Held is that this curvilinear reponse arises from polar amplification, not ocean mixing in high latitudes – do I understand you correctly?

11. AOGCMs exhibit a curvilinear response in outgoing global flux with respect to average temperature change and PaulK shows by way of Armour et al that varying values of lambda and of tau by latitude can produce a curvilinear response and therefore many observation-based estimates of climate sensitivity, founded on a global linear feedback assumption, might represent no more than a lower-bound.

    “The landmark papers on the partitioning of feedbacks in the GCMs produce arbitrary and misleading results. Note that this is true whether or not the curvature is a realworld phenomenon, since the curvature most definitely does exist in the models.”

    So what else is new?

12. KAP, My gosh, where have you been? The paper you refer to seems to be all about homogonizing the data. Apparently, knowledge of the “truth” permits a critical assessment of the ability of the system to recover the large-scale trends and a reinterpretation of the results when applied to the real observations. So, if we know the truth we can adjust the data to match?? Basically, the data does not match the climate models, but “remaining baises may account” for this fact.

    There was a big paper on this in the last 5 years with about 10 authors and I read it. The bottom line is that if you find every source of error in the data and adjust it, you still disaagree with models. So, you instead use wind speed as a proxy for temperature (ignoring the thermometer itself) and this proxy just barely is within the error bars of the models. Since wind speed data is notorously noisy, its error bars are large. This is just ridiculous science.

    The hot spot is not in the data despite a decade of trying to rationalize why its not there.

    Andy Lacis who I regard as more honest than your average climate scientist, when asked said that “we need better data.” That means to me he knows the hot spot is not in the current data. I’ll trust him over some anonymous person who has not track record or credentials that can be found.

13. KAP, I’m tempted to say you must be badly ill informed and ignorant of how models actually work. The important mechanisms such as convection, turbulence, etc. etc. etc. are all handled by subgrid models. Climate models use low resolution spatially, so even large scale dynamics can be missed. Subgrid models are always wrong, the question is by how much. They are tuned based on data, often incomplete or noisy. The experts with rigorous training (obviously not you) say that there is far too much dissipation and that the basic equation being solved (without this dissipation) is unstable. In all numerical simulations, this is quite deadly to dynamics and may in fact result in false long term trajectories.

    This doesn’t say climate models are useless, but to me means we are probably investing far too much in them and too little in finding better data and understanding basic mechanisms.

14. I’m more referring to the difference between NH and SH responses. SH is a lot less response despite much lower aerosol forcing. Seems to me this data would result in a lower sensitivity than global data.

15. Re: AMac (Comment #109798)
    February 9th, 2013 at 1:14 pm
    Hi Amac,

    Please have a look at Figure 1 in this previous article to see a graphical explanation of how “effective climate sensitivity” is normally calculated.
    http://rankexploits.com/musings/2012/the-arbitrariness-of-the-ipcc-feedback-calculations/

    “Effective climate sensitivity” is a mathematical construct. Saying that the effective climate sensitivity is increasing (in a given run) is identical to saying that the outgoing flux response to temperature change is curvilinear rather than linear in that run.
    The fact that in a GCM or in my latitude model the effective climate sensitivity does increase with temperature (and time) does not mean that the “climate sensitivity” in 100 years will be different from today. In fact for the latitude model as presented, the temperature response to a forcing X in 100 years will be exactly the same as the temperature response to a forcing X today.

16. So Paul_K, I believe that I have heard that climate models possess little skill at regional climate response. I don’t understand then how this response can be used to validate models. By this test, it would seem that all fail!! If that’s true, then any paper such as the one you review are called into question, n’est pas?

17. David Young #109812

    Gosh, where have you been? It has been known for years that single-instrument MSU data disagree with equivalent massively multi-instrument radiosonde data, due to warm bias in the way readiosondes were built and deployed during some decades of the late 20th century. And although MSU data has insufficient vertical resolution to detect a tropospheric hotspot, radiosonde data does – provided the multi-instrument radiosonde data is homogenized to account for changing instrument construction and deployment over the decades. Doing so finds the elusive hotspot. Just because you don’t like the answer for political reasons, that doesn’t mean the science is wrong. And I’m sorry, but a vaguely-recalled paper without a citation is an insufficient rebuttal. If you don’t like Tichner 2009, try Haimberger et. al. 2008, or Sherwood et. al. 2008, all of which find essentially the same thing. The hotspot “issue” has been resolved, although the news probably never made it as far as WUWT.

18. David Young #109813

    I asked what fundamental mechanisms climate models got wrong. Instead of answering the question, you spend a paragraph arguing that since we don’t know everything, therefore the things we do know must be wrong.

    Fail on both counts.

19. David Young,

    So Paul_K, I believe that I have heard that climate models possess little skill at regional climate response. I don’t understand then how this response can be used to validate models. By this test, it would seem that all fail!! If that’s true, then any paper such as the one you review are called into question, n’est pas?

    Well I agree that GCMs have very little skill at regional climate response. Try “D. KOUTSOYIANNIS, A. EFSTRATIADIS, N. MAMASSIS & A. CHRISTOFIDES “On the credibility of climate predictions” Hydrological Sciences–Journal–des Sciences Hydrologiques, 53 (2008).”

    However, what I am doing here is not trying to validate the GCMs. It is to explain a specific and very important phenomenon which they produce. Having clarity on what produces the phenomenon in the GCMs allows us to then test whether it is real or not, which is why I am suggesting that it is imperative that modelers shift focus from the global aggregate response to careful examination of how well the regional and latitude responses match the real world.

20. I looked at your cited paper and it directly admitted that the data did not show the hotspot and stated that the errors “might” account for the difference. If other papers find the same thing, then the case is closed and you are wrong. But perhaps you subscribe to the usual doctrine that if data and theory disagree, the data must be wrong. As I said before, you look to me like a typical activist who quotes things out of context and claims that papers say what they directly contradict. I won’t take you seriously without some idea of who you are or evidence of some slight regard for the truth. Goodbye

21. Re:HaroldW (Comment #109800)
    February 9th, 2013 at 1:40 pm

    Hi HaroldW,

    While the Arctic may not be a large factor in the overall energy balance or sensitivity determination, it always seemed to me that it was improper to apply equation (1) there. The area of sea ice is not time-invariant, so why should the Arctic behave as a linear time-invariant system?

    Well according to Armour, the Arctic is actually a very large factor, but you are probably right to assert that it doesn’t behave as an LTI. There are two latitude zones in the GCM tested by Armour where the feedback is net positive i.e. the positive feedbacks overwhelm the Planck response. One is the Arctic and the other is the subpolar or sea-ice region of the Antarctic. These can only reach steady-state in the GCMs with a change (net outflow) of sensible heat flux. My toy model is only an approximation for illustration, and it is quite possible that these regions do not conform to LTI.

22. The AOGCM’s don’t exhibit a “exp(-t/tau)” curve. My crappy model finds a cubed root of time a better fit.

    https://sites.google.com/site/climateadj/simple-model-of-models/crap-echam5-1pctto2x.png?attredirects=0

23. Re: Paul_K (Feb 9 21:05),

    There are two latitude zones in the GCM tested by Armour where the feedback is net positive i.e. the positive feedbacks overwhelm the Planck response. One is the Arctic and the other is the subpolar or sea-ice region of the Antarctic. These can only reach steady-state in the GCMs with a change (net outflow) of sensible heat flux.

    Yep. That’s the way it works. The meridional temperature distribution becomes flatter as the planet warms. The tropics won’t warm as much as the forcing there and the poles will warm more. The tropics already emit less LW than the SW they absorb and high latitudes emit more ( see graph ). The crossover latitude is about 40 degrees N and S.

24. Re:-AJ (Comment #109825)
    February 9th, 2013 at 11:03 pm

    Too esoteric, AJ. Is this a functional curve fit to the temperature time relationship or a convolution using estimated forcing data plus a response function with a cube root of time in it? I suspect it is the former, given that your temperature function is monotonic increasing with time. In that case, it really isn’t comparable.

25. Re:-AJ (Comment #109825)
    February 9th, 2013 at 11:03 pm

    Too esoteric, AJ. Is this a functional curve fit to the temperature time relationship or a convolution using estimated forcing data plus a response function with a cube root of time in it? I suspect it is the former, given that your temperature function is monotonic increasing with time. In that case, it really isn’t comparable.

26. DeWitt, when you include the ocean versus land by latitude your chart is more interesting. Let’s see. water is around 4 J/g, moist soil is around 2 J/g and dry soil around 0.8 J/g.

    Since heat capacity is required to maintain OLR, it might be worth considering.

27. Paul_K,
    Thanks for an interesting analysis. As you say, testing based on regional data would seem to be required. But the magnitudes of the differences you use in Lambda and Tau to duplicate the modeled response over time seem rather extreme.
    .
    Thermohaline circulation depends on high latitude cooling of ocean surface waters to ~0C. Therefore, the surface of the ocean is more-or-less “regulated” in temperature to near 0C by convective overturning, at least in the winter months. Warmer surface currents from low latitudes arrive, cool to ~0C, and sink to the deep. At constant ocean surface temperature (eg. ~0C), the rate of heat loss ought to be fairly constant, at least over the ocean. There is a summer-time only stratification in many high latitude parts of the ocean, which allows the rapid seasonal warming that is observed.. the heat capacity of the ocean is reduced in summer by stratification. I would expect substantial wintertime ocean surface warming at high latitudes to reduce thermohaline circulation.
    .
    DeWitt,
    IIRC, that nifty graphic comes from a discussion we had over at Jeff’s. Maybe an interesting extension would be a graph of emission versus sine latitude along with average temperature versus sine latitude to get some indication of the emission versus temperature function.

28. Tichner 2009:

    “Based on the available evidence we cannot reject the underlying hypothesis that under recent climate change the average tropospheric relative humidity in the Northern Hemisphere remains unchanged while temperature and specific humidity increase.”

    http://journals.ametsoc.org/doi/full/10.1175/2009JCLI2879.1

29. Re: SteveF (Feb 10 08:12),

    An annual temperature average plot may not be completely relevant. The average temperature range at high latitude is much greater than at lower latitudes. It’s over 30 K near the poles and less than 3 K near the equator. I should probably average T^4 rather than T. A quick search, though, says that average temperature looks like average insolation rather than TOA LW emission.

30. There’s a related article just released in PNAS.

    http://www.pnas.org/content/110/6/2058.abstract

    Ka-Kit Tung and Jiansong Zhou
    Department of Applied Mathematics, University of Washington, Seattle, WA 98195

    The observed global-warming rate has been nonuniform, and the cause of each episode of slowing in the expected warming rate is the subject of intense debate. To explain this, nonrecurrent events have commonly been invoked for each episode separately. After reviewing evidence in both the latest global data (HadCRUT4) and the longest instrumental record, Central England Temperature, a revised picture is emerging that gives a consistent attribution for each multidecadal episode of warming and cooling in recent history, and suggests that the anthropogenic global warming trends might have been overestimated by a factor of two in the second half of the 20th century. A recurrent multidecadal oscillation is found to extend to the preindustrial era in the 353-y Central England Temperature and is likely an internal variability related to the Atlantic Multidecadal Oscillation (AMO), possibly caused by the thermohaline circulation variability. The perspective of a long record helps in quantifying the contribution from internal variability, especially one with a period so long that it is often confused with secular trends in shorter records. Solar contribution is found to be minimal for the second half of the 20th century and less than 10% for the first half. The underlying net anthropogenic warming rate in the industrial era is found to have been steady since 1910 at 0.07–0.08 °C/decade, with superimposed AMO-related ups and downs that included the early 20th century warming, the cooling of the 1960s and 1970s, the accelerated warming of the 1980s and 1990s, and the recent slowing of the warming rates. Quantitatively, the recurrent multidecadal internal variability, often underestimated in attribution studies, accounts for 40% of the observed recent 50-y warming trend.

31. Carrick (#109833) –
    I haven’t read that paper (paywall), but it sounds a lot like Zhou&Tung, “Deducing Multidecadal Anthropogenic Global Warming Trends Using Multiple Regression Analysis”. There was an excellent discussion of the latter at Marcel Crok’s blog here and here, and was also mentioned at the time at WUWT.
    .
    Tamino predictably trashed the paper at his blog: “When fools fools themselves” and later

    I find the whole “AMO is responsible for part of the global temperature change” idea to be mathturbation of the worst kind, exploiting obvious correlation between temperature and temperature to fly in the face of the laws of physics.

    .
    Tamino calling someone else out on “mathturbation” is, almost needless to say, a bit rich. That said, though, I find the paper about as unconvincing as Foster&Rahmstorf. Even if Z&T’s estimate of anthropogenic warming is accurate, the AMO isn’t an explanation for climate behavior; at best it’s an indicator of some as-yet-unexplained process.

32. Haroldw, “Even if Z&T’s estimate of anthropogenic warming is accurate, the AMO isn’t an explanation for climate behavior; at best it’s an indicator of some as-yet-unexplained process.”

    The process has been pretty well explained by Toggweiler et al. There is a shifting “thermal equator”. The oceans and land are asymmetric. The AMO just highlights the Atlantic portion. There would also be a “thermal meridian” which ENSO somewhat emulates. Earth is no ideal anything but a place to live.

    http://www.eldoradocountyweather.com/current/satellite/goeseast-wv.php

    It is fun to watch though.

33. Dallas (#109835) –
    Can you please be more specific about the Toggweiler et al. reference? I find lots of papers, but no titles hinting at AMO, 60-to-70-year oscillations, or “thermal equators”.
    .
    The patterns *are* fun to watch. 🙂

34. Re: SteveF (Feb 10 08:12),

    I took the 1961-1990 gridded absolute temperatures that I found somewhere I can’t remember now and calculated surface emission for the monthly average temperature at each 5 degree latitude band and averaged that for the year. This graph is that data plus TOA emission versus sine latitude. Looks like a pretty non-linear response to me, as would be expected by the large variation of total column water vapor with latitude.

35. HaroldW, Toggweilder doesn’t specify a frequency just an estimated mixing rate of up to 150 years and estimated impact of ~20 sverdrup for surface velocity changes at the ACC.

    He also has a short paper on how the “thermal equator” impacts equatorial “westerlies” which are linked to the Quasi-biennial Oscillation/ENSO.

    http://sam.ucsd.edu/sio219/toggweiler_bjornsson.pdf

    That is the estimate of internal oscillation.

    http://www.gfdl.noaa.gov/bibliography/related_files/jrt9502.pdf

    That is info on their model and the estimated mixing efficiency.

    http://www.gfdl.noaa.gov/bibliography/related_files/jrt0901.pdf

    And an article on shifting westerlies.

    The impact of the shifts is roughly 4 Wm-2 (my questimate), but it depends on a lot of stuff.

36. Toggweiler reference.

    https://140.208.31.101/bibliography/related_files/jrt0901.pdf

    When I found this, Firefox warned of an untrusted connection, but whereis yielded this:

    Whois information (140.208.31.101)
    NetRange 140.208.0.0 – 140.208.255.255
    CIDR 140.208.0.0/16
    OriginAS
    NetName GFDLNET
    NetHandle NET-140-208-0-0-1
    Parent NET-140-0-0-0-0
    NetType Direct Assignment
    Comment http://www.gfdl.noaa.gov
    Comment Operations Staff – 24×7 (609)452-6560
    Comment Security related issues can be mailed to oar.gfdl.itso@noaa.gov
    RegDate 1990-06-11
    Updated 2010-04-08
    Ref http://whois.arin.net/rest/net/NET-140-208-0-0-1
    OrgName NOAA Geophysical Fluid Dynamics Laboratory
    OrgId GFDL
    Address P.O. Box 308
    Address Forrestal Campus – US Route 1
    City Princeton
    StateProv NJ
    PostalCode 08542
    Country US
    RegDate 1990-01-24
    Updated 2008-02-04
    Ref http://whois.arin.net/rest/org/GFDL
    OrgTechHandle LLE17-ARIN
    OrgTechName Lewis, Larry
    OrgTechPhone +1-609-452-6568
    OrgTechEmail Larry.Lewis@noaa.gov
    OrgTechRef http://whois.arin.net/rest/poc/LLE17-ARIN
    RTechHandle LLE17-ARIN
    RTechName Lewis, Larry
    RTechPhone +1-609-452-6568
    RTechEmail Larry.Lewis@noaa.gov
    RTechRef http://whois.arin.net/rest/poc/LLE17-ARIN

    https://140.208.31.101/bibliography/related_files/jrt0901.pdf

37. HaroldW:

    Tamino calling someone else out on “mathturbation” is, almost needless to say, a bit rich. That said, though, I find the paper about as unconvincing as Foster&Rahmstorf.

    Yes it is a bit rich, given the unphysical assumptions made by F&R ,that he would turn around and trash another paper, no doubt because he didn’t like the message, rather than because he had substantive criticisms of it.

    Some people are very predictable, he being one.

    This is a different paper though.. just in print. The title which I neglected to include was “Using data to attribute episodes of warming and cooling in instrumental records.”

38. jim2, that’s weird.

39. Paul K

    Thanks for taking this on in more detail. I don’t have any time to review Armour or your post in detail, but I do have a few thoughts to kick around until then.

    1) in Armour, they show an initial downward trending Teff which they don’t explain (based on my skimming)

    2) after 300-years of simulation, Teff is asymptotic, yet still ~0.2C below T2x.

    Regarding Carrick’s pointer to Ka-Kit Tung and Jiansong Zhou, my comment is to look at the Holocene temperature plot on wikipedia. It shows an oscillating downward trend over the past 8K years, so having periodic ups and downs is to be expected in geology.

    This brings me to my final point.

    If we were to expect the Holocene downward trend to continue and we have overprinted on top of this anthropogenic forcings that started with carbon black with the industrial revolution, then increasing water vapor via irrigated agriculture starting in the early 20th century, then the combo of CO2 increases and O3 decreases of the post-war era, it is easyh for me to understand why the models and predictions are way off.

    On the one hand, maybe the Teff(t) for CO2 is pretty good, but since it’s overprinted on a long-term cooling trend, it looks like it’s wrong. However, when you see that warming started way before CO2 and O3 showed up, one has to suspect that other anthro forcings with entirely different regional feedback responses that change with time and temperature, etc.

    My hunch is that the sensitivity for CO2 is overestimated and other anthro forcings which likely play a large role in arctic feedbacks are underestimated. Also, the long term trend of downward Holocene GAT coupled with short term oscillations makes figuring out transient and equilibrium climate responses to forcings a very very comp-licated problem. Also, our current situation cannot be compared with the LGM and previous climate responses.

    Welcome to the rabbit hole known as geology.

    Also, I love your optimistic hope that Armour’s approach will spur the alarmists will lay down in green pastures with the deniers.

40. Paul_K: I’m a little confused by the large number of adjustable parameters in your model (which could permit you to fit almost any set of data).

    If I understand correctly, your regional values of lamba include the Planck feedback (immediate and preset), water vapor, lapse rate and cloud feedbacks (which are fast), and slow uncertain feedbacks (like albedo and vegetation). If I understand the model correctly, the depth of the mixed layer in each region will change the e-fold response time.

    If I’m not too far off-base, it seems to me that short-term observations (such as after Pinatubo) and measurements on the mixed layer place constraints on possible choices for lamba and tau in each region.

41. Oddly enough, no previous AMO cycle in history caused the Arctic Ocean to melt. I wonder what’s different about this one?

42. KAP – you are an awesome historian. History is such a long time ago. Congrats to you!

43. Paul_K, A couple of points. If I understand this correctly, the equilibrium sensitivity is higher because of the slow response of the polar regions. I can imagine albedo feedbacks having something to do with this? Are there other factors? Why would these factors be different than the recovery from the last ice age?

    I read Annon’s paper on LGM temps. He calculates a linear sensitivity of 1.7K but says that nonlinear response to CO2 makes this an underestimate for present day climate. I’m just wondering if this makes sense. The difference between LGM and today is so large that it seems his linear sensitivity should cross any nonlinearity very well. I also don’t understand why CO2 response should be more nonlinear than insulation forcing changes.

    I also second your point that we have no real way of knowing how large this effect is or if the climate models are showing it correctly. I would be concerned that what we are seeing here is a witchhunt in which the models show something, we are not sure its right because models are rather bad anyway, and are searching for a possible simple explanation for it.

44. “given the unphysical assumptions made by F&R”
    Hmm, the paper assumed surface temperature is influenced by GHG, ENSO, solar and volcano. Given the caveats on Solar/volcano in Rypdal (2012), I am interested in knowing what what the “unphysical” assumptions are?

45. KAP:

    Oddly enough, no previous AMO cycle in history caused the Arctic Ocean to melt. I wonder what’s different about this one?

    My advice is don’t go into shock over something nobody here is suggesting.

46. @Phil Scadden: The first incorrect assumption of F&R is that anything OTHER than ENSO, Solar, and Volcano is “the global warming signal”. They’re only looking at 30 years, and there are obvious quasi-cycles of 30 years and longer, so this is a huge assumption.

    The second incorrect assumption of F&R is that the effects of ENSO, Solar, and Volcano are linear, that their limits are observed in the 30-year period, and that all interactions and phases that might result from them have been observed in the 30-year period.

    The third incorrect assumption of F&R are the measurements they use for ENSO and Solar. ENSO is more complex than a single number, and their Solar data series is known to have problems in recent times. They ignore the AMO, which is part of the discussion in this thread as well.

    The fourth blunder is in the last sentence of their paper, where they extrapolate their results for 30 years into the future. Based on only 30 years of data. No, they don’t give a numeric result — they simply say that their results will be valid for decades to come.

47. The paper examines what is left when you remove effects from well-understood physical processes. It finds the result “consistant” with what you expect from physics of GHG warming. It rightly leaves out quasi-cycles without established physics as its contentious whether these are unforced cycles at all.

    I would say you assume near-linear unless you had strong evidence for non-linearity because otherwise you risk over-fitting. ENSO is more complex than one number but then not all noise is accounted for either.

    http://www.skepticalscience.com/news.php?n=1824 has someone pushing this back through 130 years, (actually Rypdal rather than F&R) though this is clearly work in progress and not yet published.

    It isnt clear to me that AMO drives rather than follows climate but I interested to see papers that contradict that view.

48. Phil, the biggest problem from my perspective is that they assume a linear, delayed relationship between the forcings and global mean temperature.

    That’s a whopper.

49. You can look at correlation between ENSO3.4 and zonally averaged temperature:

    figure

    to see that a single lagged relationship does not properly describe the relationship between ENSO variability and the climate response.

50. The SKS article I pointed to above by the way assumes exponential response to look at issues highlighted by Rypdal, and also calculates effects of adding in AMO and PDO.

51. Re: KAP (Feb 10 17:43),

    1. The Arctic Ocean hasn’t melted yet. Even last year at the minimum there was still 3.49 Mm² of Arctic sea ice extent.
    2. There was, in fact, significant melting on the Atlantic side of the Arctic Ocean starting in 1918 and lasting into the 1940’s. Guess when the AMO started increasing and when it started decreasing again. A sine wave fit to the AMO index with a period of 66.6 years reached a minimum in 1911 and a maximum in 1944. See also Chylek, et.al. 2009.

    Our analysis suggests that the ratio of the Arctic to
    global temperature change varies on multi-decadal time scale. The commonly held assumption of a factor of 2–3 for the Arctic amplification has been valid only for the current warming period 1970–2008. The Arctic region did warm considerably faster during the 1910–1940 warming compared to the current 1970–2008 warming rate (Table 1). During the cooling from 1940–1970 the Arctic amplification was extremely high, between 9 and 13. The Atlantic Ocean thermohaline circulation multi-decadal variability is suggested as a major cause of Arctic temperature variation.[my emphasis]

52. Carrick – that is an interesting figure. What is the year range covered by that analysis? Its certainly interesting in light of the Armour paper but somewhat more tricky to use in a statistical model.

53. In terms of physical processes associated with ENSO, this would seem to be expected. However, I am not convinced that this invalidates a model of using single lag in calculating the effect on the global temperature anomaly. You could partition the data and look to see if significant differences in lag between partitions but my quartaloos would be very little difference in the calculated lag.

54. Dallas, jim2 –
    Thanks for the references. I’ve added them to my reading list.

55. Phil, I started with 1960 and ended in June, 2012. I used HadCrut3 5°x5° gridded data to generate the zonally averaged data & the ENSO3.4 series.

    My suggestion for a simple intermediate solution would be to break out the temperature series in zonal bands, use a simple lagged constant for each band for each forcing (essentially F&R zonal-band by zonal-band), then recombine the residuals into a new global series.

56. Phil Scadden, you can get notably different lags via any number of simple changes to F&R’s methodology. Heck, Tamino (Grant Forster) published an updated version of his code that gets notable differences. And that’s ignoring the glaring problem of F&R’s lags: Inexplicable differences in lags exist amongst the data sets.

    Nevermind there isn’t the slightest shred of statistical validation for the calculated lags. The fact one set gets the best AIC score is meaningless if you don’t examine how close the score is to that of other sets. Once you do that, it becomes evident the lag fitting is a joke as lag sets with six month differences have near-equal AIC scores.

    I think it’s weird an obviously flawed approach is accepted (mostly) without question by so many people when the authors didn’t do the slightest thing to justify it.

57. Re: RB (Comment #109810)
    February 9th, 2013 at 5:48

    Hi RB,
    I broadly agree with the position you express in your first paragraph.
    You then wrote:-

    Where Armour seems to diverge from Held is that this curvilinear response arises from polar amplification, not ocean mixing in high latitudes – do I understand you correctly?

    The divergence between the Armour explanation and the Held explanation is much more fundamental. It is the difference between effect and cause. The paper that Held co-authored with Winton in 2010 is (politely) taken apart piece by piece in Armour 2012, and correctly so in my opinion. The fact that there is curvature in the relationship between net flux and temperature for a fixed step-forcing means de facto that the rate of ocean heat uptake is changing with time, since the flux balance is changing with time. The cause of the curvature however is not that the rate of ocean heat uptake is changing with time; this is an effect. The cause of the curvature is that there are some regions which have a very low surface response to increased temperature (total radiative feedback is small in magnitude) and they respond very slowly in time (they have high values of tau). These regions may respond slowly in time because they have a higher-than-average heat capacity, as seems to be the case with the Southern Oceans, or for other reasons such as the long timeframe for ice-melt and adjustment of meridonial currents. But the point is that this is then sufficient to explain the curvature in flux response. It is really worthwhile reading the full analysis of the Winton and Held 2010 paper in the Armour 2012 paper.

