Some thoughts on climate sensitivity

Semyorka posted a comment on my previous post that highlighted a paper that I had’t seen before. The paper is Global atmospheric downward longwave radiation over land surface under all-sky conditions from 1973 to 2008 which tries to determine (as the title might suggest) the change in downwelling longwavelength flux, over land.

The abstract concludes with

We found that daily Ld increased at an average rate of 2.2 W m-2 per decade from 1973 to 2008. The rising trend results from increases in air temperature, atmospheric water vapor, and CO2 concentration.

Ld is the global atmospheric downward longwave radiation and the observed trend (2.2Wm-2) suggests it increased by 7.7Wm-2 between 1973 and 2008. Initially I was somewhat confused by this (still am maybe 🙂 ) as it seemed rather high, but over the same time interval, land surface temperature increased by almost 1K (see the Skeptical Science trend calculator). This would increase the outgoing surface flux by

dF = 4 \sigma T^3 dT = 5.5 Wm^{-2}.

So, the outgoing surface flux over land has increased by about 5.5Wm-2, while the downward longwave flux has increased by about 7.7Wm-2. If you consider a typical forcing dataset, then the radiative forcing has increased by maybe as much as 1.5Wm-2 since the mid-1970s. If you do a simple transient temperature response calculation, that would suggest that the transient response over land is

TCR = \dfrac{3.7 \Delta T}{\Delta F} = \dfrac{3.7 \times 1}{1.5} = 2.5 K,

which is somewhat higher than the expected global value of slightly below 2K (okay, maybe it should be 3.44, rather than 3.7, but that won’t change this all that much. Also, the change in forcing I’ve used is probably a bit too high anyway.). It’s possible that the system is just too complex for such a calculation to be reasonable, but given the low thermal inertia of the land – compared to the oceans – it’s not that surprising that the land-only TCR is greater than the global TCR.

However, the equilibrium response (with fast feedbacks only) shouldn’t depend on the thermal inertia (it will just take longer to reach if the thermal inertia is high, than if it is low). Therefore, if the above calculation has some merit, that the downward longwave flux over land exceeds the outgoing flux (as the paper mentioned above suggest) could suggest that the equilibrium response has to exceed 2.5K.

Admittedly, I’m ignoring uncertainties and all sorts of caveats. It’s also possible that such a calculation doesn’t really make any sense given the complexity of the system. That’s why I thought I would write this post – someone can point out where I’ve gone wrong and why this hasn’t been suggested before (assuming that it hasn’t). In my experience, when you notice something apparently simple that noone has noticed before, it’s probably because it’s not as simple as you initially thought 😀 .

Update: I knew I was going to do something silly in this post. As Chris Colose points out on Twitter, you need to close the surface energy budget using non-radiative terms too; like evaporation and convection. That the downwelling flux exceeds the upgoing flux doesn’t mean that the surface is out of energy balance. So, the latter part of this post is probably slightly nonsensical, or – rather – you can’t really use this to argue for an ECS above 2.5K.

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

110 Responses to Some thoughts on climate sensitivity

  1. Is there such a thing as “land-only TCR”? Shouldn’t you take the ocean and land surface temperature increase? The greenhouse effect is not something local and instantaneous.

  2. Lucifer says:

    Hello ATTP,

    This is brief.

    Review paragraph 1: http://www.gfdl.noaa.gov/bibliography/related_files/sm8001.pdf

  3. Shouldn’t you take the ocean and land surface temperature increase? The greenhouse effect is not something local and instantaneous.

    Well, yes, quite possibly. I was just surmising here. I actually just found the results in the paper interesting – that there’s been such a large increase in downwelling longwave flux. So, I was just extending that a little. I’m now remembering that we discussed on an earlier post why the land was warming faster than the global average, but I can’t remembered what was said.

  4. Lucifer,
    Well, yes, quite possibly. I was going to quote the paragraph, but all the formatting gets messed up and I can’t face retyping it. A good point.

  5. In fact, judging by the figure in this article, non-radiative effects make up about 30% of the upgoing energy flux. Hence that the increase in upgoing surface radiative flux is about 30% smaller than the increase in downwelling radiative flux, might suggest that the surface is roughly in energy balance (unless I’ve gone and made another silly mistake).