58. Re HR (Comment #109805)
    February 9th, 2013 at 3:25 pm

    “Paul_K it seems like paleo estimates of climate sensitivity are also thrown into question by this work, another point to add to your inferences.”

    I’m not sure about this. If estimates are made over a sufficiently long time-frame, then they should be somewhat immune to the problem introduced by a nonlinear flux response. They are however subject to other data uncertainty problems.

59. SteveF
    “But the magnitudes of the differences you use in Lambda and Tau to duplicate the modeled response over time seem rather extreme.”
    They seem pretty extreme to me too. I don’t suggest that they are correct! Just that you can duplicate the curvature this way.
    Paul

60. Re: Frank (Comment #109844)
    February 10th, 2013 at 5:26 pm

    Hi Frank,

    Paul_K: I’m a little confused by the large number of adjustable parameters in your model (which could permit you to fit almost any set of data).

    As I emphasised I am not trying to show a right answer here just to offer some insight into an effect.

    If I understand correctly, your regional values of lamba include the Planck feedback (immediate and preset), water vapor, lapse rate and cloud feedbacks (which are fast), and slow uncertain feedbacks (like albedo and vegetation). If I understand the model correctly, the depth of the mixed layer in each region will change the e-fold response time.

    That’s essentially correct, although there are some additional physical explanations for slow response besides just the difference in heat capacity.

    If I’m not too far off-base, it seems to me that short-term observations (such as after Pinatubo) and measurements on the mixed layer place constraints on possible choices for lamba and tau in each region.

    Yes they clearly do. Unfortunately, such observations only weakly constrain these parameters for the high latitude regions, which is why they may only yield lower bound estimates of climate sensitivity.

61. Re:David Young (Comment #109847)
    February 10th, 2013 at 6:03 pm

    Paul_K, A couple of points. If I understand this correctly, the equilibrium sensitivity is higher because of the slow response of the polar regions. I can imagine albedo feedbacks having something to do with this? Are there other factors? Why would these factors be different than the recovery from the last ice age?

    This is a bit loosely put, David. To be pedantic, the slow response of the polar regions on its own shouldn’t change equilibrium sensitivity – just the time it takes to get there. It is the combination of the fact that the response is slow and the outgoing radiative response is relatively weak that causes the curvature in the aggregate global net flux response. The factors controlling the strength and speed of temperature response are different between the different GCMs, but the common elements appear to be (for strength of response) the fact that the Planck response starts from a small negative value and is overwhelmed by albedo change and positive lapse rate feedback; this leads to “polar amplification”. For speed of response, it is mostly that ice-albedo change takes place very slowly with changing temperatures, high heat capacity associated with the polar oceans, especially the Southern Oceans and counter-current changes to meridonial heatflow which moderates the rate of change. Should these factors be different from recovery from the last ice-age? I don’t know, but it wouldn’t surprise me if the parameters were different between one climate state and another.

    I read Annon’s paper on LGM temps. He calculates a linear sensitivity of 1.7K but says that nonlinear response to CO2 makes this an underestimate for present day climate. I’m just wondering if this makes sense. The difference between LGM and today is so large that it seems his linear sensitivity should cross any nonlinearity very well. I also don’t understand why CO2 response should be more nonlinear than insulation forcing changes.

    I am puzzled by this comment. I could not find anything similar in the Annan paper, although he did caveat his results with (a) the limited ability of intermediate complexity models to
    adequately represent the spatial pattern of temperature changes which is related to what we are now discussing and (b) the problem of forcing efficacies. Can you provide a quote, perhaps?

    I also second your point that we have no real way of knowing how large this effect is or if the climate models are showing it correctly. I would be concerned that what we are seeing here is a witchhunt in which the models show something, we are not sure its right because models are rather bad anyway, and are searching for a possible simple explanation for it.

    I agree, but I don’t think we afford to ignore it, since we might now reasonably expect the effect to exist to a greater or lesser extent from simple physical principles.

62. AMO doesn’t drive anything, it’s just a (detrended) SST of North Atlantic Ocean. Global temperature indices should be detrended and renamed Global Multidecadal Oscillation (GMO). It wouldn’t be very scientific, but at least consistent.
    http://www.woodfortrees.org/plot/esrl-amo/plot/hadcrut4gl/detrend:0.755/plot/hadcrut4gl/detrend:0.755/trend

    Any climate region will show similar oscillations.

63. DeWitt, SteveF
    Here is a useful paper by Zelinka showing the estimated sensible heat fluxes from the AR4 models.
    http://www.atmos.washington.edu/~dennis/Zelinka&Hartmann_2012.pdf

    David Young,
    The same paper calculates feedbacks by latitude for the AR4 models, so you can see whic effects are common and which tend to vary between the models.

64. Re:DeWitt Payne (Comment #109855)
    February 10th, 2013 at 8:02 pm

    DeWitt,
    The data in the Chylek reference is very interesting. For the curvature in net flux response due to the Arctic to be “real” requires a very slow response time to get to steady-state. A very slow response time implies a very subdued amplitude response to high frequency variation. In the GCMs, the 60 year temperature cycle is a long wavelength/low frequency cycle relative to the response times over most of the planet, but a short wavelength/high frequecny cycle relative to the long putative response times for the Arctic.
    The Chylek data shows a large relative amplitude response in the Arctic. Ergo it is behaving like a single-pole system with a relatively short response time, or more likely it is behaving like a 2 pole system with a fast feedback response and a much slower longterm response. Either way this looks like good news.

65. Re: Carrick (Comment #109808)
    February 9th, 2013 at 3:45 pm

    Hi Carrick,
    I agree – the temperature amplitude response in time is a function of the frequency of input forcings.
    Have a look at my response to DeWitt above re the Chylek data in the Arctic. Any thoughts on the reason for the high amplitude of variation in the Arctic?

66. Phil Scadden (Comment #109851)
    February 10th, 2013 at 7:31 pm
    blockquote>The paper examines what is left when you remove effects from well-understood physical processes. It finds the result “consistant” with what you expect from physics of GHG warming. It rightly leaves out quasi-cycles without established physics as its contentious whether these are unforced cycles at all.
    Phil,
    Your last sentence is illogical. If they are unforced cycles, then they need to be abstracted before looking at the effect of the “known” exogenous forcings. If they are forced cycles, say net heat flux cycles into the mixed layer, then it is possible to estimate the equivalent input forcings by inversion of the temperature dataset, and their effect still needs to be accounted for before assessing the effect of other exogenous forcings. Whichever approach is adopted the temperature gradient through the later part of the 20th century after accounting for these cycles ends up around 0.1 deg/decade.
    The alternative – the Kevin C or F&R approach – is to deny the existence of the cycles altogether. The consequences on a short term dataset are not very evident. However the clear evidence of model mis-specification becomes readily apparent when this flawed methodology is applied to any longer-term datasets.

    The Rypdal work which has been published does not address this issue at all. It focuses on demonstrating that the apparent amplification of solar insolation in the temperature record can be attributed to the coincidence of volcanic cooling during expected lows of solar cycles.

67. re:Kenneth Fritsch (Comment #109811)
    February 9th, 2013 at 6:43 pm

    Hi Kenneth,
    “So what else is new?”

    Maybe not a lot for you since I know you have been following the debate! But I suspect some of this is new for some people.

    The additional thing which is new for me is that it paves the way towards establishing a new suite of necessary conditions for validating temperature forecasting in GCMs, a suite which needs to be included in the CMIP (project). It is no longer enough to say that a GCM honours global aggregate data in history and then handwave away the poor match to regional and zonal behaviour. This paper reveals that accurate matching of each of the zonal latitude bands is a quite fundamental necessary condition for long-term forecasting of global mean temperature.

    Have you seen this demonstrated before? A serious question.

68. Paul_K, I apologize for formatting difficulties in the copied text below.

    From Annon’s conclusion section:

    A first order estimate of the equilibrium climate sensitivity can be
    provided as the ratio of temperature change to radiative forcing. Our new temperature25
    anomaly of 3.9
    ± 0.8◦ C, combined with estimated forcing of 6–11W m−
    2
    (Annan et al.,
    2005; Jansen et al., 2007) would suggest a median estimate for the equilibrium climate
    sensitivity of around 1.7 ◦ C, with a 95 % range of 1.2–2.4 ◦ C. However, such a simplistic
    estimate is far from robust, as it ignores any asymmetr y or nonlinearity which is thought
    to exist in the response to different forcings (Hargreaves et al., 2007; Yoshimori et al.,
    2011). The ratio between temperature anomalies obtained under LGM and doubled
    CO2 conditions found in previous modelling studies varies from 1.3 (Schmittner et al.,
    2011) to over 2 (Schneider von Deimling et al., 2006a). Understanding and quantifying
    the relationship between past and future climate changes remains a major challenge,
    but our robust estimate of temperature change at the LGM, based on current under-
    standing of proxy data, is an impor tant step towards this goal.

69. North Atlantic ocean heat content 1955 to 2012. Cycling down now.

    http://s7.postimage.org/7t3fvl17v/North_Atlantic_OHC_Dec_2012.png

    North Atlantic OHC down to 400 metres back to 1900 – the red and blue lines – looks like the AMO cycle to me.

    http://s14.postimage.org/dq96zg4b5/North_Atlantic_Temp_400m_01.jpg

    And then North Atlantic OHC back to 1960 which goes all the way to the bottom in a recent paper. Cooling below 2000 metres.

    http://s10.postimage.org/4cn5joaux/ngeo1639_f1.jpg

    Tamino gets spanked in Zhou and Tung’s new paper and the North Atlantic really does have some deep-seated cycles and variability which influences the global climate.

70. re:David Young (Comment #109872)
    February 11th, 2013 at 8:23 am

    Thanks, David. No problem with the formatting.

    Yes I did see that in the Annan paper, but interpreted it differently from you I think.
    From the referenced Hargreaves paper which is here:-
    http://www.clim-past.net/prefaces/preface4.pdf

    In addition, an asymmetry in climate
    sensitivity, calculated by decreasing rather than increasing
    the greenhouse gases, with 80% of the ensemble having a
    weaker cooling than warming, was found.

    Annan also includes in his paper the comment:-

    While part of this discrepancy [in LGM temperature anomalies] may be due to methodological
    differences (in particular the limited ability of intermediate complexity models to
    adequately represent the spatial pattern of temperature changes)…

    I believe that this is the “asymmetry or nonlinearity” to which Annan refers, rather than to a nonlinear response to CO2 per se.

71. Carrick et al,

    Here is a pdf version of the Tung and Zhou article:-

    http://depts.washington.edu/amath/research/articles/Tung/journals/Tung_and_Zhou_2013_PNAS.pdf

    I don’t like the article even though it arrives at an answer not too dissimilar to my own. Many of the criticisms that should be leveled against F&R should also be made against this paper – with the exception of ignoring the multi-decadal oscillation.
    But when they do include the multidecadal oscillation they do so by including the AMO as a proxy regression variable. (Uggh!) The AMO may already be carrying some component of late 20th century heating due to a forced response, and cannot be subtracted out like this. I recall that Zeke produced a Blackboard article making this same point. Also it is not necessarily a good proxy for global temperature oscillation.
    They do however get some marks for proving the existence of the oscillation over almost 400 years.

72. Re: Edim (Feb 11 02:40),

    AMO doesn’t drive anything, it’s just a (detrended) SST of North Atlantic Ocean

    Correct, but you’re missing the point. The AMO index appears to be a proxy for the strength of the Atlantic thermohaline circulation. It’s also highly correlated to the entire northern hemisphere temperature anomaly, not just the north Atlantic. NH temperature fluctuations drive the global temperature in the short term because the SH with it’s much higher percentage of ocean area has a much longer time constant. So yes, it’s not a driver per se, it’s a proxy for the actual driver. It’s also pretty clear that the AMO has been present since before the industrial revolution so it’s unlikely to be anthropogenic in origin. The GCM’s have to use an aerosol kludge to get even close to twentieth century temperature behavior and still don’t show the rapid rise in the early twentieth century.

    This should give you a clue that there are significant physics that aren’t included in the GCM’s.

73. Paul_K (Comment #109871)

    “This paper reveals that accurate matching of each of the zonal latitude bands is a quite fundamental necessary condition for long-term forecasting of global mean temperature.

    Have you seen this demonstrated before? A serious question.”

    Paul_K, I would agree that a realization that the models must be able to resolve spatially where the models have not previously is a major change from looking at global averages. I also find that exciting and appreciate very much your efforts to demonstrate and explain these effects to interested laypersons.

    My smart aleck remark was aimed more at the general propositions in your post that the models do not agree and that we cannot rely on model agreement with the observed historical data to predict the future. Being able to propose a reason why this is the case is new. Has not Isaac Held proposed something along these same lines recently?

    As an aside Paul, I find your writing style very easy to follow and almost lyrical at times. I like to guess a person’s occupation/profession based on how they write. I would suspect that at least part of your job is explaining technical issues to fairly intelligent non technical people. Your polite methods of scolding would indicate some position of authority.

74. KAP:

    “I asked what fundamental mechanisms climate models got wrong. ”

    err calculating absolute temperature correctly.
    err calculating the response to volcanic forcing ( in every case they over estimate the cooling response )
    err calculating ice loss in the arctic

    and now of course the argument will be that these are not fundamental or something like you havent pointed out the exact error. The first is silly quibbling about what counts as fundamental, and the second is just head in the sandism. All models are wrong, but unless you can find the error that we cant, then the error cannot be fundamental.

    In truth, all models are wrong. The current models get some fundamental metrics wrong. Take absolute temperature as an example. These errors in fundamental metrics may stem from

    1. compounding small errors
    2. fundamental errors

    Arguing that the errors are not fundamental unless one can find them and point them out, is a silly approach. One can just as easily say that we assume fundamental errors until the small compounding errors are identified. The point is the models get a fundamental metric wrong. I think it is not anything to be proud of, and diagnosing the reason seems to be beyond the capabilities of those who claim that it can’t be a fundamental error. The typical engineering approach of having independent teams find the errors might be a suggestion worth trying. The planet is at stake after all.

75. Paul_K:

    They do however get some marks for proving the existence of the oscillation over almost 400 years.

    This is the part I thought was most interesting.

76. Interesting to see Kenneth’s compliment of Paul_K’s writing style. I agree, except sometimes he can be a bit harsh on people who come up with the wrong answers in the literature. Nevermind, full speed ahead!

    Brandon – Tamino is “Foster”, not “Forster” a distinction which may be important when skimming a climate blog – viz. “Foster and Rahmsdorf” vs. “Gregory and Forster”. I probably have a typo in there somewhere too.

    I don’t have time to re-read the Armour paper in depth but I haven’t heard anyone comment about clouds on this thread yet except in passing. Most of the discussion seems focused on the polar regions as well. I’m not sure if “high latitudes” includes say 60 south, or how much Armour gets into clouds, but recall that there has been discussion of the nonlinearity in spatio-temporal feedback distribution related to clouds ; e.g. Held’s original post on this subject and a “dispatch from AGU” from John Nielson-Gammon relating to a poster explaining temporal changes in ocean currents due to (saturation of?) ocean heat uptake yielding a cloud feedback that went from negative to positive as equilibrium was reached. This fairly well screams “Dig Here, there Be Witches!” (h/t David Young) because right or wrong, it’s a mystery since unlike arctic ice coverage we have scant data to verify anything like this.

77. BillC, thanks. I don’t know why, but I mix those two names up all the time.

78. Brandon – I have in the past and I’m guessing others do. I only bother correcting others’ typos for this sort of reason.

79. Paul-k, Not sure I believe the ensemble data based on I guess simplified models. Certainly the albedo feedback is equally important from LGM to 1880 as it is going forward.

80. steven mosher,

    All models are wrong, but unless you can find the error that we cant, then the error cannot be fundamental.

    There is indeed a fundamental error in models: they arbitrarily adjust the temperature of the added CO2.

    Or, someone is able to demonstrate that this is physically justified?

81. phi

    ‘There is indeed a fundamental error in models: they arbitrarily adjust the temperature of the added CO2.
    Or, someone is able to demonstrate that this is physically justified”

    huh. Well I can say that in my review of modelE which I started back in 2007 when gavin pointed me at the code that I can eliminate this as a potential error. There is no adjustment of the temperature for adding C02. But perhaps I missed it. I could be wrong and I don’t want to fool myself. Could you point me at the routine in the code that does this? What I found is a physical model, tested and verified against field data, that calculates the the radiative effect of C02 and other atmospheric constituents. I did not find what you describe. If you are certain that you have found this error, please point me to the code that makes this error.

    Let me put it in a way you can understand. If you were to allow me to adjust temperature for adding C02, then I could hindcast perfectly. A true skeptic actually looks for errors rather than compounding errors by arm waving.

82. phi,

    you mean they assume the CO2 to take on the average temperature of the atmosphere?

    it’s been pointed out to me that there are something like 10^9 molecular collisions during the average time it takes a CO2 molecule to emit an IR photon. this is the strong evidence that CO2 takes on the surrounding temperature, unlike some arguments that CO2 is cooler than the surroundings because it can radiate IR. Sure it’s cooler – by a factor of about 1-10^-9.

    is this what you mean?

    wouldn’t hold for the upper atmosphere, I guess, but it should hold for the turbulent troposphere where the mixing occurs.

83. steven mosher,

    This is not an implementation error but a theoretical error. The added CO2 is arbitrarily thermalized. It is illicit because only the existing matter is concerned by the fixing of the temperature profile. Any matter created ex nihilo can only arbitrarily be brought at any temperature as long as no energy interraction takes place with the system.

84. BillC,
    Indeed, at first glance this is a minor problem. In reality, it is fundamental because it is the instantaneous temperature of the CO2 that is used to calculate the radiative effect.

85. I love it when phi gives us physics lectures.

    Just saying.

86. I love when Carrick has no idea but can not stop talking.
    Just to say.

87. @Kenneth Fritsch,
    Thank you for the very kind words, Kenneth. I wish they were true.
    @BillC
    Closer to the truth. I am arrogant, opinionated and hate being proved wrong. Occasionally I lapse into total A**hole status. (Ask Carrick.) I have all the necessary qualifications to seek training as a climate scientist.
    Paul

88. Re: phi (Feb 11 12:44),

    The added CO2 is arbitrarily thermalized. It is illicit because only the existing matter is concerned by the fixing of the temperature profile. Any matter created ex nihilo can only arbitrarily be brought at any temperature as long as no energy interraction takes place with the system.

    Complete nonsense. Matter created ex nihilo, hysterical.

89. Wouhaou, that’s argumentation!

90. “Brandon – Tamino is “Foster”, not “Forster” a distinction which may be important when skimming a climate blog – viz. “Foster and Rahmsdorf” vs. “Gregory and Forster”. I probably have a typo in there somewhere too.”

    I remember the name Grant Foster by way of the sunglasses – and then forget and refer to him as Foster Grant.

91. phi, let’s try this one more time then I’ll stop and let someone take over, or not.

    CO2 comes from combustion reactions, so it is mostly emitted in the form of hot exhaust gas. a combination of radiation and collisions will quickly bring this temperature in line with ambient. thereafter, it might approach a state something like Tambient*(1-10^-9) to account for its one-in-a-billion ability to radiate off heat faster than it accumulates it via collisions. Oh yeah and occasionally grabbing a stray IR photon from the surface which happens even less frequently because of absorption spectrum overlaps with WV and other IR-absorbing gases.

    In the troposphere, it seems reasonable to propose that GHGs are the same temperature as the non-GHGs. And subsequently verify via experiment; e.g. measuring the emitted IR to see that it is, indeed, at that temperature. Since the time scales of local thermal equilibrium and large-scale radiative-convective steady state are separated by many orders of magnitude, I don’t see that there is a problem with this conceptual model.

92. Dewitt I did get a laugh out the ex nihilo argument too, to be honest.

    I’ve never seen this particular drivel before though.

    Is this sky-dragon speak?

93. PaulK, you’re doing fine, keep up the good work.. You have all the qualities to be a scientist in any field, not just climate. 😉

94. BillC:

    In the troposphere, it seems reasonable to propose that GHGs are the same temperature as the non-GHGs.

    More than just reasonable, it’s absurd to argue on physics grounds they should be different.

95. BillC,

    I don’t see that there is a problem with this conceptual model.

    Yes, there is one. The initial temperature differs little from the equilibrium temperature but this difference is of the order of magnitude of the calculated effect. This is why the calculation must be performed instantaneousely.

96. BillC,

    In the troposphere, it seems reasonable to propose that GHGs are the same temperature as the non-GHGs.

    Yes, of course. But these are not the conditions of the radiative calculation.
    The principle is to add CO2 and achieve an instantaneous radiative balance before any energy interraction.

97. Paul_K (Comment #109875),
    .
    Yes, the increase in the North Atlantic surface temperature is “contaminated” with the AGW signal. But that contamination ought to be in large part removed by de-trending of the data. Clearly linear de-trending is imperfect, and a more sophisticated approach to de-trending the data may be better. I find it convincing that the North Atlantic temperature undergoes substantial oscillation around its trend, in concert with changes in global average temperatures, while the remainder of the world’s oceans seem less connected to changes in the global average. This is perfectly consistent with variation in northward heat transport.

98. Paul_K
    “The alternative – the Kevin C or F&R approach – is to deny the existence of the cycles altogether.”
    F&R only considered short term. Kevin C is using forcings for a 130 year approach, which is the approach of saying “cycle” is forced. He also tested significance of putting in AMO and PDO.

    The big problem is determining past forcings. Without tying those down, it is hard to come up with a conclusive statement on cycles. Aerosols are particularly tough but I dont accept the IPCC just fudges them. The magnitudes are tough but the approaches used are better than assuming aerosols are constant or insignificant. I would also say that F&R, Benestad, Rypdal etc approaches make support the position of “its just a natural cycle” to explain away post-1970 warming, somewhat hard to find.

    I commented to Kevin C on Carrick’s interesting graph. It would be interesting to repeat with other data sets. Kevin had this to say:

    “… suggested offline the use of two lags for ENSO – one for the positive phase and one for the negative phase. He was right, it does improve fit.
    The single lag model is certainly wrong – all models are – but it is parsimonious and has considerable skill. I’d certainly encourage anyone who wants to look for better relationships. Carrick’s latitude suggestion is a good one. I also got some interesting results fitting lag by longitude – the event progresses westwards round the world. I didn’t get time to do this with the positive and negative lags separately.”

    In short, seems like a lot more interesting research can be done and this is physically reasonable – unlike mysterious ocean cycle with 1st law violations or cosmological cycles without a physical mechanism.

99. Phil Scadden:

    mysterious ocean cycle with 1st law violations

    Ocean cycles which redistribute heat to latitudes with different T/TOA flux relationships (lambda) don’t violate the first law.

100. Phi, if I understand correctly, you’re saying that the the CO2 added from emissions is arbitrarily thermalized, and that this is a fundamental error?

     If so.. I don’t follow your point. This is a seriously tiny amount of heat. Heck, by definition it will be smaller than the heat released by the burning of the carbon, which is itself way too small to be significant in comparison to the GHG effect.

101. Phil:

     I would also say that F&R, Benestad, Rypdal etc approaches make support the position of “its just a natural cycle” to explain away post-1970 warming, somewhat hard to find.

     Not many here (that are not basis physics challenged) would argue with you on this point.

     (KevinC:)

     “I also got some interesting results fitting lag by longitude – the event progresses westwards round the world. I didn’t get time to do this with the positive and negative lags separately”

     I liked this idea too. What is needed to flesh out longitudinal correlation is a smaller grid, 1°x1° and a shorter time discretization, 5 days would be ideal.

     Regarding the single-lag model, we’ve all heard the admonition to make things as simple as possible but no simpler. I really think in that respect that the single lag model is at least one simplification too far. (I also think more work needs to be done to relate the statistical model to underlying physics, and more work to examine the effect of the F&R procedure on the uncertainty in the residual.)

102. I am all for better models. However, just by looking at the data, you see lag between max/min of ENSO index and global temperature anomaly max/min in the ENSO cycle. Single lag models assume this lag is approximately same, good enough for the model to have skill. A lat/long grid adds a lot more variables, though also a lot more data. I agree it would be interesting to see how much more skill this has compared to simple model. I very strongly doubt that better models will challenge the fundamental assumptions that most of the surface temperature variation can be accounted for by ENSO, solar/volcanoes – and anthropogenic forcings. I’d happy bet money against Scafetta’s cycles (for examples) being a good predictor for future surface temp. I would bet against another ocean cycle having a major influence. I wouldnt bet against Foster. I do not think this is non-physical mathturbation.

103. Phil Scadden (Comment #109902),
     “Aerosols are particularly tough but I dont accept the IPCC just fudges them.”
     .
     Maybe not a strait-out fudge, but damned close. Each modeling team chooses the “aérosols du modèle” which best re-creates the historical temperatures. The inverse correlation of assumed model aerosol history with model diagnosed climate sensitivity is simply ridiculous, and ought to illicit nothing much more than laughter….. yet it seems piously accepted by the climate science community. If Tung and Zhou show that the AMO was present in the central England temperature series long before the industrial revolution, and others show oscillatory behavior in Greenland ice cores over thousands of (pre-industrial) years, are you suggesting we accept absolutely undocumented and mysterious aerosol influences over all of this time just because the modelers deem it true? Please. I think most reasonable people will reject the aerosol kludge for what it is: a pure and hideous kludge.

104. Reply to Paul_K (Comment #109864)

     Thank for the explanation. I gather you have chosen values of lamba and tau that reflect short term observations for low and mid-latitudes. How did you choose these numbers? They seem to reflect little feedback in the tropics and subtropics, but much larger feedback in the midlatitudes where climate change will impact people and agriculture.

     In the mid-latitudes, constraints for lamda and tau might be obtained from seasonal changes in insolation and temperature (not temperature anomal).