  6. CMIP5 GCMs showed that global Ld increased at a rate of 1.54 Wm-2 per decade (p<0.01) from 1979 to 2005.

    If the mitigation sceptics are allowed to say at every discrepancy that the models are running hot, without any need for further study or understanding, then surely I can now state: the models are running cold. 🙂

    Erring on the side of least drama? Or simply large measurement uncertainties?

  7. Victor,
    Interesting, so an observation that shows that it has increased faster than the models suggest. The observation is obviously correct, because all models are wrong 🙂

  8. Arthur Smith says:

    Ah, I see you updated based on Chris’s note – yes, that’s exactly what I was going to say also, an increase in downwelling flux goes not just to upwelling radiative flux, but also the upwelling latent heat and convective heat flows, so you expect that sort of difference. That’s why estimates involving surface radiative fluxes are very difficult to interpret; it is all much cleaner high up in the atmosphere where you have purely radiative effects.

  9. Arthur,

    Ah, I see you updated based on Chris’s note – yes, that’s exactly what I was going to say also, an increase in downwelling flux goes not just to upwelling radiative flux, but also the upwelling latent heat and convective heat flows, so you expect that sort of difference.

    Yes, I’m feeling particularly silly now 🙂 It seems obvious now, but it wasn’t when I was writing the post.


  10. The observation is obviously correct, because all models are wrong 🙂

    Actually everyone on the internet is wrong when it comes to the Box quote. They miss the context, which is numerical accuracy when approximating.

    From the Box and Draper book the quote came from, interesting to look at the original meaning:

  11. Mann (along with co-authors Steinman and Miller, “Atlantic and Pacific multidecadal oscillations and Northern Hemisphere temperatures”) is back to using regression to estimate the warming signal, as he is pulling out the variability due to AMO and PMO from the time-series.

    They just have to watch for the expected backlash that they are using a temperature to regress against a temperature.

    I don’t understand why they don’t use non-temperature factors to remove the variability. Good candidates are the ENSO SOI and delta LOD. These have excellent reversion-to-the-mean properties in that over the long-term these always revert to zero, meaning that they do not add to the trend. The result is that one can estimate an equivalent TCR of about 2.1C for the GISS temperature series.

  12. Robert Way says:

    My comment was that removing variability without updated forcings makes the post-2005 data appear to overestimate the PDO and underestimate the AMO signs. I was having a back and forth but then my comments started getting flagged as spam and I decided it wasn’t worth the work to search through it to see which words were triggering the filter.

  13. Robert,

    I was having a back and forth but then my comments started getting flagged as spam and I decided it wasn’t worth the work to search through it to see which words were triggering the filter.

    Here, or somewhere else?

  14. Robert Way says:

    There at realclimate. This was my response to Dr. Mann that got flagged even when I used small sections of it – I don’t really know what words trigger these things:

    [Response: In all fairness Robert, no, I’m afraid you’re missing the point. There is no amount of shifting of global mean temperature or interpolation of Arctic temperatures that is going to explain away…]

    I did not bring up coverage bias – in fact I believe it to be a small contributor not a major one, something that kevin and I have tried to make clear. I think you’re implying something when you make a comment like that which is no one’s argument. When Kevin started this work there was no intention of ‘explaining’ some pause or anything of the sort – in fact at the time we thought it was a reasonably obscure problem that wasn’t likely to garner much interest at the time of submission. By the time it was in press there had been a number of articles on a so-called pause which had been published in Nature and Science because it was a high profile issue by then. For us, we were more interested in robustness than attention and that is why we went directly to a specialist journal where we would be sure to get heavy scrutiny from those whose record we were using. Hence the four online supplements we have added in the time since (The last of which shows an anomalous cooling in GISTEMP’s Arctic data that explains much of why they’re cooler than our record). The idea that you believe we tried to ‘explain away’ anything is unfortunate.

    [Mike…we have seen anomalous increase in tropical Pacific trade wind strength, sustained tendency for La Nina-like conditions, and enhanced tropical Pacific ocean heat burial, over the past decade+. There is a healthy body of research, as you know (and as is cited in our article) showing that these factors have contributed to a slowing of global-mean warming over the past decade… so I’m rather perplexed that the point is being lost…]

    I’ve never made the argument that it doesn’t contribute to the reduced rate of warming. My point is that when you’re trying to directly quantify the magnitude and relative importance of the multidecadal variability to recent temperatures you need to have an accurate estimate of the role of external forcing over the past decade. Most if not all of the models being used in these studies would show lower temperatures over the past decade if they had updated forcings for volcanic activity and the weak solar cycle. This would result (with your methodology) in a more positive AMO and a less negative PDO that in would reduce the total role of multidecadal variability. This is seemingly straightforward.
    As I pointed out above – the updated forcings are relevant and a few of Santers paper show that – not to mention Schmidt et al (2014) which shows the updated forcings playing as large a role in the discrepancy as the phase of ENSO.