105. Re: SteveF (Feb 11 14:41),

     Yes, the increase in the North Atlantic surface temperature is “contaminated” with the AGW signal. But that contamination ought to be in large part removed by de-trending of the data.

     IMO, the contamination is mostly high frequency. It’s things like ENSO and short term weather patterns. That’s why you don’t want to use the AMO index directly as a proxy. You’re better off fitting a sine wave and using that even though the frequency and phase are likely not constant. Or you could simply low pass filter the data. I suspect that the ~66 year period of the AMO index being pretty close to three times the ~22 year solar cycle is not a coincidence. Better de-trending would help too.

106. If you mean aerosols other than volcano, then obviously not, however the operative question is how much influence (IF in fact T&J show that) it is having at moment. Not much looking at Kevin C analysis.

     The best part of this is that the analysis makes simple testable predictions. Want to guess what the surface temperature will be next time we get an El Nino of say 1.8? Or do think you need AMO to nail it?

107. Phil Scadden, just make sure you use the same parameter values (including lags) and data sets when testing things. F&R’s code recalculates these every time, and Tamino has already changed data sets once.

108. DeWitt Payne (Comment #109909)
     February 11th, 2013 at 5:47 pm
     IMO, the contamination is mostly high frequency. It’s things like ENSO and short term weather patterns. That’s why you don’t want to use the AMO index directly as a proxy.
     ———————

     There are times when the ENSO packs a bigger punch than it normally does. Like in 1877-78, 1939 to 1944, 1973-1978, 1997-98, even in 2010 (not really recognized).

     Normally, the ENSO’s impact is just 0.07*Nino3.4(lagged 3 months) but in these cases, it is more like 0.2*Nino3.4

     This is because of the ENSO’s extra impact on the AMO (which only shows up perhaps half the time). Leaving out the high frequency AMO data leaves out these higher frequency impacts which occur despite our belief that it is just a long-cycle phenomenon.

     It is both long-cycle and short-term high frequency.

109. Phil:

     A lat/long grid adds a lot more variables, though also a lot more data.

     Actually you add a larger number of data points than you do model parameters, with little cost in complexity and with the certainty your model is more closely related to the underlying physical system you wish to model. Since the extra algorithmic complexity amounts the additional of a for loop (and an extra index on parameters), I don’t see how one can argue that the single lag global mean temperature model is in anyway a better model.

     I very strongly doubt that better models will challenge the fundamental assumptions that most of the surface temperature variation can be accounted for by ENSO, solar/volcanoes – and anthropogenic forcings.

     What you mean by “most”? I.e., you have a break down of how much variance is left compared to total variance before the procedure? I.e.,is “your strong” doubt based on objective measures or is this a subjective impression?

     If you think about it carefully, F&R is just a type of optimal filter with an particular choice of basis functions. (While the choice of basis functions is rooted in relevant physical processes, the way they are applied to the data I find to be contrived and unphysical. From experience, this can and does lead to “issues”.)

     For filters, reducing variance may come at a cost of the introduction of bias, especially near end points (end point bias of filters being a known and studied problem). Remember this filter isn’t any different than any other low-pass filter, as long as you stay away from the end points. If it can’t perform any better near the end points in terms of not generating a bias in the trend… I’d just use an acausal Butterworth, it’s a lot simpler to implement and much more well studied. Simply because it is “straighter” doesn’t mean it’s more correct. (An OLS fit straight line is even straighter after all. ;-))

     In any case, I don’t see this problem is neatly put away as you are presuming. But, believe it or not, I’d be as happy to be proven wrong.

110. SteveF:

     Maybe not a strait-out fudge, but damned close

     I think it’s not a fudge, more like a circular logic loop. Or put it this way:

     The great thing about forcing histories is there are so many of them, you can pick which one you want to use with your GCM to make your back-cast temperatures look reasonable.

     The great thing about models is there are so many of them, there’s gonna be at least one that uses your forcing history.

     So everybody is happy.

     But maybe I’m a bit cynical on this one.

111. Good point on the lat/long grid. Worth doing.

     Well my “strongly doubt” would be based on based on my understanding of “physically plausible”. “Most” is quantifiable but havent done that calculation – I would if was betting. Your comment on “just a set of basis functions” is taken but I think also testable for physical significance if you partition the data set.

     I wouldnt rule out an oceanic distribution effect compounding an aerosol effect but I dont buy any “aerosols constant or aerosols insignicant” argument. I would say that prior to clean-air type acts, aerosol forcing need to show some relationship to FF burning so dont but the idea that it is completely free fitting. On the otherhand OHC isnt tied down well enough in older data so that models can completely avoid a “fitting”.

     The good news: testable in a short time frame. My intuition (not worth much I know) would also suspect that you should be able sort out “AMO – cause or effect” from Argo data before long.

112. Phi

     “The added CO2 is arbitrarily thermalized.”

     That is easy to prove. Point me to the line of code where you found that. easy peasy. You can look for it in the MIT GCM ( very nice code, good manual ) or ModelE or the UCAR model. The the UCAR model you have to register, but its easy.

113. Phil Scadden,
     ” the operative question is how much influence (IF in fact T&J show that) it is having at moment.”
     .
     Fair enough.
     Nobody really knows how much influence natural oscillatory (or pseudo-oscillatory) behaviors have right now. And some careful investigation into causal influences for ‘natural’ oscillations is called for, so that the influence can be quantified.
     .
     But IMO the weight of the evidence is that natural variation over the past 35 years has exaggerated whatever warming may have taken place due to GHG forcing, and a reasonable estimate is a peak to trough influence of ~0.2C…. which means 0.2C of the warming since 1975 is probably ‘natural’, and the remainder GHG forcing. Which puts the GHG warming closer to 0.1C per decade, not 0.17C per decade. Does this matter? You bet it does, especially when (hugely expensive!) public policy is being based on estimates of future warming.
     .
     “Want to guess what the surface temperature will be next time we get an El Nino of say 1.8? Or do think you need AMO to nail it?”
     .
     The influence of ENSO seems pretty well defined: about 0.09C per degree of change in the Nino 3.4 index, lagged 3 to 4 months, with nearly all of the temperature influence between 30S and 30N latitudes (half the Earth’s surface). Short term effects due to ENSO seem pretty much independent of the influence of long term oscillatory behaviors.

114. Hmm, but if you have a downward natural cycle now operating (as T&J contend), then F&R and similar approaches (esp Kevin C which discounts AMO as primarily a forced response) are going to overestimate the value of future temperature anomalies. Of course to further complicate matters, aerosols may be underestimated. However, do you agree that Argo data should separate a forced response (increased aerosol) from change in AMO?

115. Re:Frank (Comment #109908)
     February 11th, 2013 at 5:30 pm
     Reply to Paul_K (Comment #109864)

     Thank for the explanation. I gather you have chosen values of lamba and tau that reflect short term observations for low and mid-latitudes. How did you choose these numbers? They seem to reflect little feedback in the tropics and subtropics, but much larger feedback in the midlatitudes where climate change will impact people and agriculture.

     Hi Frank,
     It’s probably time to repeat my health warning. As I stated in the main body of the text:-“Please note that my sole objective here is to add insight into how differences in regional temperature gain can combine to define a curvilinear flux response. I am not trying to estimate any “true” climate parameters.”
     Having said that, I used the data from Figure 3b in Armour 2012 as a starting point for estimates of total feedback (including Planck), and adjusted them to match very coarsely the normalised temperatures at equilibrium in Figure 3a of the paper. I adjusted relative response times between the latitudes to yield an approximate temporal order of effect, being tropics/subtropics then Arctic then Antarctic. Maintaining the relative timings between latitudes, I then rescaled the absolute timings to give me the best fit to the observed history using the F2 forcing dataset. There is no reason to believe that these values are particularly meaningful, but they yield a result which is not wildly different from the CCSM4 data tested by Armour 2012. The sole purpose was to demonstrate that we can generate a curvilinear flux relationship from the mathematical form proposed, and to explore its properties a little.

     In the mid-latitudes, constraints for lamda and tau might be obtained from seasonal changes in insolation and temperature (not temperature anomal).

     I agree. Previous attempts to match seasonal variation, at least those which I have seen, have focused on the very small net global change. Armour 2012 suggests to me that a repeat of this approach by latitude zone may be a lot more informative.

116. Another point, if you think AGW is exaggerated over last 35 years, you would expect ENSO contribution to differ with time period. Again, using Kevin C calcs, he shows:
     El Nino (MEI) coefficient
     Foster & Rahmstorf 0.079
     35 year calculation 0.071
     130 year calculation 0.083

117. Phil:

     Your comment on “just a set of basis functions” is taken but I think also testable for physical significance if you partition the data set.

     Yes, I think an approach like would work, once you’ve recognized that the problem area, if there is one, is near the end points.

118. Phil:

     Another point, if you think AGW is exaggerated over last 35 years, you would expect ENSO contribution to differ with time period.

     Actually, we’re talking about a linear summation here, not a bias error. That is:

     T_total(t) = T_AMO(t) + T_AGW(t)

     here’s one version of the fit.

     (The blue line is HADCRUt4, the dashed red line is T_AMO … best fit … and magenta line is T_AMO + a trend line.)

     If you take the AMO seriously, for the period from 1980 to 2000, it would increase the apparent rate of warming, if you attributed all of the warming to anthropogenic of course.

119. Phil Scadden:

     Another point, if you think AGW is exaggerated over last 35 years, you would expect ENSO contribution to differ with time period. Again, using Kevin C calcs, he shows:
     El Nino (MEI) coefficient
     Foster & Rahmstorf 0.079
     35 year calculation 0.071
     130 year calculation 0.083

     F&R calculated a MEI parameter of 0.070. Or 0.083. Or any of four numbers in-between. And that was with basically the same time period. Heck, with just the changes in data over a single year, their MEI parameter increased by ~5%.

     It makes you wonder what would happen if you used F&R’s code on different subsets of their data.

120. Windchaser,

     This is a seriously tiny amount of heat.

     Yes, very low. The issue is not there. The calculation of the effect of the added CO2 is based on the arbitrary initial temperature. To avoid this, it should be left to the CO2 time to thermalize naturally (it would be very short). Unfortunately, in this case, there would remain no more radiative imbalance.

     steven mosher,

     Point me to the line of code where you found that.

     I’ve already said, this is not a problem of implementation.
     Otherwise, ask yourself the question: what is and what should be the temperature of the CO2 added?

     There is none of physically definable. The calculation made ​​by the models is underdetermined, an input data is arbitrary, and the results are arbitrary.

121. Phi. There is none of physically definable.?
     on the contrary. There is. The calculation is not underdetermined it is over determined and the results are not arbitrary.

     Gainsaying is fun.

122. steven mosher,
     If you say so …

123. Paul_K, a couple of questions from novice, if you don’t mind.

     I’m concerned about the derivation of F3. Since it comes “from the inversion of the Hadcrut3 temperature series into the flux domain using a global linear feedback model” doesn’t this mean that all the tests are doing to validating that the model you are testing does the inverse of the inversion? In which case, common mode problems will not be ruled out.

     Secondly, am I right in understanding that the basic problem being tacked here is that “most of the AOGCMs exhibit a curvilinear response in outgoing global flux with respect to average temperature change” and that this is an emergent phenomenon from the models that basically wasn’t actively coded into them by anyone? I.e. what you are trying to determine is whether the model behaviour can be mapped onto some real physical effect?

     This seems odd given that the modellers tell us that all the rules in the models are based on the underlying physics which is well understood.

124. The whole story in a single paragraph.

     Electromagnetism and thermodynamics are not unified science. You can not build a concept in one of the science and use it carelessly in the other. It is licit to distinguish backradiations and radiations under electromagnetism, it is no longer possible in thermodynamics because it would lead to violate the second principle. Modeling the greenhouse effect by backradiations is therefore not allowed. Radiative forcing, which is an emanation of the concept of backradiations is then not a legitimate notion. This is verified by noting that radiative forcing represent an arbitrary value calculated using an arbitrary initial temperature of the added CO2.

125. Re: phi (Feb 12 06:23),

     I’m beginning to suspect you’re not a real person. You’re a sock puppet for Doug Cotton.

126. DeWitt Payne,
     Whether I am a sock puppet or not does not seem to be a very interesting question, I am phi and I’m an anonymous. Otherwise, do you have any comments?

127. phi, you do realize you’re speaking to chemists, engineers and physicists on this blog, don’t you?

     There are working devices that simultaneously use thermodynamics, electrodynamics & quantum mechanics. (See thermoelectric effect.) But I digress.

     You pretty much screwed the pooch with me long ago by admitting that you thought treemometers were more accurate than thermometers (devices actually intended to take measurements, imagine them working worse than tree rings as temperature sensors :shock:). At least that was original, even if still pseudoscience dreck.

     Your most recent little exercise on Jeff’s blog of trying to say what Hansen’s 2001 paper says while showing the inability to actually quoting where it said what you claimed it said, does suggest if you are if not sock puppeting, you’re now into parroting other people’s dreck.

     So seriously, wherever you’re getting this stuff, you need to find another source.

     Next I predict we’ll get an explanation from you that photons aren’t real and that the Earth, indeed, is banana-shaped.

128. PaulK, “I agree. Previous attempts to match seasonal variation, at least those which I have seen, have focused on the very small net global change. Armour 2012 suggests to me that a repeat of this approach by latitude zone may be a lot more informative.”

     Latitude and altitude. Poleward transfer can get complicated.

129. Carrick,

     There are working devices that simultaneously use thermodynamics, electrodynamics & quantum mechanics.

     And then? What I wrote is in contradiction with that? No! You are missing the point.

     …you thought treemometers were more accurate than thermometers…

     No connection. And this is not the view that I defended and still defend. Treemometers are more reliable than thermometers of stations for medium and long term trends. I proved it and I can still prove it. Yet you are missing the point.

     Your most recent little exercise on Jeff’s blog…

     Once again, no connection ! I said on Jeff’s blog that in the current state of scientific knowledge, adjustments were illegitimate because the only paper that explains their prinipal and surprising characteristic concluded at their illegitimacy.

     There may be chemists, physicists and engineers here, the only thing I know is that you, you personally in this comment, show no one of the essential qualities of a scientist

130. phi–
     If it is your intention to be anonymous, you should be aware that you are probably *less* anonymous than most people. I just looked up your home address… and I strongly suspect I have it right. It’s much to far from Chicago for me to stop by, but really… if you are using that email address at other blogs, some other blog host or hostess is going to figure out who you are.

     DeWitt,
     It is highly unlikely phi is Doug Cotton. They aren’t even on the same continent.

131. Phi

     … it is no longer possible in thermodynamics because it would lead to violate the second principle.

     This is incorrect. Back radiation does not violate the 2nd law of thermodynamics. Anyone who thinks so does not understand the 2nd law of thermodynamics.

     Also: While engineering training often involves learning the 2nd law of thermo, some specialties don’t use it much. For example, civil engineers like my father-in-law don’t use it much. Civil engineers should tread carefully when making proclamations about the 2nd law of thermodynamics, particularly if they are being contradicted by s chemists, chemical engineers or mechanical engineers– those are the people who routinely use the 2nd law of thermodynamics when dealing with engineering applications.

132. phi:

     You are missing the point.

     No, not really. The idea they aren’t “unified” is a silly one. We know how to dance simultaneously with thermodynamics, electrodynamics and Q&M just fine. Your help in attempting to further improve insight into their intersections, while appreciated, is unneeded.

     Treemometers are more reliable than thermometers of stations for medium and long term trends

     !? Plants that respond to half a dozen different variables, and to temperature in a non-monotonic fashion make better thermometers than devices that only respond to temperature and do so both monotonically and approximately linearly. (Who would have though. :O)

     I said on Jeff’s blog that in the current state of scientific knowledge, adjustments were illegitimate because the only paper that explains their prinipal and surprising characteristic concluded at their illegitimacy.

     You also said “discontinuities indicate a bias a posteriori” then got all hysterical when I pointed out that statement makes little sense. You couldn’t provide the context yourself and in fact got rather huffy about being pressed on the issue.

     There may be chemists, physicists and engineers here, the only thing I know is that you, you personally in this comment, show no one of the essential qualities of a scientist

     lol. I’m so *crushed*.

133. Lucia,

     I know and it is voluntary. I do not want to reveal my identity, it has no interest. But I do not specifically seek to hide it. I speak as I would in my name. And if by chance, you go in the area, I would be happy to meet you and introduce you to the country.

134. DeWitt,

     Real person or not, it does not matter. He and Doug are two peas in a pod. Trying to dissuade him is not worth the effort.

135. Carrick,

     You might as well refrain from comment, your prose is empty and irrelevant.

136. Re: steveta_uk (Comment #109928)
     February 12th, 2013 at 3:25 am

     Great questions. I’m surprised no-one else called me out on the use of the F3 forcing dataset. I was expecting to be quizzed.
     On the one hand, you can think of F3 as an arbitrary choice which has about the right level of forcing gain over the 20th century to have credibility. I could equally well have chosen the forcing data from NCAR or from GISS; my conclusions would have been the same. In particular, you might note that the F3 dataset has nothing to do with the curvature seen in the step forcing case in Test#4, which is the main point of the article.
     So why didn’t I grab an off-the-shelf dataset for F3? Firstly, it was because I thought I could learn something additional if I used a dataset which I knew produced an exact known answer for temperature under a linear assumption if the correct parameters were chosen. (Have a look at the differences between Test#2 and Test#3.) Secondly I wanted a dataset which carried high frequency forcing so I could compare temperature amplitude attenuation between the fast response tropics and the slow response poles.

     Secondly, am I right in understanding that the basic problem being tacked here is that “most of the AOGCMs exhibit a curvilinear response in outgoing global flux with respect to average temperature change” and that this is an emergent phenomenon from the models that basically wasn’t actively coded into them by anyone? I.e. what you are trying to determine is whether the model behaviour can be mapped onto some real physical effect?

     This is essentially correct. It was observed in the early 90’s that “effective climate sensitivity” apparently increased with time and temperature in most of the GCMs. In addition, it was recognised that ECS values from slab-ocean models or atmosphere-only models gave ECS values which were different from their fully coupled counterparts. This did not prevent the publication of hundreds of papers which relied on the assumption of a constant (global) linear feedback. Almost two decades later there still wasn’t a clear explanation for why this happened in the models. Armour 2012 summarises as follows (Teff = effective climate sensitivity):-

     While the time-dependence of Teff has been widely
     demonstrated, there is little agreement on the magnitude or
     mechanism of its variation. Senior and Mitchell (2000) sug-
     gest that time-dependent cloud feedbacks arise from inter-
     hemispheric warming differences associated with the slow
     response of the Southern Ocean. Williams et al. (2008)
     instead argue that the time-dependence of Teff can be
     largely accounted for by the use of an `effective forcing’
     in Eq. (4). Recently, Winton et al. (2010) have proposed
     an alternative interpretation of Teff in terms of a time-
     dependent `efficacy of ocean heat uptake’, analogous to
     the distinct efficacies of different radiative forcing agents
     wherein each may drive a different global temperature re-
     sponse (per unit global radiative forcing) depending on its
     geographic forcing structure (e.g., Hansen et al. 1997, 2005;
     Yoshimori and Broccoli 2008).

     So, yes, the objective is to try to understand why it happens in the models and then to test whether it has physical validity. However, none of this challenges the underlying physics in the model. That comes later.

137. phi, that was almost as good as

     “my hair is a bird, your arguments are invalid.”

     I’d suggest using that one next time. It’s way cooler.

138. SteveF,

     I raised a number of statements that are not trivial and should be easily refuted if wrong. Why not just do it instead of holding disparaging remarks unworthy of scientists?

139. phi–

     And if by chance, you go in the area, I would be happy to meet you and introduce you to the country.

     Thanks. But I’d probably stay with my Aunt Lydia like I did the last time I visited your country. Honestly, for vacations, I prefer a warm sea shore. 🙂

     I do not want to reveal my identity, it has no interest. But I do not specifically seek to hide it. I speak as I would in my name.

     You used to use your real name– I found it on a 2010 comment at JeffIds.

     Phi/DeWitt/Carrick,
     Tone stuff down. Phi is saying quite a few obviously wrong things about the 2nd law and so on. It might be best to just ignore him.

140. Lucia,

     Phi is saying quite a few obviously wrong things about the 2nd law and so on.

     The problem is that you would be well unable to prove it.

     It might be best to just ignore him.

     You might even, it would be more polite, just ask me not to interfere here. I am generally civilized and I would try to please you.

141. phi–

     The problem is that you would be well unable to prove it.

     Prove your claim that back radiation violates the 2nd law is wrong? I don’t need to prove any such thing.

     In the first place, if you are going to claim something violates the 2nd law, you need to prove your claim.
     Neither you, nor anyone who has ever made the claim back radiation violates the 2nd law has ever shown their proof. Until such time as you prove it, no one is required to prove it doesn’t violate the 2nd law! (And, btw, proof of what does violate the 2nd law are possible. For example students routinely do proofs to show that mach number can only drop across an normal shock wave and so on.)

     If you want to go around making this claim, you ought to either do the proof or go find one instead of just spouting it. (Tip: you will fail in proving the claim.)

     In the second place: back radiation — as the term is used when describing warming due to CO2 in the atmospehre– does not violate the 2nd law. If you think it does, you either (a) don’t understand what “back radiation” is or you don’t understand the 2nd law of thermo.

     In any case: back radiation just flat out doesn’t violate the 2nd law. It is routinely accepted as occurring in furnaces, and fireplaces. Some applications use various sorts of reflective inner surfaces on some special toaster or oven designs. They use foil to help keep food warm– and these work precisely because back radiation works. Here’s another nice word: Thermos. That’s a great “back radiation” application. Mylar emergency blankets are nice too. All work on the concept that “back radiation” exists.

142. Lucia,

     I missed it.

     This is incorrect. Back radiation does not violate the 2nd law of thermodynamics. Anyone who thinks so does not understand the 2nd law of thermodynamics.

     You obviously do not prove anything. Backradiations can not be considered independently of ascending radiation within a thermic problem because they bring low temperature flux to a warm body which can not do anything of that. This does not mean that a cold body does not slow down the cooling of a hot body. Simply, you can not distinguish emission and absorption as do models of the greenhouse effect.

     I will not stop on the following of your prose, it’s inelegant and regrettable.

143. Lucia,

     I do not comment on your last post. I hope you now understand the nature of the problem. It is obviously not to deny the effect of a cold body on a hot body but to note that radiation of the two bodies can not be treated independently.

144. phi

     Backradiations can not be considered independently of ascending radiation within a thermic problem because they bring low temperature flux to a warm body which can not do anything of that.

     The italics portion is a mis-application of 2nd law; misstating the 2nd law does not constitute any sort of proof. The 2nd law has to be applied to the net heat flux.

     As for “Backradiations can not be considered independently of ascending radiation”, this is simply nonsense. Analysts routinely break down the net flux into constituent parts for convenience. The analysts choice to break things down for convenience of analysis has no physical significance– it is an analytical step. Period. The 2nd law doesn’t magically come into blocking photons emanating from a CO2 molecule and preventing those with a component in the direction of the earth from heading toward that surface simply because the analyst decided to write down the expression for the rate of emanation from the CO2 on one line on a sheet of paper and wrote the expression for photons emanating from the surface of the earth on another line on a piece of paper.

     Simply, you can not distinguish emission and absorption as do models of the greenhouse effect.

     This is total drivel. There is no reason that an analyst cannot break a problem into constituent parts writing down the mathematical expression for “emission from a surface on one piece of paper” and the expression for “absorption” on another. This is booking and down for convenience. It is certainly permissible to distinguish the two things in this way if one wishes to organize an analysis this way.

     (What’s next? Is there going to be some rule saying that accountants can’t keep track of debits and credits in different columns if they want to? Or use red ink for one and back for the other? Or what?! Of course people can break up ‘debits’ and ‘credits’ — if they find it convenient– and those doing heat transfer can break up ‘absorption’ and ’emission’ if they so desire.)

     I will not stop on the following of your prose, it’s inelegant and regrettable.

     Possibly, the French version of this sentence is withering, elegant and not to be regretted. I speak French and can often back-translate to figure out what mangled translations were intended to communicate. But I admit I am at a loss. Maybe one of the native French speakers can help out (but I’d rather they didn’t as it will derail the thread. )

     I should warn you: Defending your view on the 2nd law of thermo by telling ‘she who has access to the ban button’ that her prose is ‘inelegant and regrettable’ is not a very convincing method of supporting your view of the 2nd law. Moreover, it is not the best way to remain unmoderated!

145. Lucia,

     The 2nd law has to be applied to the net heat flux.

     Exactly. This means in other words that the disconnection between OLR and DLR is inappropriate.

     This is total drivel. There is no reason that an analyst cannot break a problem into constituent parts writing down the mathematical expression for “emission from a surface on one piece of paper” and the expression for “absorption” on another.

     What an analyst can do and what is appropriate to implement in a model are two different things. No one has ever solved a problem of conduction by separating interractions in two opposite flows. There is no reason that this is done for IR radiation in the case of greenhouse effect.

     Is there going to be some rule saying that accountants can’t keep track of debits and credits in different columns if they want to?

     What they can not do is pass a dollar credit on a Yen debit at their convenience. This is what modelers does.

     I should warn you: Defending your view on the 2nd law of thermo by telling ‘she who has access to the ban button’ that her prose is ‘inelegant and regrettable’ is not a very convincing method of supporting your view of the 2nd law. Moreover, it is not the best way to remain unmoderated!

     Indeed, but I think you know what I meant.

146. Phi (Comment #109948)
     “radiation of the two bodies can not be treated independently”

     I normally stay out of these, but you do realize that what you are saying contradicts 100 years of radiative physics don’t you?

     The Stephan-Boltzmann equation specifically treats the radiation from a black body as independent of any other radiating bodies. Phi, if you could prove what you say, you would be overturning a huge body of physics and would likely be a candidate for the Nobel prize.

147. Phil Scadden,
     ” However, do you agree that Argo data should separate a forced response (increased aerosol) from change in AMO?”
     Assuming you are addressing me with that question (it is not clear, since you don’t reference my earlier comments): Sure, measured evolution of heat content over depth via ARGO could help clarify what is happening. However one needs to keep in mind that the Gulf Stream flows fast enough (>5 Km per hour near Florida!) to carry water from the tropics to 55-60 degrees north in only about 90 days, so the heat content of the North Atlantic is potentially subject to pretty rapid change if the flow of the Gulf stream changes significantly.
     .
     “Another point, if you think AGW is exaggerated over last 35 years, you would expect ENSO contribution to differ with time period.”
     I suspect the observed rate of warming was increased above the secular trend from about 1975 to about 2005, and especially from ~1980 to ~2000 (but not over 35 years). The current observed rate of warming, and the observed rate over the next couple of decades, is likely being reduced due to the downward part of the cycle, so in that sense we are currently being “misled” about the secular trend being very low at present, just as we were misled between ~1980 and ~2000 about the trend being very high. The true secular trend is unlikely to be either.
     .
     But I don’t see any rational for thinking the ENSO contribution to short term variation would change over that period. The ENSO influence is mainly short term (several months), and limited to between 30 N and 30 S, with ~3 months of lag, with only much smaller effects at higher latitudes at longer lags… more like 6 months. It seems to take time for tropical warming to make its way to high latitudes. Which makes perfect sense if high latitude warming in response to the El Nino phase of the cycle is in large part due to warmer tropical surface waters reaching high northern latitudes.

148. phi

     to note that radiation of the two bodies can not be treated independently.

     I have no idea which “two bodies” you refer to here or precisely what problem you are formulating and what you think you mean by “independently” here. The reason I don’t know is that you have not stated anything with sufficient precision to permit me to know.

     That said: the mathematical relation for emission from the sun can be expressed entirely independently from the presence of the earth. The mathematical relation for the emission from a point on the surface of earth (or any other planet) can be expressed based on the properties and temperature or that point on the surface of the earth. Under the continuum point of view, mathematical relation for emission from a “point” in the atmosphere containing CO2 molecules can be expressed based on the temperature at that point– and totally independently of the surface of the earth. And so on.

     If you want to prove back radiation somehow violates the 2nd law, you are going to have to go to web site, write up your proof– with figures– and post it. Feel free to put the link here and we can read it. Feel free to write it in French if that is easier for you.

     Making ambiguous claims and trying to support the claim with vague statements isn’t going to cut it.

149. Well, I think it is interesting. The “no-backradiation” gang thinks that there is some sort of teleconnection/interference where the net radiative flux is the only physical flux. I sort of gathered this but it’s being stated more clearly here.

     It’s such a shame that thermodynamics was (were?) invented before quantum mechanics.

150. Sorry Lucia, I can’t resist it. My guess is that it comes from an overliteral translation of something like “Je ne m’arrêterai pas sur la lisibilité (compréhension ou compréhensibilité?) de votre prose, c’est inélégant et regrettable.”
     Which is probably better translated as “I won’t dwell on the intelligibility of your writing style, which is inelegant (or unladylike/unrefined/clumsy) and regrettable.”

     It’s not a lot prettier in French.

151. Phi,
     “I raised a number of statements that are not trivial and should be easily refuted if wrong.”
     .
     Non-trivial? Worse than trivial, they are pure nonsense, and they reveal your lack of understanding of basic science. Your statements are more comical than anything else. Easily refuted? Yes, but you would never be able to understand that your nonsensical claims have been refuted, no matter how clearly explained… I note Lucia and others have (unwisely) tried, and it is obvious from your responses to their explanations that you have not even a clue what they are saying. You go on to insult those who do understand the science you do not whenever they try to help you.
     .
     Rather than waste more of your time here, as well as that of those who try (in vain) to help you understand basic concepts, I suggest you spend your free time at any of the Skydragon blogs. Your lack of understanding of basic science, combined with your complete unwillingness to learn, and general hostility toward scientific knowledge, will be completely normal and warmly embraced at those blogs.

152. BIllC:

     Well, I think it is interesting. The “no-backradiation” gang thinks that there is some sort of teleconnection/interference where the net radiative flux is the only physical flux. I sort of gathered this but it’s being stated more clearly here.

     I think you’re correct. For a while the response was to point out it would require sentient photons and violation of causality for what they claim to work. Then they started doing away with photons and things got really silly.

153. SteveF:

     I suggest you spend your free time at any of the Skydragon blogs

     Wondering if you know where they are?

     Maybe phi can point us to one of his sources for this so we can get edjumicated.

154. John Vetterling,
     “I normally stay out of these, but you do realize that what you are saying contradicts 100 years of radiative physics don’t you?”
     .
     Better to avoid communication with those who are nuts.

155. John Vetterling,

     …what you are saying contradicts 100 years of radiative physics…

     I do not think so. Radiative physics is not thermodynamics. And this is exactly the point.

     …if you could prove what you say…

     Easy. You can not make act DLR independently of OLR without violating the second law.

     Lucia,
     The problem with an unique body is obviously not concerned.

     Paul_K,

     I’m sorry to master so bad English language. No, it has nothing to do with style.

     To others,

     Start to justify the arbitrary setting of the initial temperature of the CO2 added in the models (start of the discussion). Then, if you are mature enough, we can move on to more fundamental problems.

     I am sincerely sorry for the tone, I am committed to more moderation if everyone plays the game.

156. Carrick,

     “For a while the response was to point out it would require sentient photons and violation of causality for what they claim to work.”
     .
     Violations of causality are only acceptable to the crazed, and pointing out violation of causality is a waste of time with the self-same crazed. No photons?!?!

157. Regarding the existence of the AMO, Delsole et al (2011): Delsole_A Significant Component of Unforced Multidecadal Variability in the Recent Acceleration of Global Warming, J. Climate, found an Internal Multidecdal Pattern that he identified as being closely related to the AMO (see his Fig.4). Admittedly Delsole only used 160 year long data, but his results are supportive of the Tung and Zhou paper.

     A version is available at ftp://cola.gmu.edu/pub/delsole/dir_ipcc/dts_jclim_2010.pdf

158. I think it was Claes Johnson who finally coughed up the statement that IR photons don’t exist, on the Air Vent?

159. SteveF – while I agree that ENSO contribution is expected to be fairly constant, if there is a hidden contributor not included in the model, (AMO) then I would expect the regression to give significantly different coefficient values depending on which data segments you use. Again Kevin C splits the 35 year in two segments and gets similar values for both segments and the whole. As far as I can see, while there may be another influences contributing to flat temperatures at moment (eg aerosols, AMO), F&R and similar approaches dont give you much statistical support for that.

160. phi (Comment #109960) says : “Easy. You can not make act DLR independently of OLR without violating the second law.”

     Im not entirely sure what you are saying here… although i can guess, from seeing similar statements… But how exactly do you understand the second law?

     Entropy increases or stays the same in an enclosed system.(from top o my head, without looking it up) What this is saying, is chaos increases(on average) when looking at the full picture(as opposed to looking at just a part of a process) Back radiation isnt in violation of this, it is a result of this, of the tropospheres chaos increasing(energy spreading), because its cooler than the surface of course the net flow is from the surface up, due to T differences… at no stage, is there decreasing entropy… which would mean energy was clumping together(as happens say when you compress air, of course the energy required for the compression balances out the decreasing entropy, so its only decreasing entropy if you look at part of the process), for emissivity to be equal to absorption, its impossible for decreasing entropy through radiative mechanisms… because it radiates proportionately to its T. This is simple stuff.

     This is long way off topic mind. I generally just read this and a few other blogs, because there can be quite interesting discussion on some o the more complex stuff… and dont comment, because i dont really have anything to contribute. Perhaps some in depth research on the established science is in order, before you dismiss it out of hand.

161. Brandon, you make a lot of differences in coefficients/lags between datasets but given the difference in coverage, I would expect some difference. The big difference is between surface temperature datasets and the UAH/RSS. These are measuring different things and the lower troposphere does have a stronger response to ENSO signal than the surface temperature record.

162. OHC is also the reason why I would suspect aerosols more of a factor than natural variability in the 130 year record. While OHC data has large error bars, it certainly seems to have dropped post 40s in line with the surface temperature record. This easier to account for by aerosols than some ocean process.

163. Phil Scadden, since I only looked at NCDC/GISS/CRU, I would hope there isn’t that much difference. Assuming the calculated parameters are accurate, that would mean there are significant differences in the variability of the three series. That is opposite what we should hope to expect.

     And it wouldn’t have anything to do with the differences between parameters for the different editions of their code.

164. I seriously doubt that OHC can distinguish between changes in the overturning circulation and aerosols. OHC is not the heat content of the entire ocean, it’s the heat content of the upper 1-2 km. The thermohaline circulation goes much deeper than that. If you don’t know the heat balance below 2 km, you don’t know whether it’s aerosols cooling the surface or if more heat is being pumped into the abyss.

165. No one has ever solved a problem of conduction by separating interractions in two opposite flows.

     Wrong. It’s an exercise often done in statistical thermodynamics to derive estimates for thermal conductivity for perfect gases.

     The more correct statement is, “People who haven’t taken very many thermo or heat transfer courses are unaware that this technique is used is classroom exercises showing ab initio calculations of thermal conductivity.”

     There is no reason that this is done for IR radiation in the case of greenhouse effect.

     Wrong. There is a perfectly good reason: Separating the problem into two opposite flows makes it more convenient to work out the geometry involved in the problem. That’s the same reason it’s done in statistical thermodynamics.

     Indeed, but I think you know what I meant.

     Evidently, not. Paul_K who speaks French fluently had a crack at back translating and he’s not sure either.

     John Vetterling,

     …what you are saying contradicts 100 years of radiative physics…

     I do not think so. Radiative physics is not thermodynamics. And this is exactly the point.

     Radiative physics is not thermodynamics. And this is exactly the point.

     Of course radiative physics is not thermodynamics. But John Vertterling is correct when he observes that you are saying contradicts what we have long known about radiative physics. Observing that radiative physics is not thermodynamics is a pretty lame attempt to avoid his point. (For what it’s worth: radiative physics — which is not itself thermodynamcis– does not contradict thermodynamics. It also doesn’t contradict mechanics, or conservation of mass while also not being radiative physics.)

     Start to justify the arbitrary setting of the initial temperature of the CO2 added in the models (start of the discussion). Then, if you are mature enough, we can move on to more fundamental problems.

     This doesn’t even make sense. I doubt if anyone has the slightest clue what you are trying to communicate. I suggest you do the work to support your claims instead of trying to assign “homework” to others. (The latter — like trying to argue by rhetorical questions– is a violation of the blog rules. Oddly, though few have heard the rule expressed this way, one of the reasons arguing by rhetorical questions violates ‘the blog rules’ is that it is an attempt by the person trying to use that technique to avoid providing his own arguments for his own claims and instead try to suggest that somehow it is other people’s job to provide the arguments.)

     I am committed to more moderation if everyone plays the game.

     I have no idea what you are trying to say. I also suspect you don’t understand what I mean by moderating you. I will now moderate you and you will come to understand.

166. KingOchaos,

     Back radiation isnt in violation of this…

     No, of course. What violates the second law is the representation which would that DLR can act on matter independently of OLR.

     Perhaps some in depth research on the established science is in order, before you dismiss it out of hand.

     Unfortunately, the physics of the greenhouse effect is not a science that can be described as established. This is a bricolage which take concepts from right and left without asking too many questions about the legitimacy of the assembly.

     This comment (# 109929) specifies a bit what I think.

167. Dang. One slipped by before I got the “moderate” button set! Phi is now moderated.

168. Brandon, I am missing something here. Looking at F&R, there is no significant difference between the coefficients for the 3 data sets. (fig 4).

169. As far as I can see, while there may be another influences contributing to flat temperatures at moment (eg aerosols, AMO), F&R and similar approaches dont give you much statistical support for that.

     This is actually the question I was trying to answer in a recent post:

     http://troyca.wordpress.com/2013/01/25/could-the-multiple-regression-approach-detect-a-recent-pause-in-global-warming/

     Essentially, I used a simple model with a simulated low frequency oscillation (call it the AMO for argument’s sake) creating a recent pause and tried to see if the F&R method would detect the pause or simply “reconstruct” it out of existence. My tentative conclusion so far is that the F&R result does not adequately detect such a pause, although the issue might not be as big if that simple model is not accurately simulating the residual effects of volcanic eruptions, which has sent me down a bit of a tangent. Regardless, I would not put too much stock in the F&R results disproving some other significant contributions to the recent “pause”…at the very least, the F&R results almost certainly put too much weight on the decrease in TSI over the last decade.

170. Lucia,

     Wrong. It’s an exercise often done in statistical thermodynamics to derive estimates for thermal conductivity for perfect gases.

     So I was wrong. These two flows are independent?

     Wrong. There is a perfectly good reason: Separating the problem into two opposite flows makes it more convenient to work out the geometry involved in the problem.

     More convenient does not mean legitimate.

     Evidently, not.

     So forget it.

     But John Vertterling is correct when he observes that you are saying contradicts what we have long known about radiative physics.

     Really ? I said : You can not make act DLR independently of OLR without violating the second law.
     Is that wrong ?

     This doesn’t even make sense. I doubt if anyone has the slightest clue what you are trying to communicate.

     It is however clear. Models thermalize arbitrarily the added CO2. Until proven otherwise, there is no justification for this operation which however sets the value of the radiative effect.

171. Paul_K,

     Just wanted to say this is a very interesting post. It seems to have gotten thread-jacked a bit, but I hope to have more questions/comments after I’ve had a chance to digest this and Armour.

172. BillC:

     I think it was Claes Johnson who finally coughed up the statement that IR photons don’t exist, on the Air Vent?

     Something along those lines. This all reminds me of people who have a religious doctrine that when you point out the problems with their doctrine, redefine reality to conform with their doctrine (e.g., geocentric universe).

     By the way there are those who still think, that because of their biblical interpretations, that the Earth is indeed the center of the Universe.

     Lucia might be interested in this, because there is a church in the Chicago area where you can still hear this doctrine. Among other things they decry Galileo.

     “Galileo was wrong.”

     The Old Testament also mentions the “pillars of the Earth”. I am only glad that nobody thinks the Earth is on pillars, which itself resides on a giant tortoise, which, when it moves, causes the Earth to shake. Or at least I haven’t met anybody who believes that and admits to it.

     /backtoourregularlyscheduledprogram

173. Surely the implications of the Armour et al paper are only important and wide-ranging if its conclusions are correct. I am somewhat doubtful whether they are.

     The authors state that “the time-dependence of lambda_eff has been widely demonstrated”, which I do not believe is true. What they seem to mean is that the time-dependence of lambda_eff has been shown in a wide range of global climate models, which is an altogether different matter. This confusion of model-simulations with reality does not engender confidence. (Using “GCMs” to mean global climate models rather than, as standard, general circulation models is also a bit odd.)

     Armour et al cite Bates (2010, Climate Dynamics, Climate stability and sensitivity in some simple conceptual models), but fail to point out that its zonal models (zone 1 tropics 30S-30N, and zone 2 both extratropics combined) are based on the opposite pattern of zonal radiative response coefficients to that implied by Armour et al, and reflected in Paul_K’s final model. Bates’s model explicitly involves meridional heat transport, which is a major factor in climate stability. He explores extratropics lamda_eff high (he says most authors agree that the extratropical coefficient is close to the blackbody value) and tropical lambda_eff low, even negative. He cites GCM simulation based evidence for this latitudinal pattern.

     I can’t find a free version of Bates paper, but presentation slides are available at http://www.ima.org.uk/_db/_documents/Bates.pdf

174. Just realized my comment #109974 did not include that the quote I was replying was Phil Scadden in #109964. Sorry about that!

175. Lucia,

     Wrong. It’s an exercise often done in statistical thermodynamics to derive estimates for thermal conductivity for perfect gases.

     So I was wrong. These two flows are independent?

     What do you mean by independent? They are not equal to each other.

     Wrong. There is a perfectly good reason: Separating the problem into two opposite flows makes it more convenient to work out the geometry involved in the problem.

     More convenient does not mean legitimate.

     There is nothing illegitimate about breaking an analysis into smaller pieces for the sake of convenience. Nothing. Absolutely nothing.
     If you want to find something illegitimate in this, you have to identify the specific step that is illegitimate. So far, you have only said breaking it into pieces is somehow wrong. You are simply incorrect about that.

     But John Vertterling is correct when he observes that you are saying contradicts what we have long known about radiative physics.

     Really ? I said : You can not make act DLR independently of OLR without violating the second law.
     Is that wrong ?

     Yes. What you said is wrong. It does not violate the 2nd law to write formulations of DLR and OLR that are functionally independent from one another. That’s what they do when estimating the greenhouse effect. That’s what has been done in radiative physics for a long time. And it’s perfectly ok — because among other things– it does not violate the 2nd law of thermo.

     This doesn’t even make sense. I doubt if anyone has the slightest clue what you are trying to communicate.

     It is however clear. Models thermalize arbitrarily the added CO2. Until proven otherwise, there is no justification for this operation which however sets the value of the radiative effect.

     This continues to make no sense and I doubt anyone has the slightest clue what you are trying to express. The fact that you think you know what you are saying doesn’t mean you have managed to say it clearly.

     As for CO2: That CO2 is an emitter is an observed physical fact. If you don’t believe this fact, doesn’t mean it isn’t so.

176. I have never had any problem with back radiation myself.
     However, as a picture says a thousand words does anyone have a picture of a red-hot electrical furnace showing that the air, heated by the walls, emits light as the same frequency as the metal?
     Any one have a picture of red-hot air?

177. Nic Lewis, why shouldn’t the default assumption be that lambda_eff varies? Seems to me the onus should be on people to show it doesn’t, not that it does.

     A couple of other papers probably worth paying attention to are Feldl and Roe “The nonlinear and nonlocal nature of climate feedbacks” and Karen McKinnon’s AGU presentation “The Spatial Pattern Of Temperature Change.”

178. Troy, thanks for the link and nice work as always.

     I had suspected this is what you would find .

     If you are interested in playing more games with end-point bias, try having your “true signal” curve upwards.

     Then try it again with a zero total trend (except at the end).

     Finally put an up-and-down wiggle around the mid-point of the series (the version with a net trend).

     (Changing the frequency of the wiggle so it is closer to the ENSO should have predictable results too.)

179. Very interesting Troy. That is indeed a good question to ask and needs more work.

180. DocMartyn (Comment #109981)
     February 12th, 2013 at 4:52 pm

     I have never had any problem with back radiation myself.
     However, as a picture says a thousand words does anyone have a picture of a red-hot electrical furnace showing that the air, heated by the walls, emits light as the same frequency as the metal?
     Any one have a picture of red-hot air?

     Kirchhoff’s… law Emissivity is equal to absorptivity… red light, by definition, is not longer than red (infra red) So, our atmosphere (N2 O2 co2 ch4 etc is largely transparent to the radiation, emitted by the furnace walls, that’s not to say it dosnt get hot through conduction, and absorption of the wave lengths emitted in the lines its opaque to… it just means its emitting in lines (not as a black body) according to its absorptivity, transparency. It still conducts o course.

     Sorry for strayin so far off course. Ill stop posting now

181. Lucia,

     You ask me questions, I guess you leave me to respond.

     Let me begin by apologizing to Paul_K, I guess that I annoy him at the highest point.

     I will not take all the points.

     There is nothing illegitimate about breaking an analysis into smaller pieces for the sake of convenience.

     Indeed. But, between parts are links, you can not break those links.

     If you want to find something illegitimate in this, you have to identify the specific step that is illegitimate.

     Radiative forcing is supposed to represent a DLR with its own existence combinable with other energy flows. This is completely unacceptable.

     As for CO2: That CO2 is an emitter is an observed physical fact. If you don’t believe this fact, doesn’t mean it isn’t so.

     Do I deny it???

     I could return the compliment: Models thermalize arbitrarily CO2, it is a fact. If you do not believe this fact, does not mean it is not so.

182. Lucia,

     phi said: [“Models thermalize arbitrarily the added CO2. Until proven otherwise, there is no justification for this operation which however sets the value of the radiative effect.”]

     To which, you responded: [“This continues to make no sense and I doubt anyone has the slightest clue what you are trying to express.”]

     With due respect, I think you are being somewhat obtuse with phi. What he is saying (IMO) is this:

     The people who build the models have arbitrarily (i.e. on a whim or at personal choice) allocated a thermal quantity (hence ‘thermalize’) to the radiative effect of CO2 in the atmosphere. There is no proof (‘justification’) for this allocation (‘operation’) of a thermal quantity in the atmosphere but, in spite of there being no proof, or evidence, the people who build the models have used the models to ‘set the radiative effect’, or allocate a thermal quantity – such as climate sensitivity – as a consequence of CO2 increase. The thrust of phi’s point (made in a non-native language) makes sense to me even if I were to disagree with it.

     You then say:[ “That CO2 is an emitter is an observed physical fact. If you don’t believe this fact, doesn’t mean it isn’t so.”]

     The first sentence is correct. The second is unneccesary. The point phi is making is that CO2 being an emitter does not necessarily mean that the radiation emitted by CO2 has any evidential thermal consequence in the real atmosphere (as opposed to modelled atmosphere).

     To paraphrase you, if you believe that CO2 does have a thermal consequence, it doesn’t mean it is so. Without direct and real evidence, the idea of climate sensitivity remains a theoretical construct.

     Finally, moderating someone who is expressing an opinion without being rude, personal or pejorative, is, in my opinion, cyber-bullying.

     With due – if slightly diminished – respect…

183. Phil Scadden,
     “F&R and similar approaches dont give you much statistical support for that”
     How could an analysis based on ~3 decades of data possibly find “statistical support for” a ~60 year period cyclical variation? F&R look for a linear trend over a short period and (surprise!) they find one. Had they the good sense to extend their analysis back through 1900 or so, they would find that their results, except for the ENSO influence, turn out to be largely spurious, especially the large, zero time lag solar influence. IMO, F&R is essentially a curve fit exercise, and completely unpersuasive.
     .
     You want to bet how much warming there will be over the next decade? I say it will probably be under 0.1 degree, and there is a reasonable chance under 0.05 degree. If that turns out to be true, will you then agree there is a substantial cyclical component?

184. Carrick,
     The Earth rides on a turtle, and turtle that rides on another. Yes, it’s turtles all the way down… 😉

185. King Ochoa

     air, heated by the walls, emits light as the same frequency as the metal?

     Such an image shouldn’t exist because the air doesn’t emit at the “same frequency” as the metal. Emission by the air depends on the temperature of the air and the properties of the air. The air temperature is different from the metal and so does not emit at “the same frequency” as the metal.

     A similar observation can be made for the atmosphere. The CO2 in the atmosphere emits at the temperature of the air which is lower than both the temperature of the sun and of the earth’s surface.

     our atmosphere (N2 O2 co2 ch4 etc is largely transparent

     Of course it’s largely transparent. That’s why we can see stars and the moon at night and the sun looks like a ball by day. But it’s not entirely transparent. Among other things, clouds can block the view of the stars, moon and sun. Haze… less so.

186. Arfur

     With due respect, I think you are being somewhat obtuse with phi. What he is saying (IMO) is this:

     The people who build the models have arbitrarily (i.e. on a whim or at personal choice) allocated a thermal quantity (hence ‘thermalize’) to the radiative effect of CO2 in the atmosphere. There is no proof (‘justification’) for this allocation (‘operation’) of a thermal quantity in the atmosphere but, in spite of there being no proof, or evidence, the people who build the models have used the models to ‘set the radiative effect’, or allocate a thermal quantity – such as climate sensitivity – as a consequence of CO2 increase. The thrust of phi’s point (made in a non-native language) makes sense to me even if I were to disagree with it.

     Unfortunately for your theory, I cannot square it with phi’s reply (currently in moderation) which is:

     As for CO2: That CO2 is an emitter is an observed physical fact. If you don’t believe this fact, doesn’t mean it isn’t so.

     Do I deny it???

     So, while it seems to me that your interpretation assumes he denies that CO2 is an emitter, he does not deny it. Whatever he means by setting it “arbitrarily”, it seems he considers it “arbitrary” for a modeler specify that CO2 is an emitter when it is indeed an emitter.

     The rest of his comment is utter nonesense. I’ll release it from moderation after I hit submit.

187. lucia (Comment #109990)

     Yes… 😉 that was exactly the point i was making to DrMartyn. My formatting may have caused some confusion… But you have no arguments from me.

188. Arfur-

     Finally, moderating someone who is expressing an opinion without being rude, personal or pejorative, is, in my opinion, cyber-bullying.

     With due – if slightly diminished – respect…

     First: It is not cyber-bullying to moderate someone who comes to my blog. Bullying would require me to go after him.

     Second: Phi is being rude and perjorative. His english isn’t very good, but see for example:
     http://rankexploits.com/musings/2013/observation-vs-model-bringing-heavy-armour-into-the-war/#comment-109940

     phi (Comment #109947) where he is directly insulting to me. The only reason it’s obscure is that it is a mangled translation from the French.

     Worse, he is attempting to be condescending, and trying to argue by assigning people “homework” instead of providing his own arguments to support his own claims.

189. Phil Scadden, I assume you mean Figure 3. I’m not sure what’s going on with that one, but it doesn’t seem to match F&R’s published parameter values. Of course, even if it did, it only shows three parameters. F&R’s linear regression uses nine parameters, and that’s after lags were already selected. Heck, the difference in lags alone is enough to make my point. Why should NCDC have an ENSO lag half that of GISS?

     Anyway, if you download the code F&R published, you’ll find a file that shows CRU’s solar parameter is ~135% that of GISS’s (~120% that of NCDC’s). That doesn’t match Figure 3, but it is taken directly from F&R’s results file. The same pattern can also be found in the results file of the updated code dump. It’s undeniable.

     Should we take that to mean solar changes affect CRU more strongly than NCDC or GISS? If so, why would they? If not, why would F&R’s method say they do?

190. Lucia, why wouldn’t the air be at thermal equilibrium with the electrical furnace? We are to believe that the atmosphere is in thermal equilibrium with incoming solar radiation and atmospheric IR.