    Santer et al (2014)

    “…We show that climate model simulations without early 21st century volcanic forcing overestimate the tropospheric warming observed since 1998…”

    Ridley et al (2014)

    “…Finally, the SAOD above 15km most noticeably underestimates the total SAOD at high latitudes following an eruption, particularly a high-latitude eruption…. “

    Santer et al (2015)

    “Our findings show that the hiatus is not due to internal variability alone…Of particular interest is our positive detection of volcanic cooling signals in observed tropical SST data… Volcanic cooling signals are therefore aliased into the observed tropical SST changes specified by Kosaka and Xie [2013]… While prescribed SST simulations are useful for many purposes [see, e.g., Gates et al. , 1999], our study shows that they cannot reliably quantify the contributions of individual factors to the “warming hiatus.””

    You demonstrate in this contribution and the supplemental materials what appears to be a reasonable method for discriminating between forced and unforced variability and that itself deserved to be published in a specialist journal for covering the period where we have known forcings. The section post-2005 is where you have the least certainty in the forced response and external forcings and I just don’t see that reflected in the analysis, error bars and conclusions.

  15. Robert,
    Thanks, interesting. I was thinking of writing something about this whole internal variability thing, but maybe I should give it all a bit more thought before doing so.

  16. Meow says:

    I was thinking of writing something about this whole internal variability thing

    I’m interested, particularly in what constraints there are on internal variability. What does paleoclimate tell us? How about spatial differences between the expected effects of known forcings and spatial effects of internal variability?

  17. Meow says:

    Also I’d like to see an exploration of the relationship between climate sensitivity and the amplitude of internal variability. I’ve often heard it said that high/low climate sensitivity implies high/low internal variability, and vice versa. This makes intuitive sense, but I’d like to understand it more quantitatively.

  18. Meow,

    I’ve often heard it said that high/low climate sensitivity implies high/low internal variability, and vice versa. This makes intuitive sense, but I’d like to understand it more quantitatively.

    It’s probably too late for me to do justice to this, but the basic argument is that internal variability isn’t just about moving energy around in the system and changing temperatures, it’s also about the response to those internally driven temperature changes. The response will be changes to the atmospheric water vapour and to cloud cover. These are essentially the same physical processes that drive the feedbacks to changes in forcings. So, if internal variability is high, then it implies that the feedback response to a change in forcing must also be high – and vice versa.

    So, if the system doesn’t respond much to internally driven changes in termperature, it shouldn’t respond much to externally forced changes in temperature; or if it does respond strongly to internally driven changes, it should respond strongly to externally driven changes.

  19. Frank says:

    As you increase the amount of GHG’s in the atmosphere, several different factors increased the GHG flux. First, the number of GHG’s emitting DLR increases. Second, the temperature of those GHG’s has increased. Third, DLR only reaches the surface when it isn’t absorbed by GHG’s on the way to the surface. As GHG’s rise, the average photon reaching the surface is emitted from a lower altitude – where it is warmer.

    This online MODTRAN calculator may allow you to take all three factors into consideration:
    http://climatemodels.uchicago.edu/modtran/modtran.html

  20. An important observation to make is that the short term variability of AMO lines up with the global temperature variability. Use the WoodForTrees “isolate” function to experiment with it:

    http://www.woodfortrees.org/plot/esrl-amo/mean:12/from:1920/isolate:60/scale:1/plot/hadcrut4gl/mean:12/offset:0.6/from:1920/isolate:60/scale:1

    The next observation is that same variability in the AMO is a mix of the ENSO signal along with volcanic diturbances plus small fraction of TSI variability

    The last step is to remove the long term variability. I wouldn’t recommend doing it the way Steinman and Mann do it just for the fact that you will be accused of making circular regression errors. The solution is to find a proxy for the long-term variations that doesn’t represent temperature directly and has a reversion to zero property.

  21. Meow says:

    ATTP: Sure, but that explanation relies upon some similarity between drivers of change that are external to the climate system and drivers that are internal. Is that always so?