191. Carrick – I think it was Karen McKinnon’s presentation I was referring to when I said “Dig, Here be Witches”.

192. Troy_CA,
     Very nice work over at your blog analyzing the weaknesses in the F&R approach.

193. BillC, Carrick,

     I can;t find any information on Karen McKinnon’s presentation. Can you give me a link?

194. Lucia, why wouldn’t the air be at thermal equilibrium with the electrical furnace? We are to believe that the atmosphere is in thermal equilibrium with incoming solar radiation and atmospheric IR.

     Air near a furnace isn’t in thermal equilibrium because of a lack of sufficient time to equilibrize (by conduction, radiation, and/or convection).

     Likewise, the atmosphere is most certainly not in thermal equilibrium – note the substantial temperature differential between the poles and the equator, for instance. Any body with an atmosphere that is near a source of heat (like the Sun) won’t be anywhere near thermal equilibrium.

195. Doc

     Lucia, why wouldn’t the air be at thermal equilibrium with the electrical furnace? We are to believe that the atmosphere is in thermal equilibrium with incoming solar radiation and atmospheric IR.

     Read the description

     However, as a picture says a thousand words does anyone have a picture of a red-hot electrical furnace showing that the air, heated by the walls, emits light as the same frequency as the metal?

     The way I read this description: The “red-hot electrical furnace” is a different thing from the “walls”. The electrical furnace is at one temperature (red hot, T_furnace). The walls are heated by radiation from the furnace– and are some temperature T_wall which is different from the furnace. The air between the furnace and the walls is at another temperature T_air.

     The rest of the system was not specified. The air might be sucked into to the furnace from an inlet and blown out (as in a home heating furnace.) In which case, the air is not in “equilibrium”– but it is still heated.

     There are other possible geometries in the problem. I don’t know which KingOChaos meant. But off hand I can’t think of any where air would be at the same temperature as the furnace element if this is a furnace and there is a heat flux from the furnace heating elements to the walls.

196. ” If that turns out to be true, will you then agree there is a substantial cyclical component?”

     I would agree that there is another substantial factor at play. The cyclical bit has a long way to go. I’d assume you would ditch the cyclical theory should temperatures rise substantially by 0.17 or more?

     Brandon, issues with the solar have been raised (Rypdal 2012) and I accept that problem. Prefer Kevin C’s approach. Fixing it makes little difference the conclusions.

197. Phil Scadden, “fixing” anything with F&R’s approach won’t change results much. All F&R did is fit a linear regression with extra parameters to remove noise. The results are guaranteed to be little different than just doing a linear regression.

     The fact F&R’s inexplicable parameter results (like “seasonal cycles” that are nothing like seasonal cycles) don’t change much isn’t a defense of the methodology. If anything, it shows the weakness of the methodology.

198. Phil Scadden,
     “I’d assume you would ditch the cyclical theory should temperatures rise substantially by 0.17 or more?”
     I sure would be inclined to.
     .
     “I would agree that there is another substantial factor at play. The cyclical bit has a long way to go.”
     I am surprised that you continue to suggest ~100% forced response, in spite of lots of evidence to the contrary. Really, in light of the appearance of pseudo cyclical variation over a wide range of time scales in historical data, from multiple sources, I don’t understand your reasoning at all. The most reasonable explanation for the historical temperature trend is a significant contribution from internal variability.

199. Brandon Shollenberger

     fit a linear regression with extra parameters to remove noise

     And showed the AR4 projections using a baseline not selected by those who made the AR4 projections. This forced the AR4 projections into better agreement with data through the method of “subtracting the difference from the projection and the observation from the projection”.

     FWIW: I don’t have any particular theory about cyclical. I merely note that the model mean projection for the early 21st century was above 0.20C/dec, so 0.17C for the decade would still be on the lower 1/2 of projections. (You can verify the model mean in a variety of ways.)

200. Lucia.. I was quoting Doc in regards to the furnace … They are not my words. I apologise for the poor formatting.

     Doc, the air in the furnace, even at thermal equallibrium, will neither absorb or emmit in visible light. It will be emmitting what it absorbs, in the wave lengths its opaque to. So for air it will be in infra red… Longer than red, so not visible.

201. arfur

     ” Without direct and real evidence, the idea of climate sensitivity remains a theoretical construct.”

     Several points. First I don’t think you know what a theoretical construct is. The physical world is a good example. But that’s a philosophical point. But let’s assume for a minute that there are things that are not constructs or posits. And lets assume there is something called “direct evidence”

     Climate sensitivity is a metric. Like miles per gallon is a metric. Like speed is a metric. for miles per gallon you measure Distance and you measure a volume and you come up with this “thing” called
     MPG.. hmm, does miles per gallon exist? do we have direct measurement of that thing? Wait, what about the speed of sound.
     we measure a distance and a time.. but never speed “directly” what ever that is. is speed a theoretical construct? Anyway look at sensitivity:

     Its: the change in C per change in Watts.

     Now measuring that metric and characterizing that metric is hard. But, you have delta C, you have delta watts so its as theoretical as those two entities. Its as theoretical as speed. What is known less well is how stable that metric is ( ratios are nasty business ) and how well one can measure it.

     What things can we actually measure directly? and is our knowledge limited to things we can measure directly. hmm.. nope.

202. It should also be noted that F&R 2011 used the old faulty PMOD composite data for solar forcing TSI.

     One of the instruments it relies on is suffering from degradation and it has an inaccurate decline of -0.5 W/m2 starting in about 2008.

     Obviously, they knew that but used it anyway because the accurate data would start to show a decline in the adjusted ENSO-Solar-Volcano trendline.

     I mean they even detrended the PMOD composite data so that it would not give them the wrong regression sign. Anyway, we have seen this sort of thing from Tamino before. He gets away with it in some venues but I would call it bush league (ball player from way back and that is what we would call it).

203. KingOchaos–
     Ahh! Well, then that’s 3 of us responding to Doc and all 3 of us seem to be in broad agreement. 🙂

204. SteveF,

     Karen’s poster should be in your email.

205. lucia:

     Worse, he is attempting to be condescending, and trying to argue by assigning people “homework” instead of providing his own arguments to support his own claims.

     He was doing that on Jeff’s blog. He actually expected me to dig through Hansen’s 2001 paper to try and decipher where he was getting what he was claiming from. Which wouldn’t have been a real problem (I’ve read it several times), but I’m pretty sure Hansen wouldn’t agree with him on his interpretation (at least the English version of it), making such a “homework assignment” pretty much impossible.

206. Carrick–
     Around here, if ‘X’ (i.e. phi) claims Hansens 2001 says “y”, and someone asks him to supply the quote, figure or whatever that shows Hansen got that, then it’s ‘X’s (i.e. phi’s ) responsibility to get the quotes, figures or whatever that shows Hansen said “y”. It’s not appropriate for ‘”X” (i.e. phi ) to tell the person disputing X (i.e. phi’s) claim to go hunt down the proof of X’s claim. This rule is required because in addition to enforcing the general issue that people should required to support their own claims rather than act as if they have a right to assign the “homework” of finding the support to others (who in fact dispute the claim), it also avoids permitting “X” to try to force people to read his mind.

     After all: In the case you describe, where phi was supposedly demanding you hunt down the evidence for phi’s claim in Hansen 2001, for you to attempt it, you not only have to do phi’s work, but you are working at the disadvantage of having to to guess how phi interpreted what Hansen said and then — based on that guess– guess which bit phi thinks supports his claim Hansen 2001 says whatever it is phi claims.

     Given the (in this case strong) possibility that phi badly misinterpreted Hansen, that means that phi has not only dumped his job on you, but in doing so, he has created a job that he could presumably do easily (i.e. find the quotes and figures he thinks support his case) and turned it into a job that would be impossible for you to do (i.e. try to guess precisely how phi misinterpreted Hansen and while doing so figure out which bits of misinterpretation map into phi’s claim of what Hansen did.)

     And then should you guess wrong, “you” are suddenly the one who was unable to “find” that which phi insists is there (but refuses to reveal on the grounds that he has assigned dredging that up to you.)

     The behavior is very disrespectful. Even though bystanders might not notice the level of disrespect being handed out, this “assigning of homewors” is just as disrespectful as calling people names.

     This is when I strongly discourage those who “assign homework” of this sort. Luckily, few do it. Consequently, the rule is not widely known.

207. Lucia, well put.

     We had this little exchange:

     [phi] This is explained in the reference I gave

     [me] Word of advice: you should never give a reference and leave the idea that you are trying to express left inside of that reference. I suggest using references to bolster your arguments, not make them for you.

     This was followed by the ever polite response from phi:

     Obviously, you argue without even having read Hansen et al. 2001. You are not responsible for homogenisation and therefore excusable. But at least have the decency not to talk about what you do not know.

     But I had read Hansen, several times even, and I wasn’t even arguing…I was just asking him to clarify since I just had no friggin’ idea where he was coming from. It was almost like we weren’t reading the same James E. Hansen paper or something…

     (That said, I’m not sure why my not being “responsible for homogenisation” makes me excusable, but hey, I’ll take what chits I can get in this world.)

208. Lucia,

     You will certainly be relieved to learn that I do not see any interest in continuing the discussion.

     I am accused of condescension, I have been in reaction to statements more hostile against me. For the rest, I do not see the point to come back on one-sided extracts from past dialogues.

     Initials questions have not found answers, thermalization of CO2 by models and the fact that backradiations and radiative forcing are concepts that simply can not be captured by thermodynamics. There will be no answers here.

209. “Any one have a picture of red-hot air?”

     Would a neon tube do?

210. I mistakenly referred to Karen McKinnon when I was thinking of someone else. Here’s the relevant except from John N-G’s “dispatch” here.

     Eight posters away, I stopped at a poster by Elizabeth Moyer of the University of Chicago. The focus of her paper was on climate simulations responding to a sudden increase of CO2. For the first hundred years or so, the temperature seems to be gradually equilibrating to a new value, but then it keeps getting warmer and warmer, finally stabilizing after 3000 years or so.

     Previous researchers have framed this as though the ocean was getting steadily more and more efficient at taking up excess heat, but the heat uptake is never really proportional to the radiative imbalance. Instead, Moyer showed that the warming rate made perfect sense if you looked at warming in individual latitude bands. The nonlinear behavior is just a consequence of different latitudes warming up at different rates.

     The slowest to warm up was the Southern Ocean, because (as I saw a minute ago), it has a lot of vertical mixing and takes a long time to warm up. Furthermore, that time delay was associated with a temporary change in radiative forcing: the relatively cool ocean encouraged more low-level clouds to form, thereby reflecting extra solar radiation back to space.

211. “The warming and cooling of the IMP matches that of the Atlantic Multidecadal Oscillation and is of sucient amplitude to explain the acceleration in warming during 1977-2008 as compared to 1946-1977, in spite of the forced component increasing at the same rate during these two periods.”

     They’re getting there. The so-called IMP should be called GMP or GMO, it’s unknown what’s forced and what’s unforced (or internal). The cyclical (or oscillatory) pattern with the ~60-year period is global and correlates with solar cycle frequency. The Null Hypothesis (no (significant) AGW) is getting stronger and stronger.

212. “While the IMP can contribute significantly to trends for periods of 30 years or less, it cannot account for the 0.8C warming trend that has been observed in the twentieth century spatially averaged SST.”

     I disagree – it can account easily for the 0.8 °C. Furthermore, the AGW started in the ~1960s.

213. Bill Illis (Comment #110007)

     February 12th, 2013 at 9:01 pm
     It should also be noted that F&R 2011 used the old faulty PMOD composite data for solar forcing TSI.
     One of the instruments it relies on is suffering from degradation and it has an inaccurate decline of -0.5 W/m2 starting in about 2008.
     Obviously, they knew that but used it anyway because the accurate data would start to show a decline in the adjusted ENSO-Solar-Volcano trendline.

     Are you a mind reader?

214. BTW Lucia, was your change of King OChaos to King Ochoa intentional? It’s funny either way.

215. bugs (Comment #110018)
     February 13th, 2013 at 4:31 am
     Bill Illis (Comment #110007)
     Are you a mind reader?
     ————————

     Tamino and Rahmstorf (and Skeptical Science and Hansen) are open books. Not difficult to assign thought processes once you’ve gone through their methodology a few times. [wrong sign on a regression? – detrend TSI? – from people who know what they are doing – easy call on that one].

216. steveta_uk (Comment #110014)

     “Any one have a picture of red-hot air?”

     Would a neon tube do?

     Just for the record…no, a neon tube wouldn’t do, not if by ‘red-hot’ one is referring to the temperature of a blackbody. The red of a neon sign is arises from electronic transitions (excited state emissions) in neon atoms. Also recall that blackbody radiation doesn’t depend on the material composing the blackbody, so what sense does a ‘neon’ sign make in that context.

     As far as the original comment (question) goes…why bother.

217. BIllC:

     I mistakenly referred to Karen McKinnon when I was thinking of someone else. Here’s the relevant except from John N-G’s “dispatch” here.

     OK, now I understand the comment 😉

218. phi:

     I am accused of condescension, I have been in reaction to statements more hostile against me. For the rest, I do not see the point to come back on one-sided extracts from past dialogues.

     To keep the record straight, you have reacted hostilely and sometimes in a rather ugly fashion, to mundane requests for clarification. The thread I was referring to started here. I don’t think I particularly distorted the record on that one, it’s not necessary.

     I don’t think you can accuse me of intentionally provoking the response I posted. And yes, I can come up with other examples of similar “testiness” on your part to what seem at the time sensible questions to and issues raised with your claims, so for you this is by now a well-established pattern.

     Initials questions have not found answers, thermalization of CO2 by models and the fact that backradiations and radiative forcing are concepts that simply can not be captured by thermodynamics. There will be no answers here.

     You’ve been answered already, the problem is what you say just doesn’t make any sense. I don’t know why that’s so hard for you to grasp. I suggest taking a course in radiative physics, and with an open mind. My prediction is you probably wouldn’t recognize your own “theories” afterwards.

219. Brandon:

     “fixing” anything with F&R’s approach won’t change results much. All F&R did is fit a linear regression with extra parameters to remove noise. The results are guaranteed to be little different than just doing a linear regression.

     .
     You’ve been making this claim several times. IIUC you’re saying that F&R’s main result (linear, non-plateauing temperatures once you remove fitted ENSO, solar and MEI) is just an artifact of the method, as opposed to an indication of a real relationship between the series.
     .
     If that is true, you should get a similar result using just any input series that have similar spectral properties.
     .
     Maybe you could provide some evidence for that claim? For example, try to run the same analysis with phase-scambled series? Heck, for a quick and dirty check, you could start by simply using the original series, but time-reversed (checking for non-time-symmetry, of course).
     .
     If you get similar flatness in the final series after subtracting the fit for these fake series, then I’ll admit that F&R’s approach doesn’t tell us much about the flatness of the real “underlying signal”. If you don’t, well, what would you conclude?

220. toto (Comment #110024)

     “Maybe you could provide some evidence for that claim”

     Way up thread TroyCA provided a link where he has done just that. But beyond that, it is definitional that, if you perform a linear regression you will get a linear result. If you perform a polynomial regression you get a polynomial result. The form of the regression can not tell you the form of the underlying parameter. The F&R analysis does not inform us as to whether there is an underlying linear trend. All the method can do is tell us what the best fit is IF there is and underlying linear trend. And Foster should know this, as does anyone who understands statistics.

221. toto (Comment #110024),

     You should look at Troy_CA’s blog for a detailed analysis which shows the F&R method can’t detect a known trend in synthetic data.
     .
     A secondary problem is that the F&R methodology pretty much falls apart, especially WRT solar response, if extended to earlier periods.

222. John Vetterling,
     Great minds… reply at the same time.. 🙂

223. For me the biggest problem here is using a filter, obtaining a smoothed curve and then assuming that smooth curve has anything to do with the underlying system, without any of the sorts of testing that is needed to know you’ve done it right.

     I wouldn’t be shocked if for a system that has a net trend, you got the right trend in the center of the time interval. I would be somewhat surprised if you got the right trend near the end points, as it’s been my experience is the filter tends to extend the trend from the center towards the endpoints.

     (In econometrics this issue is known as the “endpoint problem”.)

224. @toto: You say that F&R’s main result is “linear, non-plateauing temperatures once you remove fitted ENSO, solar and MEI”, which assumes that their data for ENSO, solar, and MEI are, among other things, accurate and meaningful. We know that a single ENSO number is ridiculously simplistic, the “solar” number doesn’t reflect any kind of realistic sun mechanism, and their MEI data has errors in it. Working with such data doesn’t bode well for whatever technique is used or whatever findings are reported.

     You seem to imply that temperatures apparently were plateauing, but by subtracting out the three factors, it was more linear. From what I understand, the data they used and the way they processed it meant that there was no plateauing from the start: all they did was to find a way to factor out some noise (the three factors) which had no significant effect but did drag down their R^2. That’s how they were able to say in their final sentence that the linear increase would continue for decades to come — because their three factors were too trivial to bend the trend.

     So they had inaccurate and vaguely-meaningful data which they found to have weak effects and having assumed that these are the only possible factors and having subtracted them out, they fit a straight line to the residuals and called that “the global warming signal”. There’s no room for any longer-term quasi-cyclical mechanisms, there’s no way to have plateauing because they fit a straight line, the factors they attempted to include are not properly represented… They proved nothing.

225. toto

     You’ve been making this claim several times. IIUC you’re saying that F&R’s main result (linear, non-plateauing temperatures once you remove fitted ENSO, solar and MEI) is just an artifact of the method, as opposed to an indication of a real relationship between the series.

     Your the sentence starting IIUC is incorrect. He is saying the multi-linear regression which assumes the trend over the full period is similar will return a trend similar to the trend in a simple linear regression which assumes the trend over the full series is similar. That is: you learn little more than you learn from a simple linear regression.

     If you get similar flatness in the final series after subtracting the fit for these fake series, then I’ll admit that F&R’s approach doesn’t tell us much about the flatness of the real “underlying signal”. If you don’t, well, what would you conclude?

     With respect to the exercise you suggest Brandon does: Maybe you should elaborate. Because it seems to me that you are suggesting he creates synthetic series that *by definition* will not have flatness on the end *except* in those randomly occurring cases where the noise happened to result in flatness. Of course he is unlikely to get flatness in your synthetic exercise.

     But I think his point is that F&R’s method will remove flatness in both of the following cases:
     1) The flatness is part of the signal.
     2) The flatness is due to noise.

     But of course: If he does your exercise and any flatness that might occur in a synthetic series occurs only because it’s noise, then it’s unlikely he will create a series that is flat at the end. But that exercise doesn’t address the problem with F&R which is that F&R’s results will not show flatness even if it’s due to the signal.

     A better exercise would be for Brandon to create a “signal” that is flat at the end. It could just be a sinusoid, or end near the top curve of a parabola or something. Add “noise” to this signal. Then apply F&R method and see if the flat end is removed from the thing F&R method identifies as “the signal”. After all– I think that’s what he’s claiming will tend to occur. (And I’m pretty sure if he does this, he will remove the flat end. The reason is that I think the F&R method assumes the trend is linear.)

226. Lucia:

     And I’m pretty sure if he does this, he will remove the flat end. The reason is that I think the F&R method assumes the trend is linear.

     Actually, I think all that’s required is that the “bend” have an overlap in frequency content with the basis functions (ENSO,MEI, etc) used to construct the filter. (I believe this assertion can be proven using a variation on the correlation theorem, which relates cross-correlation to the product of the Fourier transforms of the two series, but normalizing it for an inner product rather than a cross-correlation.)

     If you make the bend over over 10-years, you’ll find extra amplitude absorbed by solar (what troy finds), if you shorten it to say five years, ENSO should be enhanced.

227. While this runs dangerously close to “assigning homework”, in line with Lucia’s #110030, perhaps a good underlying signal would be
     s(t) = 0.01*(t-2005)+0.12*cos(2*pi*(t-2005)/65), where t is the time in years.

228. If this depends on ice cap melting to change polar albedo, then the ECS will not be reached for thousands of years, because most of the ice caps (Greenland, Antarctica) won’t melt enough to show bare ground until then. Instead, it melts from the top down. There is no way the projected warming will melt winter ocean ice at N. pole (only summer melt sooner), so this can’t be invoked.

229. HaroldW–
     For clarification:

     If A claims ‘x’ and provides no evidence for ‘x’, it is ok for anyone to tell ‘A’ that he needs to prove his claim ‘x’. That’s an appropriate ‘homework’ assignment.

     But this sort of stuff is ridiculous

     To others,

     Start to justify the arbitrary setting of the initial temperature of the CO2 added in the models (start of the discussion). Then, if you are mature enough, we can move on to more fundamental problems.

     A person isn’t allowed decree that everyone else reading is assigned some problem and follow that with the declaration that when those given the assignment have reached an appropriate level of “maturity”, “we” can move on the whatever topic the person who just dropped in thinks “we” can move onto.

     Never mind that there were all sorts of other oddities in that “assignment” some of which may be due to the fact that phi doesn’t speak English very well. For example: I have no idea whose ‘model’ might have ever set the initial temperature of CO2 to some ‘arbitrary’ value nor what value might have been chosen. I don’t know if he meant to be quite as insulting as he was when he wrote a sentence that sure sounds like he is suggesting others here are not “mature”.

     Even if his comment was a less confusing or insulting, phi sounded an awful lot like a self appointed “teacher” who dropped in and started handing out “assignments” so all his “students” could gain “maturity” and then embark on other “assignments” the “teacher” will select. This method of trying to get other people to do the work to prove whatever claim you are trying to make is not allowed. And I do think it was his method and not purely a result of poor translation into English.

     In any case, I think it best that phi find a French language blog where his difficulties where he may be able to communication without perpetual misunderstanding.

230. toto (Comment #110024)

     Here is a simple exercise to illustrate my point. You can actually do this quickly in Excel yourself. Generate a list of 1000 random numbers between 0 and pi/2, then calculate the sin and add a random number between 0 and 1/2 as noise.

     If you do a linear regression you will get a line with an R2 around 0.76. Does that mean your “true” signal is linear?

231. Troy_CA (Comment #109976)

     Troy I would appreciate your thoughts on it being the F&R methods or data pushing their result to be linear.

232. Sheesh. I come three hours after toto’s response to me, and people have already said everything I would have said (and more). You guys are too fast.

     lucia’s comment hits on the point that bugs me the most about F&R: It is begging the question. A linear regression inherently assumes an unchanging trend. That means F&R assumed there was no change and found… no change. It’s ridiculous to treat that as some sort of meaningful result.

     As for what sort of test I should do, the option I prefer is quite different than what people have suggested. Rather than create synthetic series, I tested the methodology by perturbing the system and observing the results. My favorite test was adding trends to parts of the record and seeing what happens. Does anyone care to guess what happens if you add a cooling trend to the last ~15 years of the data?

233. Brandon –
     “Does anyone care to guess what happens if you add a cooling trend to the last ~15 years of the data?”
     My guess would be that the TSI coefficient increases.

234. But I think his point is that F&R’s method will remove flatness in both of the following cases:
     1) The flatness is part of the signal.
     2) The flatness is due to noise.

     .
     I think that’s precisely the point under contention. What F&R did was fit a bunch of timeseries + a linear trend, remove the fitted time series, and look at the final results.
     .
     Do we expect the final result to look more linear that before, no matter what the timeseries are? Absolutely.
     .
     Do we expect the entire plateauing to disappear, no matter what the timeseries are? Well, I don’t think so. If the plateauing is in the underlying signal rather than the removed timeseries, there should some amount of plateauing left. How much? Well, that’s the point of the randomization experiments I suggest.
     .
     If we see that randomizing/phase-scrambling the timeseries leads to final results that are just as linear as using the real timeseries, then we can conclude that the strong linearity of the final series is just an artifact of the method.
     .
     If using phase-scambled timeseries leads to final results that are much less linear (more jitter or more plateauing) then we have to admit that there is something special about these original timeseries, namely that they do explain much of the deviation from linearity of temperatures.
     .
     Now that might be a happy coincidence. But I don’t think that was Brandon’s argument.
     .
     (PS: when I said “flatness”. I didn’t mean plateauing/zero-trend, I meant “linearity”, e.g. r(x(t), t). Sorry for any confusion.)

235. Does anyone care to guess what happens if you add a cooling trend to the last ~15 years of the data?

     .
     My WAG: You get different coefficients, different lags, and a final timeseries that has much more variance after linear detrending (or equivalently, reduced r between x(t) and t).
     .
     Depending on how much “cooling” you put, of course (considering that the last 15 years are pretty flat, small cooling is not going to change much of the final variance)

236. toto (Comment #110048)

     IIRC what F&R did was to regress
     T = a*solar+b*ENSO+c*volcanoes+d*t where t is time
     They then subtracted the first three term and low and behold – a linear trend emerged!?

     But that was implicit in their definition

     T-a*solar-b*ENSO-c*volcanoes = d*t

     No statistics required and no inferences possible.

     Now if they had added a fifth term, say e*sin((t-t0)/60) then shown that the last term did not i proved their fit then they might be able to say something. But I don’t remember that being their approach.

237. IIRC what F&R did was to regress
     T = a*solar+b*ENSO+c*volcanoes+d*t where t is time
     They then subtracted the first three term and low and behold – a linear trend emerged!?

     .
     I think you forgot the “epsilon(t)”/residual term, which IIUC they left in the final timeseries. I may have misunderstood.
     .
     If the plateauing is in the “signal” rather than in the removed extraneous components, I would expect it to show up to some extent in the epsilon, and thus in the final timeseries.

238. lucia (Comment #109991) and (Comment #109993)

     Lucia,

     First, it is your blog and you can run it how you like. I do, however find it strange that you can initiate moderation on someone who, as per your link, accused Carrick of using ’empty and irrelevant’ prose (hardly pejorative or rude), and yet you allow SteveF (#109959) to imply that phi is ‘nuts’, without any form of censure. This, IMO, strongly suggests a somewhat subjective approach to moderation.

     Second, my interpretation of phi’s argument was not that CO2 ‘was not an emitter’ (it is), but that the ‘thermalization’ of CO2 in the models was arbitrary. In this, I agree with him.