    Case 1, external: TSI increases. This causes increased ASR, which increases surface and tropospheric temperatures, causing increased atmospheric water vapor content due to Clausius-Clapeyron, causing yet higher temps due to increased atmospheric absorption/emission of IR. Got it.

    Case 2, internal: Something internal to the climate system changes (why?) Maybe it’s a little high cloud that by chance (why?) appears over Amazonia. How does this change get amplified into a noticeable climatic effect? Why doesn’t it just get washed out by the much larger diurnal/annual climatic excursions? Is significant internal variability real, or is it perhaps merely an effect of some forcing that we’ve not yet identified? If some form of “chaos” is the explanation, why does it seem that only a few butterflies out of vast flocks get amplified into noticeable climatic effects?

  22. Robert,
    Over at WUWT, Nic Lewis is saying to you that “I’m glad you realise what rubbish the Steinman paper is. “
    I hope he is not putting words in your mouth. I wouldn’t call the paper rubbish but not as airtight as it could be.

    Also look at what is happening with PDO right now, at the right hand of the graph.
    http://www.ncdc.noaa.gov/teleconnections/pdo/

    The PDO might be flipping, which means that the analysis could probably use an update.

  23. Meow,

    How does this change get amplified into a noticeable climatic effect? Why doesn’t it just get washed out by the much larger diurnal/annual climatic excursions? Is significant internal variability real,

    I suspect it is real, but I suspect it is more to do with variations in ocean currents which can have a significant effect on the temperatre distribution. There’s a Palmer & McNeall paper that looks at unforced climate simulation runs which shows that unforced variability can exceed 0.1oC per decade for periods of about a decade. The trend drops strongly with increasing time, though.

  24. Sou says:

    @Robert Way – the spam flag was probably your use of the word “specialist”, which dates back some years now, to the “cialis” pharma spam. (I think Hank Roberts mentioned this at realclimate.org, too.)

  25. WebHubTelescope said on February 27, 2015 at 9:26 pm:

    “Mann (along with co-authors Steinman and Miller, “Atlantic and Pacific multidecadal oscillations and Northern Hemisphere temperatures”) is back to using regression to estimate the warming signal, as he is pulling out the variability due to AMO and PMO from the time-series.

    They just have to watch for the expected backlash that they are using a temperature to regress against a temperature.”

    I think it may be important to say that they are using temperature along with one or more other variables to regress against temperature. See my further below for why. (It’s perfectly legal from a purely mathematical standpoint. And could this *mathematically legal* construction with the output having some part in the input be some part of the innovation in question that some recent researchers have implemented?)

    WebHubTelescope said on February 28, 2015 at 12:48 am:

    “The last step is to remove the long term variability. I wouldn’t recommend doing it the way Steinman and Mann do it just for the fact that you will be accused of making circular regression errors.”

    Would it be good to have mainstream climate scientists let those who reject mainstream climate science dictate what they the scientists do? (It seems to me to be a very bad idea to set such a precedent. I say damn the fake torpedoes, full steam ahead – especially on innovation.)

    Besides, they might not make such a claim again. When it was pointed out on Feb 18 how it was that those who made that “circular regression error” claim were the only ones making a purely mathematical mistake, explicit repeats of that claim went to almost zero almost immediately. There’s been essentially not a peep since in terms of explicitly repeating that claim, at least from those in question with the degrees in math or statistics or economics.

    See my comment on February 18, 2015 at 10:33 am under “Models don’t over estimate warming?” for details if interested. More simply: Given a composition of two functions, they were confusing the input of the outer function with the input of the inner function, where the input of the inner function is the input of the composite function formed from the composition. And (abstract) algebra tells us that on any function in any group we can always construct a composition of two functions such that the original function is the outer function and such that the inner function has an input with two or more argument variables that are allowed to vary as needed (this is a k-ary function for k > 1), where one of these argument variables in the input of this inner function is the output of the original and now outer function. And from that we can apply the substitution property of equality. So from a purely mathematical standpoint, those who make these claims don’t have a leg to stand on – they’re contradicting some basics of (abstract) algebra itself. (Perhaps this type of construction that is *perfectly legal from a purely mathematical standpoint* could actually be viewed as an innovation when applied as a regression method? I don’t know. Perhaps in some way this type of construction might be part of the innovation that these researchers speak of? I don’t know. But regardless I say again, “Damn the fake torpedoes, full steam ahead – especially on innovation.”)

  26. These are essentially the same physical processes that drive the feedbacks to changes in forcings. So, if internal variability is high, then it implies that the feedback response to a change in forcing must also be high – and vice versa.

    That may be true, but must is difficult to justify, when we notice that some of the changes may lead to negative feedbacks while some other have a positive effect.

    If the response were uniform over the whole globe then the conclusions might be more obvious, but internal variability means almost certainly that some parts of the ocean surface warm and some other parts cool, distribution of clouds changes as well.

    My understanding is that what’s known about changes in cloud cover has not provided support to the idea that observed internal variability would be to a major part due to variability in the effective forcing (defined as the overall influence of all changes in the atmosphere including changes in the cloud cover) but that such a possibility cannot be totally ruled out based on the existing knowledge. In that alternative oceans would play an essential role in the variability, but to a major part indirectly through the influence on the cloud cover and only to lesser part through the changes in the net heat flux into the oceans.

  27. Pekka,

    That may be true, but must is difficult to justify

    Fair enough.

    If the response were uniform over the whole globe then the conclusions might be more obvious, but internal variability means almost certainly that some parts of the ocean surface warm and some other parts cool, distribution of clouds changes as well.

    Yes, I agree. That’s what – I suspect – makes it a complicated issue.

  28. K&A, After looking at the Steinman paper some more with comments from Mann at the RealClimate site, I can see that they are using simulations to match the variability in AMO and PMO, and not necessarily the AMO or PMO data by itself. Mann has repeated that he doesn’t believe that traditional time-series analysis will help much in this case, which counters my original impression to what they are doing.

    I still don’t understand why they avoid using the ENSO SOI to match the intra-decadal natural variability. Put their effort into solving ENSO, which shows every indication that it can be modeled as opposed to the rat’s nest of ensemble outcomes that they average against. This becomes a time-series analysis again.

    The electrical analogy to me is like this. They have a fluctuating DC signal with what looks like hum embedded in it. An engineer would simply compensate the hum (which turns out to be a 60 Hz signal) revealing the remaining cleaned-up signal. The fact that in this case the hum signal is hard to characterize because it looks pseudo-periodic is less important than it appears.

    As long as one can isolate the source of the signal, in this case the equatorial Pacific’s ENSO process, means you go with that. And with step-wise refinement, one eliminates the other sources of modulation, such as volcanoes and what may be differential frictional heating or circulation changes from long term delta LOD changes (the JPL NASA argument).

    It always puzzles me why they don’t want to use the simplest approaches.

  29. izen says:

    @-Pekka Pirilä
    ” In that alternative oceans would play an essential role in the variability, but to a major part indirectly through the influence on the cloud cover and only to lesser part through the changes in the net heat flux into the oceans.”

    It may be a matter of memory.
    Internal variation is the emergent property of the same physical processes, negative and positive feedbacks included, that are invoked by AGW.
    Clouds can only be a response to other variation over short, 10-20 day timescales. There is no physical basis for the past state to affect future cloud conditions beyond this.
    Oceans on the other hand can change heat content distribution in ways that have an influence on the future for decades and possibly centuries.
    To generate internal variations over a specific timescale you need a physical sysytem that encodes change over similar timescales.
    So random cloud variation can lead to chaotic behavior on the short timescale, but the longer ocean changes AMO, PDO, which alter the cloud variation. Causation flows from the processes with the longest physical ‘memory’.

    @-WebHubTelescope
    “I still don’t understand why they avoid using the ENSO SOI to match the intra-decadal natural variability.”

    I think one argument against this is that the assumption that ENSO, AMO or PDO ‘hums’ embedded in the rising signal average to zero may be wrong.
    ENSO may be a thermodynamically neutral internal variation (despite Bob T’s claims) when all other things are equal, but may be skewed, or biased when other forcings are changing. So if the warming from CO2 has shifted the balance of the ENSO quasi-periodic cycles, or the AMO then removing the ENSO or AMO signal from the observed data may be removing some of the warming signal, or amplifying it.

  30. Eli Rabett says:

    It always puzzles me why they don’t want to use the simplest approaches.

    Usually because their advisers adviser tried it fifty years ago and it didn’t work. As Eli said

    Graduate training is designed to pass lore from advisers to students. You learn much about things that didn’t work and therefore were never published [hey Prof. I have a great idea!…Well actually son, we did that back in 06 and wasted two years on it], whose papers to trust, and which to be suspicious of [Hey Prof. here’s a great new paper!… Son, don’t trust that clown.] In short the kind of local knowledge that allows one to cut through the published literature thicket.