     Having said all that, it appears that he has decided to leave this thread, so any argument we may have is probably irrelevant.

239. SteveF:

     You should look at Troy_CA’s blog for a detailed analysis which shows the F&R method can’t detect a known trend in synthetic data.

     … in a model where the volcanic effects are extremely long-lived, therefore creating “room” for the supposed 60-y oscillation. Which does fool F&R’s method into believing that the oscillation is part of the volcanoes (IIUC).
     .

     A secondary problem is that the F&R methodology pretty much falls apart, especially WRT solar response, if extended to earlier periods.

     .
     Yeah, that’s my “happy coincidence” option. IIUC what I read (e.g. Van Hateren (sp?)), solar just happens to be aliased by aerosols in the recent past. I haven’t seen an estimate of how much that affects the final result though.

240. Steven Mosher (Comment #110006)

     steven,

     Wonderful, abstract fun…

     First, I don’t care if you think I don’t know what a theoretical construct is. Just because you think I don’t know doesn’t make it so.

     Second, If climate sensitivity is a metric, what quantity is it? Provide the accurate measurement of climate sensitivity. Not theoretical, not estimated, not assumed. The actual answer.

     Can’t? Nope. You would need to measure it. You would need to be able to state pretty accurately how much of the 0.8C rise in temperature since 1850 is ‘directly’ due – and proven to be so – to CO2.

     Can’t? Nope.

     So, until you – or anyone else – is able to provide that evidence, the idea of ‘climate sensitivity’ remains a theoretical construct.

241. toto

     Do we expect the entire plateauing to disappear, no matter what the timeseries are? Well, I don’t think so. If the plateauing is in the underlying signal rather than the removed timeseries, there should some amount of plateauing left. How much? Well, that’s the point of the randomization experiments I suggest.

     I’m puzzled with how you are using the term “plateauing”. When I read that word, I understand it to mean “in the early part of the series, there is a positive slope; in the latter part there is no positive trend”. Of course if you fit a straight line there is no “plateau”. You will have a straight line with the same slope from beginning to end.

     I’m also not clear about *what* you want done in the randomization experiments. Do you want someone to take a signal that looks like “positive trend for years 0-20, no trend from years 20-30″ and add noise”? That is: take a signal that has plateauing in it? And then apply the F&R method? Or what?

     Because obvious, if you add noise to a linear trend with no plateauing, then your not going to get a plateau out after you use the F&R method.

     I”m actually mystified by what your test is because you seem to be leaving some features of your test out of your description of your test. For example, when you write this:

     Well, that’s the point of the randomization experiments I suggest.
     .
     If we see that randomizing/phase-scrambling the timeseries leads to final results that are just as linear as using the real timeseries, then we can conclude that the strong linearity of the final series is just an artifact of the method.

     I understand what you hope the point of your experiments would be. The thing I don’t understand is the actual procedure that constitutes an experiment.

     Do you mean just taking a 30 year time series like GISTemp, detrending, finding the fourier components of the residuals, using those to generate new noise with similar features as the former residuals, and adding that to the trend in GISTemp? That could be called “phase scrambling”, but it wouldn’t test Brandon’s claim. That is: even if you hoped the ‘point’ of this experiment would be to learn the effect of F&R on plateauing, the experiments wouldn’t reveal anything about that.

     Or do you mean… what?

     You may actually need to do it to clarify what the test itself is and to explain how you would diagnose something from it. Failing that, you need to provide a lot more words describing precisely what you think ought to be done. (The former might be better because then we can understand what you mean by ‘just as linear’ and so on.)

     After you apply the method you have partially described (or describe what you think ought to be done more fully), then maybe I could figure out whether it tells us anything about Brandon’s claim. That said: Based on what I think you are describing, I think if you add phase scrambled or random noise to a linear trend the test will tell us absolutely nothing about whether Brandon’s claim is true or false.

242. toto –
     Here’s a basic example. Simplify F&R’s analysis: assume that the regression exactly removes “exogenic” influences such as TSI, MEI, AOD. What you’re left with is the remaining signal and noise. Now let’s say that the signal is of the form suggested in #110034 — that is, a linear signal of 1K/century and a 65-year sinusoid with amplitude (center-to-peak) of 0.12 K. In Excel I generated a temperature history of this form, with additive white gaussian noise of sigma=0.1 K. When we exclude the other terms, the F&R method becomes just computing the OLS trend. This is the result of a 30-year regression. There isn’t much difference between the true signal and the regressed one. The residuals (not shown) are nondescript. Now look at the temperature evolution over 90 years. I’ve included the original regression line, and another one which uses the first 60 years of data.
     .
     Bottom line: A regression with a linear function for the “anthropogenic influence” is going to tell you that the anthropogenic effect is linear, as John Vetterling wrote above. The residuals may not clearly show any non-linear effects, even (in this case) over half a period.

243. Re: Arfur Bryant (Feb 13 14:26),

     my interpretation of phi’s argument was not that CO2 ‘was not an emitter’ (it is), but that the ‘thermalization’ of CO2 in the models was arbitrary.

     Please explain what you mean by ‘thermalization’. I think it means something different to you and phi than it does to the rest of us.

244. When I read that word, I understand it to mean “in the early part of the series, there is a positive slope; in the latter part there is no positive trend”. Of course if you fit a straight line there is no “plateau”.

     But there will be one (possibly smaller than in the original) in the linear trend + residuals, which I think is what F&R call “the adjusted data”.

     I’m also not clear about *what* you want done in the randomization experiments.

     Take the “extraneous” timeseries (ENSO, solar, aerosols), and phase-scramble them. Then, apply (what I think is) F&R’s method: fit the real temperatures to these fake timeseries + a linear trend, then subtract the fitted fake timeseries, but keep the linear trend + residual.
     .
     If the final result is still just as linear as the original F&R final result, then the linearity in F&R Fig. 5 is an artifact of the method, which is what I understand to be Brandon’s point. Otherwise, it really is caused by a special match between temperatures and the real extraneous timeseries.

245. toto (Comment #110054),
     Humm…. so it seems you reject even the possibility of significant cyclical influence? Interesting and funny, but probably wrong.
     .
     A disinterested observer might conclude a cyclical component is far from unlikely: http://www.woodfortrees.org/plot/hadcrut4gl/from:1960
     .
     I think I will rather enjoy watching how temperatures evolve over the next decade. I think you will not enjoy it quite so much as me.

246. It’s interesting to note that if you just fit the global mean temperature to a 60-year cycle, then subtract off the residual, you get a fairly linear looking temperature trend for post-1980 too.

     figure.

     Since there is more than one way to arrive at a linear trend, linearity in the final residual doesn’t seem like a very strong test.

247. Toto,

     Take the “extraneous” timeseries (ENSO, solar, aerosols), and phase-scramble them.

     Ahh.. ok. So, phase scramble all time series except the temperature one. Now I understand what you are suggesting. I was assuming you were suggesting phase scrambling the series of most interest– the temperature series itself!

248. Paul_K,
     Your final model suggests tremendously slow response at high latitude. Is it not fairly straightforward to compare the expected warming versus latitude under a specified forcing scenario? I mean, if you take the forcing based on the 20th century warming and apply those constants, you can generate expected warming by latitude band and compare that to the observed warming at each latitude. my guess: the measured warming at high latitude has been much faster than those constants would project

249. Carrick (Comment #110065),
     Carrick, Carrick, Carrick…. If you do that it looks like forcing and warming go together pretty much has you might expect, and (worse) there is a fairly low temperature response to historical forcing. That will never do. Just ask toto.

250. You’re right Steve… definitely can’t have that. 😉

251. Humm…. so it seems you reject even the possibility of significant cyclical influence? Interesting and funny, but probably wrong.

     .
     I’m certainly not rejecting its possibility. I’m just very wary of calling “cycles!” based on two periods worth of data, especially if there are alternative explanations.
     .
     I guess the next decade or so will be interesting in that regard, especially if a strong El Nino coincides with a solar maximum.

252. toto:

     I’m just very wary of calling “cycles!” based on two periods worth of data, especially if there are alternative explanations.

     Agreed on the need to be cautious.

     The data I used HadCrut4 goes from 1850-2012+ which is 162 years. The best fit period is 56 years, which gives 162/56 = 2.9 cycles, not 2. Checkyermath.

     I don’t think that’s enough to really assume that we don’t have coincidence producing an apparent cycle, not without a model on the level of ENSO where we can really describe the phenomena anyway.

     Point wasn’t that temperature will down-turn, but it is interesting to see that fitting for it produces a very striking trend reminiscent of net forcing, as Steve pointed out..

     There is other evidence though (not sold on that either). I think there are other studies too (only the first is peer reviewed):

     http://depts.washington.edu/amath/research/articles/Tung/journals/Tung_and_Zhou_2013_PNAS.pdf

     ftp://ftp.fao.org/docrep/fao/005/y2787e/y2787e04.pdf

253. Re: toto (Feb 13 22:03),

     I’m just very wary of calling “cycles!” based on two periods worth of data, especially if there are alternative explanations.

     Did you miss this comment? Here’s the relevant quote:

     A recurrent multidecadal oscillation is found to extend to the preindustrial era in the 353-y Central England Temperature and is likely an internal variability related to the Atlantic Multidecadal Oscillation (AMO), possibly caused by the thermohaline circulation variability. The perspective of a long record helps in quantifying the contribution from internal variability, especially one with a period so long that it is often confused with secular trends in shorter records.

     353 years is nearly 6 cycles.

254. toto:

     My WAG: You get different coefficients, different lags, and a final timeseries that has much more variance after linear detrending (or equivalently, reduced r between x(t) and t).

     I find your guess interesting. You guess there would be “much more variance” in the detrended result. That is a strong guess. Significant differences in results, you say. Test it, I say.

     Fortunately, I already had the code written to do so. Opening up my testing script, I quickly pulled together three time series based on GISS. First, the original GISS data provided by F&R. Second, the same series with a negative, linear trend introduced in the last 15 years with a magnitude of -.1. Third, the same, but with a magnitude of -.2. I then plugged each of these series into F&R’s process to create an “adjusted” version. I then detrended each with a simple linear regression. These are the results:

     sd(a) = 0.119
     sd(b) = 0.118
     sd(c) = 0.119

     That’s right. Despite the linear detrending removing trends of significantly different magnitudes, the variance after was almost identical. And unlike what you might have expected, the parameters were mostly unaffected. In both test cases, the linear trend decreased (with the offset adjusting to accommodate), and as HaroldW guessed, the TSI coefficient increased. Otherwise, the parameters remained practically constant. The lags didn’t change at all.

     You could argue the perturbation here is smaller than you had in mind. Or you might argue different types of trends would have different effects. Both are true. With a large enough trend, I’d have generated results more in line with your expectations. Increasing the magnitude of my modification to .4 would have increased the standard deviation to 0.125 (and slightly altered the volcanic coefficient). A non-linear modification could have any number of effects.

     But that’s a lame defense of F&R’s methodology. F&R’s methodology finds GISS to be no less “noisy” than if one artificially adds a trend of -.13 C/dec to half the series. How can we take anything it says about the “noise” of its results seriously?

255. Toto,
     Loehle and Scafetta also identify a number of additional papers which show the quasi-60 year cycle in many different long-term proxies. Downloadable here:-
     http://benthamscience.com/open/toascj/articles/V005/74TOASCJ.htm

256. Re:SteveF (Comment #110076)
     February 13th, 2013 at 7:55 pm
     Hi SteveF,

     Your final model suggests tremendously slow response at high latitude.

     Yes, but recall my health warning.

     Is it not fairly straightforward to compare the expected warming versus latitude under a specified forcing scenario? I mean, if you take the forcing based on the 20th century warming and apply those constants, you can generate expected warming by latitude band and compare that to the observed warming at each latitude. my guess: the measured warming at high latitude has been much faster than those constants would project

     Unfortunately, it’s not straightforward. It can be done in the models because the net flux and temperature data are available. In the real world, there are no long-term net flux measurements available – and not even reliable short-term measurements available in the high latitudes north and south of 60 deg latitude. So the best you get from observational data is temperature vs time. For a constant forcing or a slowly changing forcing with time, when only a temperature time-series is available, there is no way to distinguish between a high lambda/fast response case and a low lambda/slow response case. The problem is compounded by the large local effect in the high latitudes of changing sensible heat gradients in ocean and atmosphere.

     We can speculate that with multiple frequency forcing data e.g seasonal data, it may be possible to use the amplitude vs frequency relationships to narrow down the range of uncertainty, but I suspect that this won’t be too easy or robust.

257. SteveF,
     Erratum: I meant to write”e.g. seasonal data combined with multi-decadal forcing data,”

258. Re:Nic Lewis (Comment #109978)
     February 12th, 2013 at 4:23 pm
     Hi Nic,
     You wrote:

     Surely the implications of the Armour et al paper are only important and wide-ranging if its conclusions are correct. I am somewhat doubtful whether they are.

     For the record I believe that the implications are important and wide-ranging if its explanation of the model results is correct.
     So when you say you are are somewhat doubtful that the paper’s conclusions are correct, do you mean that you doubt that the models do a good job of simulating the realworld (in which case I agree with you) or do you mean that you doubt that Armour’s explanation of why models display an increasing effective climate sensitivity with time (a curvilinear flux-temperature relationship) is a valid model (of a model)?
     The paper by Bates highlights the importance of considering regional changes rather than zero-dimensional zonal averages, but I don’t think it helps resolve this particular question.

259. HaroldW (Comment #110061)

     Harold, thanks for the analysis. It was just what I was looking for.

     F&R: IWTIT.

260. Hi Paul
     I’m not convinced that the Armour paper’s explanation of the model results is correct. At least, I think they are only a partial explanation. Surely meridional heat flow plays a major role in reducing differential latitudinal temperature changes, and moving heat to where lambda is higher? I find the fact that their CCSM4 GCM seems to have exactly the opposite pattern of latitudinal variation of the climate feedback parameter to what I understand to be the correct pattern (lambda higher in the extratropics) very disconcerting. If both their GCM and their 3-zone model results are based on the opposite of the actual latitudinal pattern of lambda, why should one think they are correct?

261. Paul_K,

     Are you using constant forcing with latitude? It isn’t, you know. Not only that, but at the new steady state, there should still be a net positive forcing in the tropics balanced by a negative forcing at latitudes above about 40 N and S.

262. ‘Second, If climate sensitivity is a metric, what quantity is it?
     change in C per change in Watts. easy.

     “Provide the accurate measurement of climate sensitivity. Not theoretical, not estimated, not assumed. The actual answer.

     Can’t? Nope. ”

     Of course you can. Your demand equivocates on the meaning of the term accurate. All measurements are estimates, so there is only measurement with its associated accuracy. that accuracy may make the metric unuseable for some purposes

     “You would need to measure it. You would need to be able to state pretty accurately how much of the 0.8C rise in temperature since 1850 is ‘directly’ due – and proven to be so – to CO2.”

     No, climate sensitivity is the system response to any change in forcing. C02 has nothing to do with the estimate of sensistivity.
     that is why, for example, why folks can measure sensisitivity by looking at the response to volcanoes. Nothing to do with C02.
     The sensisitivity to doubling c02 is the proct of two quantities.
     tell us what those quanities are? not the values.. just the quantities

     “Can’t? Nope.”

     Of course you can

     “So, until you – or anyone else – is able to provide that evidence, the idea of ‘climate sensitivity’ remains a theoretical construct.”

     it is not like you get to define what is a construct and what is not.
     You simply do not get to make up rules. Sorry.

263. DeWitt Payne (Comment #110062)

     Hello DWP,

     I explained my answer to your question in my comment #109987.

264. steven mosher…

     [“change in C per change in Watts. easy.”]

     Easy? Then quantify it! And provide real, physical, un-modelled evidence to support your assertion.

     [“Your demand equivocates on the meaning of the term accurate.”]

     Then provide your estimate of uncertainty along with your estimate of CS. Stop prevaricating. If you have any real evidence, provide it. Or, provide your estimate of uncertainty along with your estimation of how much of the 0.8C warming is attributed to CO2. Again, show evidence.

     [“C02 has nothing to do with the estimate of sensistivity (sic).”]

     steven, this is what the IPCC says about Climate Sensitivity: “It is broadly defined as the equilibrium global mean surface temperature change following a doubling of atmospheric CO2 concentration.”
     Are you sure you want to argue CS has nothing to do with CO2? Really?

     Volcanoes? Your attempt at deflection is noted.

     [“You simply do not get to make up rules. Sorry.”]

     When did I try to make rules? I am just asking you – or anyone else – to provide some evidence (real, physical, non-modelled evidence) which supports the assertion that CO2 can have a significant effect on global temperature.

     Stop evading the issue. Until you provide such evidence, the idea of a ‘Climate Sensitivity’ remains a theoretical construct.

     Goodnight.

265. Re: Arfur Bryant (Feb 14 17:50),

     Then how about defining specifically what you mean by ‘ allocated a thermal quantity ‘, because it’s still not clear what you or phi mean.

266. Paul_K, Sorry for delay and the esoteric comment. My fit to the 1% to 2x ECHAM5 experiment was actually the integral of the rate of thermal input into the pipeline (ecs*ln(1.01)/ln(2)) times my accumulation function (exp(-(n-t)^(1/3)/tau). I since have been able to find data from other models and none produced a similarly nice match. For most of these other models the quad-root of time would be a better fit as the slope after stabilization was flatter.

     My model was an exercise in curve fitting. Once I applied it to observations though, I gained an appreciation of the difficulty in constraining ECS. TCR on the other hand looks much more constrainable. In hindsight, that’s not too surprising.

     https://sites.google.com/site/climateadj/simple-model-of-models

267. Arfur.

     you make a classic mistake, Let me see if I can explain your folly

     you claim that sensitivity is a theoretical construct

     ”

     So, until you – or anyone else – is able to provide that evidence, the idea of ‘climate sensitivity’ remains a theoretical construct.”

     This mistake confuses epistemology ( what we know ) with metaphysics ( what exists )

     If a tree falls in the woods and no one is there to hear it does it make a sound?

     Lets consider the speed of light. Long before it was measured, it existed. It was not a theoretic construct, it was a thing that was tough to measure. Or take another simple one. the radius of the earth. Hardly a theoretical construct. But had you told a man who claimed the earth was roundish before he could measure it, that the “radius of the earth’ was a theoretical construct, you would have been quite wrong. So, you really have confused a question of metaphysics ( what a thing is) with epistemology ( how well do we know anything about that thing)
     So, is the age of the earth a “theoretical construct’ simply because we have estimates of it and not exact measurements of it?
     the number of stars is odd or even. is that a theoretical concept?

     I don’t mind going over these fundamentals with you if it can clarify the way you think about things.

     So, if a tree falls in the forest and nobody is there to measure the frequency of the sound it makes, does it make a sound?

268. arfur

     “Are you sure you want to argue CS has nothing to do with CO2? Really?”

     yes. Climate sensitivity is a system property that is defined as
     the response in C to a change in forcing of Watts.

     We can’t proceed unless you understand the fundamentals.

     Now, the sensitivity to a doubling of C02.. is the IPCCs clumsy way
     of speaking about the response to a change of 3.7Watts, but you need to understand the fundamentals first.

269. Mosher, would you measure the radius around the equator or through the poles.
     Thing is you get different numbers 7,926.38 mi at the equator and 7,899.84 mi at the poles.

270. Arfur

     “When did I try to make rules? I am just asking you – or anyone else – to provide some evidence (real, physical, non-modelled evidence) which supports the assertion that CO2 can have a significant effect on global temperature.

     Stop evading the issue. Until you provide such evidence, the idea of a ‘Climate Sensitivity’ remains a theoretical construct.”

     you try to make up the rules when you say this

     “Until you provide…. the idea remains a theoretical construct”
     That sure looks like a rule you want to impose on people.
     But you should now see that this rule is deeply confused.
     If you asked me how much I weighed on oct 27, 1962 and I told you that i had no idea, but that it was somewhere between 10 lbs and 100lbs ( i was 4 years old) according to your rule my weight would be a theoretical construct. You see how silly such a rule or approach is. By the way, what was your IQ on may 16th 7pm
     in 1980. No estimates, if you dont have empirical evidence and copy of the test then your intelligence is a theoretical construct.

271. Arfur, “to provide some evidence (real, physical, non-modelled evidence) which supports the assertion that CO2 can have a significant effect on global temperature” try here. Papers referenced.
     http://www.skepticalscience.com/empirical-evidence-for-global-warming.htm

     How about temperature profile of other planet? Try to compute that without GHG.

     Having extra W/m2 irradiating the surface without raising the temperature would by tough to explain with known physics.

272. Phil,

     Arfur is of a school of skeptics that seem to think that if you cannot measure it directly that it doesnt exist. This looks reasonable on the surface of things until you realize how little we know by direct observation and measurement. I would imagine that if you told him the moon was x miles away he would demand you produce the tape measure. The idea that you can measure something by using one set of values and physics to derive other values is foriegn to these types.

273. Well you can’t measure the radius of the Earth either. It’s a theoretical construct that doesn’t actually apply to our rather oddly shaped planet.

     (It’s still a useful construct though.)

274. Re: Arfur Bryant (Comment #110118) 


     When did I try to make rules? I am just asking you – or anyone else – to provide some evidence (real, physical, non-modelled evidence) which supports the assertion that CO2 can have a significant effect on global temperature.

     Arfur, the cold, hard reality of “objective” science is that no measurement is truly independent from any model whatsoever. The ways we get numbers out of an instrument is based on models. You just do the best you can.

275. Re:DeWitt Payne (Comment #110105)
     February 14th, 2013 at 2:45 pm

     Hi DeWitt,

     Yes, I have used a uniform forcing across the latitudes.
     Your point is an interesting one. Using non-uniform forcings changes the answer (obviously) but not the conclusions. If I select constant but different forcing values such that their areal weighted average = 3.7 W/m2, say, I can also change the values of lambda proportionately in each zone so that the equilibrium temperatures in each zone are retained and hence the ECS of the total system is retained. With the same response times, a plot of outgoing flux vs temperature then shows more curvature than before becasue of the larger contrasts in the assigned lambda values. It doesn’t change the more fundamental observation that to reproduce this type of curvature you need some zones with a combination of high temperature gain and a long response time.

     Not only that, but at the new steady state, there should still be a net positive forcing in the tropics balanced by a negative forcing at latitudes above about 40 N and S.

     I agree that there should still be TOA net flux differences (I wouldn’t call it a net positive forcing) in the individual zones at steady-state. These should integrate to zero across the planet. But these will exist with either uniform or non-uniform forcing if meridonial currents are accounted for.

276. Steven Mosher (Comment #110122):… “Climate sensitivity is a system property that is defined as the response in C to a change in forcing of Watts….Now, the sensitivity to a doubling of C02.. is the IPCCs clumsy way of speaking about the response to a change of 3.7Watts, but you need to understand the fundamentals first.”

     If this is true, then Climate Science is dumberer than we thought. A WMGHG forcing averaging 3.7-W/m2 is very weak compared to a planetary mechanical forcing averaging 1-W/m2 that melts and/or grows mile-thick glaciers. Asymmetrical forcings on an asymmetrical geography = huge climate change. Think of the cutting ability of a butter knife versus a razor blade with the same force applied.

     What is the forcing from aerosols? Carbon Black? Ozone? Irrigation? Deforestation? The uncertainty of these forcings is an order of magnitude greater than the *answers* to these questions. Therefore, you are right, climate sensitivity is not a theory because they haven’t even developed a conceptual model yet. We are decades away from something that could rise to the level of a theory. ;^)

     Is it any wonder that some people claim that models “thermalize” CO2. They give it the W/m2 due to physics, then based on an uncoupled, isolated water vapor feedback with little supporting field data, the models mash down on the gas pedal. Then, using the aerosol knob, they put on the brakes as hard as they can justify. With all that, the models still overshoot GAT.

     Wake me up when they have a theory that can be tested. Armour is a good start.

277. Re: Howard (Feb 15 11:07),

     A WMGHG forcing averaging 3.7-W/m2 is very weak compared to a planetary mechanical forcing averaging 1-W/m2 that melts and/or grows mile-thick glaciers.

     Did you perhaps mean strong rather than weak?

     The swing in peak insolation at 65N from the precession of the equinoxes is on the order of 100 W/m² even though the global insolation hardly changes. The onset and end of glaciation periods is driven by albedo change which changes the global forcing by significantly more than 1 W/m². Hansen’s (I think) estimate of the change in global albedo from glacial to interglacial is ~3.5 W/m², which is, IMO, on the low side. The change in CO2 from 180 to 280 ppmv adds an additional forcing of ~3.3 W/m². So now we have a lower limit on the order of 7 W/m² for the forcing change from glacial to interglacial. IMO, it’s at least 10 W/m². If the global average temperature change from glacial to interglacial is 5 C, then a forcing change of 7 W/m² implies an ECS of 0.7 C/W/m² or a temperature change from doubling CO2 of 2.6 C.

278. DeWitt:

     The albedo is a positive feedback (not a forcing) from a very small change in average forcing. The 50- to 100-W/m2 focused on the northern hemisphere is what does the work to create large feedbacks from changing albedo, ocean circulation, dust, water vapor, Ch4, CO2 etc, etc. It’s like taking a magnifying glass to fry ants. I haven’t seen any study that definitively detects a unique CO2 temperature signal in geological samples from the major glacial cycles. Also, I have yet to see a paper that clearly demonstrates how climate responds to very different forcings in exactly the same manner with the same net feedbacks and resulting GAT. What you describe are very rough estimates of global average forcings from the local effects of positive glacial feedbacks derived from simplistic models. It’s circular logic and avoids tackling a very complex field data and numerical modeling problem. Modern CO2 increasing is a forcing, along with many other human climate forcings that are currently poorly studied and generally ignored. Pielke Sr. gets pilloried for mentioning them! What is the net human caused modern forcing? It’s unknown and is very poorly constrained by advanced models that are unsophisticated compared to the complexity of the task.

279. steven mosher #110121) et al…

     Steven, the mistake is not mine.

     Epistemology. A branch of PHILOSOPHY concerned with the nature and scope of knowledge. Largely how knowledge relates to the notions of TRUTH and BELIEF. It is possible to believe something is true based on belief, but still be wrong.