  31. izen said:


    I think one argument against this is that the assumption that ENSO, AMO or PDO ‘hums’ embedded in the rising signal average to zero may be wrong.

    Seriously, how can this be “wrong” ? If it looks like a duck and quacks like a duck, it is a duck. ENSO is not just El Ninos and La Ninas but is a continuum of quasi-periodic variability that directly contributes to the global temperature signal. Every one of the peaks and valleys in the ENSO SOI time series lines up with the same peaks and valleys in the GISS time series with a 5 to 6 month lag, save for those that are clearly the result of significant volcanic events.

    I already linked to the output of my multiple regression analysis in comment #11 above and you really have to tell me what is wrong with that chart. It sure looks like a winner to me..

    Yet, as you say, about the only factor that has some uncertainty is whether the ENSO signal is amplified over time. Yet that is second order stuff. Bob Tisdale essentially soils his pants when he sees this stuff I produce, so why not rub it in his face? The simple stuff is what you want to use to argue against these people. They tend to do graphical fits against astrological cycles and use very thick lines to give the impression that there is some sort of correlation.

    Why not go for the low-hanging fruit? I just do not get it.

  32. izen,
    That’s exactly, what I had in mid. Oceans have memory. The role of clouds is as an intermediary factor in the influence of the oceans on the energy balance of the surface and the Earth as a whole.

    I repeat that I do not present this alternative as a known or even very likely mechanism, I presented it only as a partial argument against the idea that strong internal variability must be connected to high climate sensitivity. Whether such a connection exists depends on further details of the Earth system, details that can be studied empirically by accurate enough satellite observations over long enough periods.

    GCM type models give their own answers to that question. The discussion that we had on the Marotzke and Forster paper is related to this as M&F used effective radiative forcings extracted from the GCMs as input, and one of the question raised is, how well they could extract the correct ERF based on the assumption that a constant α is applicable to the calculation of ERF from TOA imbalance over the periods being considered. This is an issue M&F acknowledged as an uncertainty in their approach. A further and more important question is, whether the present GCMs can describe correctly the influence of the state of the oceans on the TOA imbalance.


  33. we did that back in 06 and wasted two years on it

    Eli, I don’t get your point. I tried it and it worked.

  34. I don’t think that ENSO, which is a result of geophysical processes, has any relation to climate sensitivity.

    Volcanic activity is another story, as that obviously tells us much about how forcing relates to temperature changes.

  35. BBD says:

    Pekka

    I repeat that I do not present this alternative as a known or even very likely mechanism, I presented it only as a partial argument against the idea that strong internal variability *must* be connected to high climate sensitivity. Whether such a connection exists depends on further details of the Earth system, details that can be studied empirically by accurate enough satellite observations over long enough periods.

    I’m missing something here. My understanding is that the range of internal variability is constrained by the rate at which energy leaves the climate system. This is fundamentally determined by the radiative properties of the atmosphere. Sustained change in GAT forced by internal variability *require* that the rate of energy loss to space is inhibited, which would inevitably result in a climate system at least moderately sensitive to radiative perturbation.

  36. David Blake says:

    @WHT,

    “It always puzzles me why they don’t want to use the simplest approaches.”

    Amen to that. LOD has some nice correlations, but then one get’s into the mysteries of magnitism, and even more correlations therin…

    On correlations: clearly the AMO is more correlated with temperature than the PDO or ENSO? (Both in the NH and SH).

    See:
    http://woodfortrees.org/graph/hadcrut3nh/mean:12/plot/esrl-amo/mean:12/plot/hadcrut3sh/mean:12

  37. BBD,

    A different state of oceans leads to a different circulation pattern in the atmosphere. that affects the properties of the cloud cover, which in turn affects both albedo and outgoing LWIR at TOA.

    All those effects are surely true, the question is only, what is the magnitude of the net effect and the relative sign of that as compared to the variability in net energy flux into the ocean. There are no general principles that can answer these quantitative questions, they must be studied using data from observations and with models known to have sufficient skill in describing these phenomena.

    I interpret the word must used in this connection to mean that there’s a strong physical principle that tells the answer. What I have tried to say is that no physical principle by itself can tell that answer, therefore must is not the right word to use.