     Metaphysics: A branch of PHILOSOPHY concerned with explaining the nature of the world (and being). Prior to modern times, scientific questions were addresses as the part of metaphysics known as ‘natural philosophy’. However, the scientific method transformed natural philosophy into an empirical activity derived from experiment. This distinguishes SCIENCE from PHILOSOPHY. To be termed scientific, a method of inquiry must be based on empirical and measurable evidence subject to specific principles of reasoning (from Newton). So, Steven, since the 17th century, science distinguishes itself by using systematic observation, measurement, and experiment, and modifying hypotheses as required. See where I’m going?

     In epistemology, belief is any cognitive content accepted as true whether or not there is sufficient proof.

     Truth is not a necessary bedfellow with belief. If something is KNOWN, it cannot be false, but belief does not mean it is true.

     Knowledge requires TRUTH. Knowledge requires that a belief should have justification in order for it to be TRUE. a belief COULD be true but it could simply be true because of luck. Reliability of some sort is required for the justified belief to be TRUE. So… Climate Sensitivity…

     I am extremely interested to hear you try to distance CS from CO2. Whether you personally accept the IPCC definition or not, you are going to struggle with defending your position that CO2 has NOTHING to do with CS. Even if you were to define CS as a response to a generic ‘radiative forcing’ increase, CO2 is considered by warmists and lukewarmers alike as being a fundamental part of radiative forcing in the context of the GHE. If you dismiss CO2 as a contributor to the notion of CS, how do you explain the CS figures which you yourself have discussed on several occasions, such as defining what a lukewarmer is by using the CS metric?

     Anyway, back to theoretical constructs. I understand your abstract arguments about measurements and I understand YOU are missing the point. If you can’t measure CS, and you can’t give any evidence to support your belief that it actually HAS a measurement, how can you say it is NOT a theoretical construct?

     I can see you want this to be an argument based on labels. I would ask that you do not disparage or patronize my intelligence or my scientific ability and simply read what I write. If you can find a hole in my reasoning or logic, then by all means try to correct me. However, you should also be aware that I reserve the right to do the same to you. I’m happy to debate the issue of CS, and I am happy that it remains a theoretical construct until you – or anyone else – can provide real, physical, un-modelled evidence to the contrary. Please try to remember that assumption is, quite possibly, the mother of all f***ups. (A quote from Under Siege 2, I think!) It would not be reasonable for you to argue that CS is NOT purely theoretical without at least some sort of evidence to back up your assertion. You would then be guilty of arguing from an unjustified belief – in epistemological terms! 🙂

     As for CS being C per Watts. What is C? The temperature change from radiative forcing? You cannot identify C, unless you consider that ALL the warming since 1850 (0.8C) represents C. That, however, infers that there is no room for warming due to natural factors. Provide (real) evidence which identifies the portion of the 0.8C which you would use in your equation for CS.

     Be objective. Stop evading the issue. Stop arguing abstracts. Either produce evidence or admit Climate Sensitivity exists in the realm of theory (and assumption).

280. Phil Scadden (Comment #110125)

     Paul, I’m not interested in other planets. Lets just stick to this one, ok?

     Your link in no way shows any direct causal evidence between increasing CO2 and warming. You have warming on one hand, and increasing CO2 on the other. This is not ‘real’ evidence. The site uses peculiar reasoning to justify its argument. “Because there has been warming… and because there has been an increase in CO2… CO2 must be the cause.” Er… Ok, I like to think of myself as objective. The warming may have been due to CO2, but I cannot be certain. So if I were you (or John Cook) I would try to justify my belief. If CO2 causes warming, why has the temperature remained essentially flat for nearly 15 years? If CO2 has a significant effect on global temperature, why has there been only a 0.8C increase in 162 years commensurate with a 40% rise in CO2 (and other non-condensing ghgs)? How much of the 0.8C rise is directly attributed to CO2? How much of the pre-1850 GHE was attributed to CO2 and how much of today’s GHE is attributed? These are questions that would have to be answered before I would even contemplate believing that the known radiative properties of a molecule of CO2 can be applied to the atmosphere as a whole! The skeptikalscience alien landing on Earth, and seeing three cars, all of which were red, would return to his planet confidently stating that Earthlings had four wheels and they were red in colour!
     You should tell John Cook that arguing from an assumption leads to an epistemological impasse! (Sorry, a bit of crossover from my chat with steven mossier… 🙂

281. “So, is the age of the earth a “theoretical construct’ simply because we have estimates of it and not exact measurements of it?
     the number of stars is odd or even. is that a theoretical concept?”

     Lucia,

     Please moderate Steven Mosher for breaking the asking rhetorical questions blog rules.

     These questions have nothing to do with climate science.

     Andrew

282. Oliver (Comment #110128)

     [“Arfur, the cold, hard reality of “objective” science is that no measurement is truly independent from any model whatsoever. The ways we get numbers out of an instrument is based on models. You just do the best you can.”]

     Oliver, I understand the spirit of your comment except for the last sentence.

     IMO, doing “best you can” is not compatible with making an assumption that a theory is correct and then desperately trying to come up with explanations why your theory isn’t working! Throw the theory out. Be objective. Start again. Accept that there was a GHE before 1850 (IPCC accurate data start) and that the total rise in global temperature is 0.8C in 162 years even though non-condensing ghgs have increased by around 40%. Work on that and re-think the theory.

     Regards,

283. DeWitt Payne (Comment #110119)

     DWP,

     This is what phi said:

     [“Models thermalize arbitrarily the added CO2. Until proven otherwise, there is no justification for this operation which however sets the value of the radiative effect.”]

     Climate models tend to allocate a warming property to additional CO2. How much warming is decided arbitrarily. There is no ‘real’ evidence to support the figure of climate sensitivity, hence it is an arbitrary figure based on the modeller’s whim.

     That is all. That’s is what I said to Lucia. It seemed clear to me. I did not initiate the word ‘thermalize’.

284. Arfur Bryant (Comment #110153)

     Phil Scadden, my apologies for typing Paul, instead of Phil! I beg forgiveness…

285. Re: Howard (Feb 15 15:50)

     I haven’t seen any study that definitively detects a unique CO2 temperature signal in geological samples from the major glacial cycles.

     Why should you expect a unique geological temperature signature from CO2 from the glacial cycles? Warming is warming. While it’s not from a glacial cycle, what about the PETM?

     You have a short (in geologic terms) spike in global temperature (~6C) coincident with a spike in CO2. The evidence for a spike in CO2 is a sharp reduction in δ13C and dissolution of carbonate deposits on all ocean basins. At that point there weren’t any ice caps to speak of as it was close to the Eocene Optimum, much warmer than current temperatures*. Post Industrial Revolution fossil fuel burning has also produced a reduction in δ13C because fossil fuel is depleted in 13C.

     *That puts comments like “Antarctica will be the only habitable continent” in perspective.

286. “Climate models tend to allocate a warming property to additional CO2. How much warming is decided arbitrarily.”
     This is demonstrably false – look at the code. The radiative properties of the gas can determined by quantum theory but were measured before that was possible. From that observation Arrhenius was able to predict global warming before it happened. Those graphs of spectral change due to CO2 are measured and match the theory. You are missing the important bits about measuring the change in DLR and changes in the spectrum. You take basic physics to make a prediction about DLR, about the spectral change to TOA, about the surface temperature and lapse-rate of earth and other planets and verify these by experiment. That is how science works. If you want to see the maths used for these codes, then see Ramanthan and Coakley 1978.

     ” How much of the pre-1850 GHE was attributed to CO2 and how much of today’s GHE is attributed? ” So why dont you look it up?

     And other planets matter because basic physics is same there too. Without GHGes our planet would be far colder.

     As to current temp, note that climate responds to total forcing. The arguments earlier on this thread are about to work those out.

     “Work on that and re-think the theory.” Matches the theory pretty well. It is clear though that you are attacking a theory you do not understand. Take some time to learn what the real theory is instead of a straw-man version of you own invention.

287. “Climate models tend to allocate a warming property to additional CO2. How much warming is decided arbitrarily. There is no ‘real’ evidence to support the figure of climate sensitivity, hence it is an arbitrary figure based on the modeller’s whim.”

     Arfur, you are simply wrong.

     There is no warming added for C02. There is a radiative transfer model. Typically it is a band model as opposed to a LBL model. These models are verified by empirical measurement.
     As C02 increases or methane or any gas, the model calculates an increase in forcing. This increase is due to the well known physics of radiative transfer.

     Sensisitivy is a model output. You double C02. That increases forcing. The system responds. When he system achieves equillibrium you measure the final temperature. If you increase C02 from 280 to 560 some models will increase by 2.1C others 4.4C

     The difference between these is a function of how aerosols are handled.

     Read the people here. There are problems with the models but they are not the problems you are imagining. They are deeper problems, more critical problems that you imagine. The problem I have with your position is that you are making a weak argument against models and sensivity, a wrong argument and you should be paying attention to Paul Ks stronger argument and steveFs stronger arguments.

     Its like people focusing on the weak arguments in climategate rather than the strong arguments. The weak arguments are easy to handle.. the strong ones.. they avoid.

288. Re: Arfur Bryant (Comment #110155)

     Oliver (Comment #110128)
     [“Arfur, the cold, hard reality of “objective” science is that no measurement is truly independent from any model whatsoever. The ways we get numbers out of an instrument is based on models. You just do the best you can.”]

     Oliver, I understand the spirit of your comment except for the last sentence.
     IMO, doing “best you can” is not compatible with making an assumption that a theory is correct and then desperately trying to come up with explanations why your theory isn’t working!

     The intended spirit is that you use the best models appropriate for the purposes intended. There’s no basis for tossing out models “on principle.” What we usually consider “measurements” are just outputs from the least-negotiable models.

     Also, any given prediction is usually based on theory but also a long chain of intervening details. A discrepancy with measurements in one such case doesn’t automatically empower one to discard the basic theory if it’s been well-established in many other test cases. Instead, the obvious first place to look is in the chain of details.

289. DeWitt:

     Warming is warming, sure. However, not all forcings are equal and they will produce different feedbacks and different net warming. Clearly, a small net forcing that increases solar on N Hem Ice Sheets has an overall huge net GAT change due in large part from the associated forcings.

     It’s important to detect a unique CO2 signal following the LGM because there are assertions being made that most of the GAT increase following the LGM was due to 100ppmv of CO2 and 300ppbv of CH4 and their associated feedbacks and not the small net insolation forcing and the huge associated feedbacks.

     The PETM does not tell us anything about what is going on now. The ocean circulation was greatly different due to an open Panama and closed Drake. While there was a huge carbon anomaly, there are many competing theories of what actually happened. The geology is not settled, no matter how attractive this event might be as an example for what me might face.

290. P.S.: In looking back at some of the comments posted over the last few days about thermohaline circulation and the AMO, I just want to register a comment that the Gulf Stream and most of the northward oceanic heat transport in the Atlantic is largely NOT “thermohaline” at all — it’s the result of the wind-driven gyre circulation. The sinking latitude will certainly influence what happens in the North Atlantic, but that isn’t the same as saying the poleward flow is “driven” or “controlled” by deep convection.

     P.P.S.: As several posters suggest above, it seems likely that the AMO is probably not “driving” variation but is a proxy for whatever is underlying the variation.

291. Re: Howard (Feb 15 20:37),

     there are assertions being made that most of the GAT increase following the LGM was due to 100ppmv of CO2 and 300ppbv of CH4 and their associated feedbacks and not the small net insolation forcing and the huge associated feedbacks.

     Those assertions would be just as wrong as asserting that CO2 had no effect at all.

292. Re: Oliver (Feb 15 20:56),

     I just want to register a comment that the Gulf Stream and most of the northward oceanic heat transport in the Atlantic is largely NOT “thermohaline” at all — it’s the result of the wind-driven gyre circulation.

     Wunsch, for one, would seem to agree with you.

293. Re:Nic Lewis (Comment #110098)
     February 14th, 2013 at 12:09 pm

     Nic,

     Surely meridional heat flow plays a major role in reducing differential latitudinal temperature changes, and moving heat to where lambda is higher?

     That would have been my first guess, but the reality , if you believe Zelinka, is that it generally does not work this way in the models. Polewards heat flow is INCREASED in the models on average with the effect of enhancing sub-polar temperature amplification. There is a large increase in sensible heat flow from the tropics to the midlatitudes. Most of this heat is then dumped into the subpolar regions.
     The change in meridonial flux solely due to surface temperature change is in the direction one would expect – a reduction of polewards heat transport. This effect is however overwhelmed by other influences in the models. See Figure 7 of Zelinka:-
     http://www.atmos.washington.edu/~dennis/Zelinka&Hartmann_2012.pdf

     Polar amplification seems to be a near-universal feature of the GCMs, but I think I am going to have to accept your pushback. If Figure 3 in Zelinka is valid, then I guess we conclude that many models explain polar amplification not with a low relative magnitude feedback, but with a high magnitude feedback and an enhanced meridonial heat flux.

294. AndrewKY

     Lucia,

     Please moderate Steven Mosher for breaking the asking rhetorical questions blog rules.

     These questions have nothing to do with climate science.

     Andrew

     If they were rhetorical questions, I’d say something to Mosher. But
     Those questions look like actual questions to me. I’m actually interested in reading Arfur’s defintion of “theoretical construct” so I hope he answers Mosher’s questions.

     As for having to do with climate science, it seems to me Mosher is asking Arfur to clarify what he meant by theoretical construct in his earlier discussion. If Mosher’s question is somehow not permitted based “not being about climate science”, then Arfur’s idea that “being a theoretical constructs” having something to do with climate science must be off limits too. Seems to me that’s a bit loopy. (Anyway, what is or is not a “theoretical construction” would appear to have something to do with science. So, it’s all ok.)

295. Arfur

     Climate models tend to allocate a warming property to additional CO2. How much warming is decided arbitrarily. There is no ‘real’ evidence to support the figure of climate sensitivity, hence it is an arbitrary figure based on the modeller’s whim.

     That is all. That’s is what I said to Lucia. It seemed clear to me. I did not initiate the word ‘thermalize’.

     The italicized sentence seems to just be incorrect by any normal meaning of the word “arbitrarily”. The last sentence would appear to be a conclusion based on the italicized one– and so is unsupported.

296. Re: Paul_K (Feb 16 04:28),

     Polar amplification seems to be a near-universal feature of the GCMs,

     That seems perfectly reasonable given that polar amplification is observed to happen in the real world. High latitudes cool a lot more during glacial epochs than the mid-latitudes and tropics according to the ice core records. Chylek, et.al. also report polar amplification during the 60 year oscillation.

     In the tropics, the energy content per cubic meter of air is dominated by the heat of condensation of the water vapor content. At constant RH, the energy content goes up exponentially with temperature. So at the same flow rate, energy transfer by circulation should also increase with temperature.

     Given that, as Oliver pointed out, ocean surface circulation is driven mainly by wind, then a reduction in poleward oceanic heat transfer also implies a reduction in average wind velocity, assuming the oceanic heat transfer is being modeled correctly.

     The other interesting result in Zelinka is that the current northward bias in energy transfer will not only remain present, but increase. I suspect that this is further confirmation that quasi-cyclic processes as exemplified by the AMO index are not present in the models.

297. DeWitt:

     I agree. CO2 has an effect, but my conceptual model is that each type and style of “forcing” has a unique set of associated feedbacks that are also influenced by the climate regime of the day. This is why identifying a specific WMGHG signal in the glacial cycles is important for our situation today as it is the closest thing we have as a paleo analog. However, it is likely an impossible task because WMGHG overall contribution may be ~10% of the total.

     This is also why quantifying all of the other possible human forcings and their associated unique feedbacks is so important. Given the pre-CO2 temperature rise correlated with the industrial revolution, it is very likely that many other human activities influence climate. With the knowledge of what these forcings might be, perhaps there are measures to take that are significantly cheaper than decarbonization that may also have immediate environmental and human health benefits.

     Triage seems to be a four-letter word in the Climate Industry.

298. “Arfur’s idea that “being a theoretical constructs” having something to do with climate science must be off limits too.”

     Lucia,

     You make the rules here, as you are fond of reminding us. 😉 The only rule I am aware of is about against asking the wrong questions, not about making the wrong statements. So it’s the questions that are the issue.

     Andrew

299. Re: Howard (Feb 16 10:02),

     my conceptual model is that each type and style of “forcing” has a unique set of associated feedbacks that are also influenced by the climate regime of the day.

     It’s possible that the transition path for a change in CO2 would be somewhat different than that caused by a change in, for example, TSI. But it’s only the path that will differ. The final steady state won’t be different, IMO. What I doubt is that the geologic record or even an ice core has the resolution to distinguish between different forcings. If it were possible, someone would likely have done it by now and we would have a much better estimate of the relative contributions of CO2 and albedo to the transition to the Holocene and much less controversy over the PETM.

300. Paul,

     You responded to my comment
     “Surely meridional heat flow plays a major role in reducing differential latitudinal temperature changes, and moving heat to where lambda is higher?”
     with
     “the reality , if you believe Zelinka, is that it generally does not work this way in the models. Polewards heat flow is INCREASED in the models on average with the effect of enhancing sub-polar temperature amplification.”

     I don’t think that these two statements are inconsistent. I was talking about the absolute effect of meridional heat flow being to reduce temperture differentials, whereas Zelinka is talking about the derivative of such meridional heat flow with increasing global mean temperature.

     “If Figure 3 in Zelinka is valid, then I guess we conclude that many models explain polar amplification not with a low relative magnitude feedback, but with a high magnitude feedback and an enhanced meridonial heat flux.”

     Yes, I think that is right. The models have only small negative, or even positive, feedback in the tropics, as with constant relative humidity the water vapour feedback is extremely strong there. Water vapour feedback decreases much faster than temperature feedback with latitude, and net feedbacks become, on the whole, increasingly negative towards the poles. Fig. 7 of Zelinka shows that the enhanced meridonial heat flux is carried from the tropics by the atmosphere, not the ocean.

301. Re: Nic Lewis (Feb 16 13:40),

     Fig. 7 of Zelinka shows that the enhanced meridonial heat flux is carried from the tropics by the atmosphere, not the ocean.

     I don’t know why this result should be surprising. About half of the current meridional heat transfer is by air circulation and the percentage increases with latitude.

302. DeWitt:” But it’s only the path that will differ. The final steady state won’t be different, IMO. ”

     By this logic, if a 1-w/m2 average global forcing is responsible for melting mile thick glaciers and raising GAT by 6-degC, then a forcing CO2 doubling will melt Antarctica and evaporate the oceans.

     Also, there are orders of orders of magnitude more signal remaining from the LGM versus the PETM. I’m sure kids in future not contaminated by CO2 inspired academic narcissism will figure it out.

303. Why does Teff decline for the first 30-years in Armour? I can’t find an explanation for this. Is it a numerical artifact?

304. Re: Howard (Feb 16 15:26),

     By this logic, if a 1-w/m2 average global forcing is responsible for melting mile thick glaciers and raising GAT by 6-degC, then a forcing CO2 doubling will melt Antarctica and evaporate the oceans.

     Nope.

     You’re still forgetting that the melting of the mainly North American glaciation is thought to be due to a very large swing in local insolation, not to a miniscule change in global insolation. The change in global insolation during the Milankovitch cycles is much less than 1 W/m² and could have nothing whatsoever to do with the glacial/interglacial cycles.

     The forcing from an albedo change from melting the continental ice cap in the NH was far larger than would be obtained from melting the Antarctic ice cap (Greenland is trivial by comparison) because it was at lower latitude and covered much more area than the the Greenland and Antarctic ice caps, which would take tens to hundreds of thousand years to melt. Antarctica has a surface area of 13.2 Mm², that’s only 2.6% of the total area of the planet or about 9% of the land area. The ice coverage during the last glacial maximum was 32% of the land surface (10% of the Earth’s surface area) and sea ice likely extended much further towards the equator as well. That’s lots of albedo change that isn’t there any more.

     Besides, we know the planet is habitable when there are no polar ice caps.

305. Phil Scadden (Comment #110159)

     Phil,

     [“The radiative properties of the gas can determined by quantum theory but were measured before that was possible. From that observation Arrhenius was able to predict global warming before it happened.”]

     That’s the whole point, Phil. Your talking about radiation, not heat. Arrhenius predicted about 6C warming from a doubling of CO2! Since then, the figure ‘predicted’ for CS has steadily reduced. Now even lukewarmers are talking about 1 to 1.5C. I have never denied a radiative property for CO2 (see my earlier posts). It is the theoretical warming attributed to CO2 that I argue with. Even then, I do not deny there may be a warming effect (I am not a ‘skydragon’). What I do say – and have said over and over on this blog before – is that no-one can ‘accurately’ state how much warming can be attributed, and that most folk overstate any warming effect. As years have gone by, the observed data has not supported the sort of warming that Arrhenius first predicted. It was a prediction based on theory. That is why Climate Sensitivity is a theoretical construct. It has no accurate physical quantification (as yet) because no-one can state with any conviction (if they are objective) how much of the observed warming can be attributed to CO2!

     [“You take basic physics to make a prediction about DLR, about the spectral change to TOA, about the surface temperature…”]

     The ‘prediction’ is based on theory. Models are not experiments in any real sense. Observations might support the models and prediction but, so far, observed data does NOT support Arrhenius’ theory.

     I understand the theory, Phil. I don’t agree with it. You are arguing from the theory forward. I am arguing from the data back. The two don’t meet.

     Nothing I am saying is a strawman argument. I am arguing ‘directly’ against the theory that CO2 (and other non-condensing ghgs) can have a ‘significant’ warming effect.

     As to 1850 and the GHE – I’ll give you an example of ‘looking it up’. Lacis states that CO2 contributes 20% to the GHE and 5% fro other nghgs. 25% of 33C (if you agree with 33 – or use your own) is 8.25C. In 1850, the GHE was 32.2 (as the enhanced GHE since 1850 is measured at appx 0.8C). 25% of 32.2 is 8.05C. So you are telling me that 280ppm CO2(+other nghgs) in 1850 equates to 8.05C and that a 40% (at least) increase in these nghgs since 1850 has only led to an increase of 0.2C [8.25-8.05]). This makes no sense UNLESS Arrhenius’ theory was wrong. In which case, start re-visiting the theory.

     Real physical evidence is required to back up Arrhenius. Models simply aren’t good enough.

306. Steven Mosher (Comment #110160)

     [“These models are verified by empirical measurement.”]

     Not for warming, they’re not. Radiation and heat are not the same thing. Hindcasting doesn’t count. How many of the pre-hindcasted models have been proved correct? Heck, Lucia has shown that!

     [“Sensisitivy (sic) is a model output. You double C02. That increases forcing. The system responds. When he system achieves equillibrium you measure the final temperature. If you increase C02 from 280 to 560 some models will increase by 2.1C others 4.4C”]

     The problem with that is that is you are not using physically-measured warming from CO2 to come up with your Climate Sensitivity! You are, in effect, admitting that CS is a theoretical construct. You are assuming that the forcing can be assumed to be a warming. This might be correct, but it has not been demonstrated as being correct. CO2 has increased since 1850, and particularly since about 1950, but the warming periods of 1910-1945 and 1975-1998 are almost identical and the cooling periods pre 1910 and from 1945-1975, along with the lack of warming since 1998, count against the assumption.

     cAGW is about warming, not radiation. Because pro-cAGW commenters assume there is ‘an amount’ of warming associated with a doubling of CO2, they support the idea of using the forcing as a warming input. There is NO physical, non-modelled evidence to support this assumption!

     I don’t mind if you think my argument is weak. Provide the physical evidence to support the CO2 = (significant) warming theory and my argument is over. (Please note I do not deny the theoretical possibility o fa negligible warming effect.)

     My argument is not against the radiative properties of a single molecule of CO2 but the ASSUMPTION that these properties enable CO2 to have a significant effect in the atmosphere.

307. Oliver (Comment #110161)

     Oliver,

     I don’t disagree with your comment in the main. However, it is not a discrepancy ‘in one such case’ in the cAGW debate. I have asked repeatedly for someone to provide evidence which indicates exactly (ish) how much of the observed warming since 1850 is attributed to CO2. No-one seems able to do that. So how can you tell how well the models are doing? Even if ALL the warming was due to CO2 (how likely is that) there are too many anomalies in the temperature datasets against the steady-ish CO2 rise to provide corroboration.

     In the area of ‘climate science’, the models have not performed well. So maybe it really IS time to re-visit the theory!

308. Lucia,

     My use of arbitrary is a follow-on from my interpretation of phi’s comment. If there is no physical, measured, quantity available for the modeller to input climate sensitivity, then he or she has to estimate one. The amount chosen is, essentially, unknown and, as such, is by personal choice. Hence arbitrary.

     As for ‘theoretical construct’, please see my response to steven mosher above.

309. DeWitt, I don’t know if this got published yet, but it was an interesting comparison of LGM to pre industrial ice albedo.

     http://kiwi.atmos.colostate.edu/pubs/Burt_Randall_Otto-Bliesner_submitted.pdf

310. Re:Nic Lewis (Comment #110182)
     February 16th, 2013 at 1:40 pm
     Nic,
     I e-mailed Dr Armour regarding the challenge raised by the distribution of feedback values shown in Zelinka et al, and asked him about the possibility of his results being model-specific to CCSM4. He sent a thoughtful and reasonable response.
     If you can be patient for a little while, I will share it as an update tomorrow (with his permission). It is very late here!
     Paul

311. Arfur–
     The climate sensitivity is not ‘chosen’ by a modeler– arbitrarily or otherwise. It is not “input” into any model. It is an emergent property.

     Andrew_KY-

     The only rule I am aware of is about against asking the wrong questions,

     You wrote this:

     Please moderate Steven Mosher for breaking the asking rhetorical questions blog rules.

     These questions have nothing to do with climate science.

     Mosher didn’t ask a rhetorical question and so did not violate the rule about rhetorical question.

312. “Nothing I am saying is a strawman argument” Your assertions about climate theory in your posts above absolutely contradict you. Since you dont know what the theory is you dont even realise this.

     Radiation is energy – if you dont know the connection between radiation/temperature and heat then its time to inspect a text book. An increase in surface radiation from GHG is measured not inferred. Arrhenius did not understand much of the physics we do now. We have moved on so how about discussing Ramanathan and Coakley? If you read those papers you will see the match between theory and observation is beyond dispute. The “model” Mosher means is not a GCM – but a solution to the RTEs and these are absolutely confirmed by observation – in the lab, from ground looking up, from satellite looking down. You either havent read the papers or you havent understood them. Further argument is impossible until do.

     What is not beyond dispute is the amount of feedback. Anything (change in solar,aerosols, GHG) that varies the amount of energy reaching the surface will change the temperature. Direct temperature increase from doubling CO2 is only about 1 degree. However, this sets up feedback that further changes GHGs (change in water vapour in short term, CO2 and CH4 on longer time scale), albedo (changes to ice, clouds) and even aerosols. Feedback can positive or negative and the net result is the climate sensitivity. GCM model the processes in the atmosphere and ocean and sensitivity comes out. It is not a number you punch in. The fact that you dont know this further demonstrates how little you understand the theory you are attacking.

313. Nic Lewis et al,
     Please note 17th Feb update appended to original article.

314. arFur said

     ” As years have gone by, the observed data has not supported the sort of warming that Arrhenius first predicted. It was a prediction based on theory.”

     arFur, Statistical mechanics was not even developed as a theory in 1900. In 1924, Bose wrote a paper that eventually became Bose-Einstein statistics. I say all this, arFur, because accounting of photons of EVERY wavelength is important to get the theory correct.

     Arrhenius didn’t have a chance, may have made a mistake, etc, I really don’t care. As David Goodstein stated in his book “States of Matter”,
     “Physics, I think, should never be taught from a historical point of view—the result can only be confusion or bad history—but neither should we ignore our history.”

315. Paul_K (Comment #110206)

     That was a thoughtful reply. Why does the science seem so less settled when you hear comments directly from the scientists’ mouths than when you hear digested versions through the IPCC? Or is that just me?

316. DeWitt:

     I agree to disagree. The asymmetrical forcing (the 50 to 100-W/m2 like a magnifying glass frying ants) due to insolation is the primary climate driver while change in albedo is merely a feedback. This is how the alarmists imagine climate disaster: CO2 warms only so much due to radiative physics, then water vapor and albedo feedbacks add up to 3-times the radiative forcing of CO2.

     Albedo and ocean current changes from insolation forcing are feedbacks. Obviously, quite a lot of ice needs to melt before albedo has a significant role. Likewise, during glaciation, the asymmetrical forcing does the heavy lifting of changing the climate regime by expanding the N Hem winter to create the changes in albedo.

     Another example is black carbon. It probably has a net primary forcing of zero and all of it’s GAT effects will be from feedbacks. Lets say the forcing is 0.1=W/m2 and the temperature change is 1-deg C, therefore, the sensitivity is 10-deg C per W/m2. Like insolation, it has a very high feedback resonnse compared with the overall energy change.

     I like to think of this as leveraged forcings: unique forcings must have unique feedbacks and will have different associated climatic responses. The earth is not homogeneous and isotropic. Armour is a great first step in addressing these complexities.

     An illustration is rainfall. Lets say we have 2-feet of rainfall in one month. In one year, the rain comes at a rate of less than an inch per day and nothing interesting happens… essentially very few feedbacks. In another region, the rain comes in 2-days causing landslides, flooding, losses of topsoil, etc. The overall “forcing” is the same, but the feedback effects are dramatically different. CO2 is like a steady rain compared with a highly leveraged biblical flood like insolation.

317. Re: Howard (Feb 17 11:53),

     In the case of glacial/interglacial transitions, albedo is more properly considered a forcing, not a feedback. Forcings cause temperature change. Feedbacks are a result of temperature change. Before a glacial/interglacial transition occurs, a significant global energy imbalance doesn’t exist until ice area increases or decreases, changing the amount of solar radiation that is absorbed. Then an imbalance exists, the temperature begins to change and feedbacks can occur.

     Anthropogenic CO2 from fossil fuel burning, cement manufacture and land use/cover changes due to agriculture and other human activities and other well-mixed greenhouse gases of human origin like CFC’s and HCFC’s are forcings. The change in CO2 between glacial and interglacial periods is a feedback. Water vapor is always a feedback.

318. Paul,

     Many thanks for obtaining Dr Armour’s informative comments (and to Dr Armour for giving them and agreeing to their being posted).

     I was aware that the Zelinka paper used the change in global mean rather than local temperature in computing net feedback. But I don’t think that makes a huge difference until one approaches 60N or 60S. Over the vast bulk of the Earth’s surface, the latitudinal dependency of net local feedback in the CCSM4 GCM used inthe Armour study seems completely at odds with that of the almost all, if not all, the 12 GCMs studied by Zelinka and Hartmann (which include the CCSM3 model).

     I have had a look at how the CCSM4 model compares with other CMIP5 GCMs in the forcing analysis by Forster et al (doi:10.1002/jgrd.50174, 2013, JGR). It has a rather large preindustrial TOA radiative imbalance (2 W/m^2), but several other models have larger imbalances. It looks middle of the road on most other measures, apart from its Adjusted forcing being towards the top of the range.

     One other interesting point is that the implicit increased poleward meridional heat flow simulated by the Zelinka GCMs is linked to tropospheric temperatures increasing faster in the tropics than the poles notwithstanding that the opposite is (at least in the NH) true for surface temperatures. Hence the famous tropical “hot spot”. This makes sense in theory, assuming relative humidity is constant, since the lapse rate will decrease with temperature. But observational evidence for the hot spot seems lacking. So could it be that all the other models are wrong about poleward meridional heat flow increasing?

319. lucia (Comment #110199)

     That’s a fair point and I was wrong to use the term climate sensitivity when I should have used ‘warming component’, or similar. However, how do the models emerge with a climate sensitivity unless they have had an input which allocates a positive warming effect due to increased CO2? That is where the ‘thermalizing’ of CO2 that phi was talking about comes from.

     You yourself have shown that virtually ALL the models have projected a higher level of warming than has been observed. Hence the ‘thermailizing’ factor must have been overstated.

     I repeat that this, IMO, was the point phi was making.

320. Phil Scadden (Comment #110203)

     Phil,

     1978? Really? I ask you to provide real, physical, non-modelled evidence and all you can give is a paper from thirty-odd years ago that uses models to come up with the temperature increase from radiative forcing? This is not evidence, despite however much belief you have in the theory. “The flux changes and the radiative-convective model surface temperatures were calculated by using the model developed by J A Coakley.”

     Well, colour me unimpressed!

     You have completely ignored my point about Lacis et al and their estimation of the contribution made by CO2 and other nghgs. All yuou can offer is ” you don’t understand the theory” and other appeals to the authority of a bunch of advocates who think that computer programming can be a substitute for the complexity of the atmosphere.

     Either provide real evidence or stop trying to tell me that I can’t see the complete lack of intellectual or logical reasoning that seems to be ‘de rigeur’ in the pseudo world of warmist climate science.

     Feedbacks? Don’t make me laugh. Provide any evidence that they actually exist, other than in your head. Then provide a reason why your feedbacks haven’t had any measurable effect in 162 years. The 0.8C rise since then includes ALL forcings and ALL feedbacks. And you still can’t say with any certainty how much of that 0.8C is due to CO2…

     The lack of objectivity in your argument is staggering. Like I say, you are arguing from the theory forward. I am arguing from the data backwards. The two don’t meet.

     So, come on, if your theory is good, how much of the 32.2C GHE in 1850 was due to CO2? Don’t tell me to look it up unless you can argue my Lacis point!

321. WebHubTelescope (Comment #110209)

     Hey, don’t blame me, WHT! I wasn’t the one who introduced Arrhenius into the discussion as some sort of justification of the cAGw theory! Speak to Phil Scadden…

322. Arfur – all science proceeds from models – the term in science means a systematic description of theory. The paper outlines exactly that model – and then uses it calculate the GHE. The spectral observations (measurements) are compared to that theory/model. Because the measurements so precisely match observation, R&C 1978 is the basis of modern radiative codes. The spectral measurements are the empirical result you are looking for. They match the model outputs from R&C. R&C does capture the complexities of the atmosphere from a radiative transfer perspective. Why do you persistently ignore those measurements. What about even the simple measurements from pyrgeometers? If you are denying the GHE, then explain why the surface of earth receives longwave radiation (measured) when the sun barely emits any?

     I have no issue with Lacis et al. The answer to attribution for present day green house in http://onlinelibrary.wiley.com/doi/10.1029/2010JD014287/abstract by same authors.

     Feedbacks? You are denying ice reflects radiation? Denying the Clausius-clapyron relationship? And you still dont get that 0.8 is roughly what is expected with feedbacks. If climate sensitivy is 3 degrees for doubling of CO2, then with current CO2 levels and ocean thermal uptake, we should be at around 34% of that number, roughly 1 degree. When you consider that 1 degree is temperature change from doubling CO2 only, you can claim 0.8 as evidence for feedbacks.

     Whether models have overestimated warming is a moot point but any error is in the calculation of feedback, not from calculating the direct thermal effect of CO2.

     By all means argue about the accuracy with which models capture the feedback processes, but arguing against the radiative properties of CO2 is laughable, “sky-dragon” country.

323. DeWitt: I really appreciate your views and insights. I won’t argue against your opinion as it is, for all I know, reflective of what is considered the standard nomenclature.

     However, if “Feedbacks are a result of temperature change.” as you say, then LGM CO2 and CH4 increases must be a forcings just like albedo because it causes temperature increases and creates feedbacks of it’s own.

     What about black carbon on snow? I think it’s the primary forcing that does not directly cause a temperature change. When the snow melts from the CB albedo, what is that? a secondary effect Feedback #1 Then temperatures are altered by changes in ocean currents and albedo from white ice to gray ocean, what is that? Feedbacks #2 and #3.

     Personally, I prefer to think of the forcing as the primary effect and the secondary, tertiary, etc etc effects on down the line that follow from the primary physical stimulus are feedbacks. For one, it’s easier to sort them out.

324. Re: Howard (Feb 17 19:01),

     However, if “Feedbacks are a result of temperature change.” as you say, then LGM CO2 and CH4 increases must be a forcings just like albedo because it causes temperature increases and creates feedbacks of it’s own.

     A forcing initiates a temperature change by creating a TOA radiative imbalance. A positive feedback increases the forcing and a negative feedback decreases the forcing as a result of the temperature change caused by the initial imbalance. Hence CO2 and CH4 can be forcings or a feedbacks depending on their sources. Increased water vapor should also cause the temperature to increase further, but since the residence time of water vapor in the atmosphere is so short, it can never be a global forcing. It can be a local forcing, however. Irrigation definitely alters the local climate.

     The IPCC concentration on CO2 is a problem caused by the UNFCCC’s focus on CO2. There are a number of other primary anthropogenic forcings like black carbon and land use/land cover changes that could well be as important. But they’re not on the table.

325. Re: Nic Lewis (Comment #110217)
     February 17th, 2013 at 4:23 pm
     Nic,
     You are raising several issues, which I suspect can be separated.

     First, is CCSM4 atypical? On one very important measure, it is well inside the pack, and that is in the degree of curvature it shows in its relationship between aggregate outgoing flux vs average surface temperature. If we use ” ε ” from Winton 2010 as a measure of this curvature, then CCSM4 would rank around number 5, and it is superseded by several of the more respected models.
     You raise the point that the Forster et al 2012 paper (http://www.atmos.washington.edu/~mzelinka/Forster_etal_subm.pdf) shows a pre-industrial TOA imbalance for CCSM4 which is not the worst, but on the high side. One reason for this, I believe, is the limited tuning protocol applied to CCSM4 (Gent 2011, Mauritsen et al 2012). I have not entirely ruled out the possibility that the curvilinear relationship in flux-temperature might in part be explained by flux leakage or other numerical problems, but if this is happening, then there is no simple correlation with pre-industrial flux imbalance. It is common practice to net off the preindustrial flux imbalance in any calculation of flux relationships, thus avoiding the most obvious error in the estimation of ECS. Nearly all of the models which show very little evidence of a pre-industrial TOA imbalance still show significant curvature in outgoing flux vs temperature, so I am inclined to look for the dominant explanation elsewhere.

     Where CCSM4 does look like a bit of an outlier is in its tropical feedback. As Armour acknowledges:
     “The cloud feedbacks in particular seem to be less positive in the tropics than in most models, leading to a net feedback that is more negative in the tropics than in the model average. ”
     You make the point:-

     I was aware that the Zelinka paper used the change in global mean rather than local temperature in computing net feedback. But I don’t think that makes a huge difference until one approaches 60N or 60S.

     I think you will find that it is the change in the high latitudes that makes the large difference. The models show a range of amplification in the Arctic of 2 to 6 times average global temperatures, with most models in the range 2-4 times, with lower-but-still-significant amplification projected for the Southern polar and subpolar regions. If these values are used to “correct” the Zelinka estimates of local feedback in the high latitudes, then Armour’s geometric explanation still fits the facts very well.

     Substituting a low magnitude negative feedback or small positive feedback in the tropics has very little effect on the argument, since the temperature change is rapid. In terms of the gradient of the net flux temperature relationship, this low magnitude tropical feedback is effectively averaged in with all of the other rapid response regions to define the gradient in the early part of the plot. The later part of the plot shows the “average” of the feedbacks from the slower responding regions, where temperatures are still changing. The plot still then displays a curvilinear relationship.

     I would like to be able to show this effect directly, with a specific example, but unfortunately I cannot do so without modifying my simple model to solve for the meridonial heat flux. In the simple model’s present guise, the tropical temperature will reach a value of F/lambda, which is clearly not tenable, if lambda is small negative or positive. In practice, in most GCMs, if you believe Zelinka, then the tropical temperature is constrained not by closing the TOA net flux difference, but by a balance being achieved with an increase in meridonial heat flux out of the tropics. (It is important to note that this is a deficiency in my simple illustrative model, which does not exist in Armour’s temperature-weighted feedback model. His basic model does not need to make assumptions about the temperature-time relationship.)

     In summary, it still looks to me like Armour’s conceptual model may be fundamentally correct.

     In answer to your last question: “So could it be that all the other models are wrong about poleward meridional heat flow increasing?” An honest answer is that I don’t know, but given the absence of the hotspot, plus the fact that I find this result counterintuitive – I would expect polar amplification to homogenise temperatures and reduce meridonial flux overall – then this seems perfectly possible.

326. Re: Paul_K (Feb 17 22:17),

     I would expect polar amplification to homogenise temperatures and reduce meridonial flux overall

     That’s not possible without a massive change in local albedo. High latitude warming should indeed decrease albedo there, but unless tropical albedo increases by a lot, the tropics will still absorb much more solar radiation than the high latitudes and emit less than it absorbs. Flattening the TOA emission versus latitude means the high latitudes will have a greater surplus of emission over absorption. Keeping the high latitudes warm will then always require more meridional flux rather than less.

327. Phil Scadden (Comment #110225)

     Phil,

     Your inability to think for yourself is understandable, as it happens a lot on this site. However, your inability to even try to understand my point is just rude.

     I ask you for non-modelled evidence. You fail to provide any. You then compound that failure by presenting yet another model-based ‘paper’ as if it were evidence. It is not. Did you not even read my argument against the Lacis’ 20% attribution for CO2? You now propose that I should accept Schmidt’s equally indefensible 20% attribution! How obtuse of you. Here is the same argument presented against your lauded Schmidt paper:
     20% of 33C (if you agree with 33 – or use your own) is 6.6C. In 1850, the GHE was 32.2 (as the enhanced GHE since 1850 is measured at appx 0.8C). 20% of 32.2 is 6.44C. So you – and Schmidt – are telling me that 280ppm CO2 in 1850 equated to 6.6C and that a 40% increase in this trace gas since 1850 has only led to an increase of 0.16C [6.6-6.44]).

     Why don’t you try to just think about it instead of regurgitating nonsense ‘papers’ written by people who seem unable to consider any other possibility?

     You keep talking about me ignoring radiation figures. This is a complete strawman argument. I do NOT ignore them and I don’t have much of a problem with them. However, radiation is not heat! When have I denied the GHE? I’ve said up-thread I am not a ‘skydragon’, although I can understand that you would want to apply a label in a pejorative sense because you can’t provide any proper evidence to support your assertions.

     And you still don’t get that you cannot state with any conviction how much of the ).8C rise has been due to CO2. All you are doing now is riding along the coat-tails of natural factors and claiming that “it’s all due to CO2”. It is a pathetic argument and about as far away from science as is possible to get.

     Feedbacks. I’ve hard it all before. Provide evidence. Typical warmist excuses to attempt to explain why the global temperature has not risen in accordance with the model projections. You say that all science proceeds from models. No, it proceeds from experiment and observation (evidence). Models are only as good as the information put in.

     The key to rational thought in this debate is to explain why the planet had a GHE of similar value pre-1850. You can either invented and unproven assertions such as “Oh it’s feedbacks…” or “Oh it’s thermal lag…”, or you can accept that the radiative properties of CO2 , whilst known, are insufficiently strong to make a significant difference to global temperature.

     Ocean thermal uptake? Nice one, sidestep the issue by inventing yet another smoke and mirror sideshow which you wouldn’t even be bothering with if the global temperature had increased like your theory predicted.

     “Oh, no, you see the trace gas in the atmosphere can significantly warm the deep oceans in less than 160 years without a corresponding rise in sea surface temperatures but this is the reason why the same trace gas hasn’t warmed the atmosphere like we said it would…”. Handwringing drivel.

     At least try to be objective!

328. Arfur – part of your problem is that you are not understanding the use of “model”. Just about everything you measure (maybe everything if getting all philosophical) is done so in context of model. Your attitude appears to be stop reading something the moment you see the word model and then miss the point completely. As to Lacis/Schmidt, you are ignoring thermal inertia, other forcing an log response. If you do the calculation wrong, then of course you get the wrong answer. GCM are what do the full calculation (and predict very close to what we have) but if like back-of-the envelope stuff, then http://www.skepticalscience.com/Earth-expected-global-warming.htm does it for you.

     As to radiation and heat, well if you irradiate an object it must warm to such temperature so that outgoing irradiated energy must match precisely that of incoming(conservation of energy). You verify this in a high school lab; you can verify on a local surface.

     So lets spell this out one more time. Our instruments (pyrgeometers, and satellites observatories) observe a change in LR radiation. Ie a measurement. A change by itself doesnt tell that tell you why. However, they also observe a change in the shape of the observed spectra. How and what accounts for these changes? Well you can measure the radiative properties of gases in a lab. R&C then tells you how to propagate the measurements through the vertical structure of the atmosphere to tell you what the strength and shape of the spectra should be if GHG theory was right. Lo and behold, the calculations exactly match the measurement.
     Thermal ocean uptake. Right, the ocean observing network of thermometers is smoke and mirrors is it? You can calculate the earth’s energy imbalance direct from change in OHC. Doesnt this count as measurement?

     So far you asserted:
     1/ models arbitrarily thermalise CO2 which as Lucia says makes no sense for any well understood definition of arbitrary.
     2/ the GHG theory predicts that earth should heated more than it has. (there is no published paper that asserts climate behaves in the way you imagine science says it does).
     3/ that climate sensitivity is an input rather than output.
     4/ and now, that climate science says “its all due to CO2” – the briefest perusal of the IPCC SPM would show that no such claim is made.

     With so many errors and misunderstandings, I would ask the question then of what basis, what facts, led you to decide that climate models (which you havent looked at) arbitrarily thermalize CO2. What information did you study to come to your conclusions? It seems to me that you decided climate science is wrong a priori, and that there is no empirical evidence a priori. If you are not prepared to look at evidence, then there is no point to further discussion.

329. Phil Scadden,
     ” If you are not prepared to look at evidence, then there is no point to further discussion.”
     .
     You will not be the first to come to a similar conclusion.

330. SteveF – thanks. Obviously newby here. About as productive as talking with Doug Cotton. Much more interesting discussions elsewhere on site with a lot more meaning.

331. Lucia moderates at this site with quite a light hand.

     I think most regular readers look at the author of a given comment and categorize:

     * A close read of this person’s writing is typically rewarding;
     * This person often has interesting things to say;
     * This commenter’s track record is uneven;
     * Caveat lector;
     * Next.

     Re: the tail end of that list, a couple of aphorisms come to mind, e.g. “You can’t reason someone out of something they weren’t reasoned into.”

332. Phil,

     Part of your problem is that you are unable to countenance the likelihood that your radiative theory is wrong in any significant sense. Are you now denying that the IPCC have asserted that the climate sensitivity is ‘very likely’ to be 3C? Are you denying also that the warming since 1850 is 0.8C? Please use your ‘log effect’ to prove to me that the expected warming is likely to be 3C without using thermal lag a an excuse. Because that is what it is. Let me be clear about this: ‘Thermal Lag’ is unproven, unlikely and totally discredited by observed data.

     Hence:

     http://www.climate4you.com/images/HadCRUT4%20GlobalMonthlyTempSince1979%20With37monthRunningAverage.gif

     for air temp, and

     http://www.climate4you.com/images/NODC%20GlobalOceanicHeatContent0-700mSince1979%20With37monthRunningAverage.gif

     For 700m ocean, and
     http://www.climate4you.com/images/HadSST3%20GlobalMonthlyTempSince1979%20With37monthRunningAverage.gif
     for sea surface temperatures.

     What do you think the lag is? 5 years? Air temp and sea surface peak in 1998, ocean peaks in 2003.
     Ok, a 5-year lag.
     So, according to you (and skepticalscience), the reason the planet hasn’t warmed since 1998 is because of the ‘thermal lag’ with the ‘missing’ temp being absorbed into the deep ocean. Unfortunately, for that assertion to be true, the rise in temperature would have had to have started 5 years later in the ocean data. Look at the graphs. See anything similar? All three show a trough at around 1985 (give or take a year). SO there was no lag when the ocean data started to show a warming. The air temp peaks in 1998, as does the sea temp. The ocean peaks 5 years later but all three show a flattening since at least 2003.
     So, if your thermal lag is the reason why the air temperature hasn’t warmed, why hasn’t the lag kicked in since 2003? Why is the starting date the same?
     Also, and this is equally important, how do you explain the warming of 1910-1945 (appx 0.7C) at a period when – according to your ‘radiation’ argument, the radiative effect of CO2 had not really started? It is almost exactly the same rise as the 1975-1998 warming. Your argument seems to be that the latter warming was due to ‘CO2 radiation’ but the CO2 radiation wasn’t there in 1910. The only logical conclusion is that the warming of 1910-1945 had to be due to ‘other factors’. This presents an unfortunate conundrum for the advocate of the CO2/radiative theory. It seems pretty clear from the IPCC graphs (which don’t match their words – but that’s another argument) that they think CO2 didn’t really start increasing until about 1950. So how do you explain the earlier warming? And, if your answer is ‘natural factors’, then how can you insist that the later warming was due to CO2? These are not rhetorical questions, by the way!
     To summarise, thermal lag is not supported by data. If you want to argue that the lag is greater than 5 years, then please provide your reasons.
     If you are going to base your argument on the skepticalscience comments, your credibility is going to reduce pretty quickly.

     Radiation. Ok, I agree with your comment (who couldn’t?) but it has little relevance in the cAGW debate. If it was as simple as that there would be no argument, would there? The global temperature would be increasing exactly as per your school lab measurement. Simple. Of course that is not the case. I keep telling you that your assumption that ALL of the observed warming is due to CO2 is completely unproven. If you disagree with me, then please provide any evidence that confirms just how much of the 0.8C warming since 1850 is due to CO2. You keep talking about measuring radiation and I say you need to talk about measuring heat. The atmosphere is not as simple as ‘an object’. If it was, there would be no discussion.
     1/I have explained that to Lucia before. The amount of warming inputted is based on the amount of radiation. The amounts used have almost universally been too high. This alone should give you cause for thought.
     2/Just about every climate model has overstated the amount of predicted warming. See several posts by Lucia on this subject.
     3/ See my post to Lucia. Yes, I should have used a different term such as warming component.
     4/”its all about CO2″. Where did I say that the IPCC said this? I was talking about you! That sort of comment is about as silly as steven mosher saying “CO2 has nothing to do with climate sensitivity”. I do, however, claim that ‘climate science’ has overstated the warming effect of CO2 and other nghgs. What I find most unpalatable about your line of argument (along with your sanctimony) is that you are completely unwilling to even countenance the possibility that you are wrong. Obviously, the production of real, physical, un-modelled evidence would support your argument.
     [“I would ask the question then of what basis, what facts, led you to decide that climate models (which you havent looked at) arbitrarily thermalize CO2. What information did you study to come to your conclusions?”]
     These facts:
     Global temperature rise since 1850 = 0.8C
     Warming period in 1910-1945 virtually identical to 1975-1998.
     CO2 rise mainly started to increase significantly after 1950
     No evidence or rational argument for thermal lag argument
     No evidence for feedback argument
     No concession by pro-cAGW commenters that the GHE before 1850 was only slightly cooler than today in spite of a very significant increase in CO2 and other nGHGS. (You keep avoiding this point while at the same time insisting that I don’t understand.)
     Every paper you have produced ‘as evidence’ has been based on models. I don’t care how much you value them, a model is only as good as the data and algorithms used to build it.
     And, finally, the fact that the known radiative properties of a single molecule of CO2 have been assumed to make a significant impact on the atmosphere as a whole. You may be right, but if you were as scientifically authoritative as you assert, you would realise that some credible evidence would be required as per the ‘scientific method’.
     .
     A word of advice, Phil. You seem like an intelligent bloke. Try not to fall into the ‘groupthink’ trap that several commenters have fallen into on this site. Several commenters here don’t like my argument but no-one (repeat no-one) has provided any real evidence to support their belief in the CO2=significant warming argument. Think for yourself. You don’t need them to give you a back-slap. It’s a gang thing…
     .
     Oh, and if you don’t like my figures for the 20% attribution, then please use a ‘log effect’ instead of a simple linear effect. See if it helps your case.

Comments are closed.