  38. Short and medium term variability in LOD tells mainly about the angular momentum of the atmospheric circulation. The more angular momentum the atmosphere has, the slower does the Earth rotate to conserve the total of the whole Earth system including the atmosphere.

    My guess is that the most important contribution to that comes from the jet streams. Thus LOD would tell about the strength of the jet streams, which is surely correlated in some way with other climatic phenomena.


  39. On correlations: clearly the AMO is more correlated with temperature than the PDO or ENSO? (Both in the NH and SH).

    AMO is a temperature measurement so it obviously correlates with temperature. Consider that it reveals the effects of volcanic disturbances.

    In contrast the ENSO SOI does not reveal any impact from volcanoes.

    So ENSO+Volcanoes+dLOD is a good way to go as it provides orthogonal factors that can be composited to reconstruct temperature variability.

  40. Meow says:

    My understanding is that the range of internal variability is constrained by the rate at which energy leaves the climate system.

    That would be a very interesting result. Do you have a citation to a paper discussing it?

  41. The LOD correlation, specifically anti-correlation as it is -dLOD ~ dT, to decadal and longer temperature trends has been known for a long-time:

    Lambeck, K., and A. Cazenave, 1976: Long-term variations in the length of day and climatic change. Geophys. J. Roy. Astron. Soc., 46, 555-573.

    This paper is not behind a pay-wall.

  42. There are two time scales to the LOD data, the shorter scale is definitely AAM.

    The longer time scale for LOD has a few interpretations as to what it entails.

    I use the AAM alongside the ENSO SOI factor to improve the model fit. AAM is wind and SOI is pressure, the link is that wind is caused by pressure differentials.. This would be problematic as the two may be considered degenerate measures but there is enough of a difference that this works out perfectly fine.

  43. BBD says:

    Pekka

    A different state of oceans leads to a different circulation pattern in the atmosphere. that affects the properties of the cloud cover, which in turn affects both albedo and outgoing LWIR at TOA.

    These changes in ocean circulation can be triggered by orbital dynamics (TSI) and internally-forced change such as sustained volcanism (LIP) and ocean gateway opening/closure. But in every case, they are feedbacks, not forcings. Tail, not dog.

  44. JCH says:

    The PDO and mid-century cooling, try to get the AMO to do this.

  45. BBD,
    Yes, they are feedbacks, but they do not necessarily correspond to a equally strong net feedback, when the forcing is spatially more uniform. Thinking that feedbacks are characterized even very roughly by a single number may be highly misleading. Locally strong feedbacks may in some case add up and in another case largely cancel in their contribution to GMST.

  46. BBD says:

    Surely palaeoclimate and observed climate behaviour points to the dominance of whole-atmosphere average radiative properties (net feedback) over regional (eg. cloud) effects?

  47. -1=e^ipi says:

    @ WebHubbleTelescope –

    A problem with your CSALT approach is that it doesn’t allow for a delayed response of climate to changes in forcing. You need to determine a climate response impulse function. A simple exponential decay impulse response function does not work because the Earth has many responses with their own response times. However, the approach by Van Hateren 2012 gives a way to get a good approximation of the true impulse response function.

    I have made some progress the past few days on trying to explaining temperature changes since 1876 as a combination of an impulse response function to changes in forcings (greenhouse gasses, solar irradiance and volcanic aerosols) as well as to natural climate variability (currently using 6 year lagged LOD, SOI, PDO and NAO as internal variability indices). My latest estimate is that the 95% confidence interval for equilibrium climate sensitivity is (2.50 +/- 1.67) C. However, the method I am using to get the confidence interval is invalid and probably overstates my uncertainty since my regression is highly non-linear.

    Does anyone have any tips for properly estimating confidence intervals for parameters in non-linear regressions (beyond a first order approximation)? Specifically if I am trying to estimate a confidence interval for A + B + C + D + E and my regression equation is:

    Y = AX1 + BX2 + CX3 + DX4 + EX5 + AFX6 + BFX7 + CFX8 + DFX9 + EFX10 + AGX11 + BGX12 + CGX13 + DGX14 + EGX15 + HX16 + IX17 + JX18 + KX19 + error, where the X’s are the independant variables and the Y is the dependant variable.

  48. harrytwinotter says:

    I thought the trend of around 2.2 W/m2 per decade downward longwave flux seemed high, but then I realised most of the excess appears to be subducted into the ocean. So a corresponding surface temperature increase has not been seen yet?

    It doesn’t make me feel any better; the downward longwave flux goes 24×7, the trend is increasing, and it would be wishful thinking the oceans will keep subducting the excess at the rate they currently are.

  49. izen says:

    @-Meow
    “… Do you have a citation to a paper discussing… that the range of internal variability is constrained by the rate at which energy leaves the climate system.”

    I don’t know if there is a paper discussing this, I suspect it is a logical deduction from thermodynamics.

    An example; a few years ago the new ARGO buoys returned data that when combined with the old XBT record APPEARED to show a big cooling in the oceans over just a couple of years. (Lyman et al 2006).
    Very few people thought the cooling was real because the rate of cooling that was apparently in the data exceeded the rate any known physical process could remove that much energy from the oceans. Crucially the rate of energy loss exceeded the rate of energy emission for a body at that temperature. The energy loss implied by the buoy measurements exceeded the Planck limit for emissions. It could be possible to remove energy from the oceans by evaporation as well as radiation, but that would require a physical mechanism that vastly increased evaporation without any increase in temperature and without any evidence of a big jump in absolute humidity.

    It was the unphysical nature of the measured cooling that prompted a close investigation of the accuracy of the data rather than accepting it as a ‘Natural variation’. That holds for all natural and anthropogenic changes in the climate, they are constrained by the laws of thermodynamics and the available physical process involved.
    Because both natural and anthropogenic effects are emergent properties of the same underlying physical processes the natural variation gives an indication at least for the constraints, or lack of them, on climate change. In other words the sensitivity of the climate.

  50. izen says:

    @-WebHubTelescope Re; the non-zero average of ENSO AMO PDO –
    “Seriously, how can this be “wrong” ? If it looks like a duck and quacks like a duck, it is a duck. ENSO is not just El Ninos and La Ninas but is a continuum of quasi-periodic variability that directly contributes to the global temperature signal. ”

    I think the argument is that the temperature can also directly contribute to the ENSO quasi-periodic variability.
    A comparison would be with the dripping tap.
    The drip rate and drip size are chaotic, a quasi-periodic process with a limited envelope of variation of time between drips and volume of drip.
    Change the driving input, the water pressure, and the envelope may not change much, but the distribution of drip size and time may alter, perhaps to smaller drips, but shorter time gaps, or larger drips with longer time gaps, but accommodating the increased input.

    This may impose non-linearities on the relationship between pressure and flow rate from the chaotically dripping spigot.

  51. David Blake says:

    @WHT,

    “There are two time scales to the LOD data, the shorter scale is definitely AAM. ”

    If I understand it correctly the shorter time scale is mostly controlled by distribution of mass of the atmosphere & oceans, while the longer scale is due to movements of the liquid magnetic core?

    On that last point, here is an interesting graph of how far the magnetic north pole has moved over the last 150years, and – just for fun- I’ve put Hadcrut 4 alongside 🙂

  52. izen

    Physical arguments provide constraints, but constraints derived from basics physics alone tend to be too weak to be of much help. That’s true also for the case of apparent ocean cooling that you mention. Your statement that the cooling rate exceeded the Planck limit for emission is off by a very big factor. The rate of cooling was roughly 1 W/m^2, while the total emission from the surface is roughly 400 times stronger.

    It may still be true that the observed rate of cooling was highly unlikely and that it was difficult to present physical mechanisms to explain quantitatively the rate of cooling, but basic principles of physics alone cannot provide relevant limits, more complex and detailed models are needed for that.

  53. Harry,

    I thought the trend of around 2.2 W/m2 per decade downward longwave flux seemed high, but then I realised most of the excess appears to be subducted into the ocean. So a corresponding surface temperature increase has not been seen yet?

    I think there has beena surface temperature response and I don’t think the oceans are relevant here. This is land-only and the land-only surface temperatures have risen by 1K, which would produce a 5.5Wm-2 increase in surface flux. Also, about 30% of the outgoing surface energy flux is non-radiative (convection, evaporation) so a 7.7Wm-2 increase in downgoing flux is probably balanced by an increase of 5.5Wm-2 of upgoing radiative flux and another 2Wm-2 in non-radiative energy fluxes.

  54. izen says:

    @-Pekka Pirilä
    “The rate of cooling was roughly 1 W/m^2, while the total emission from the surface is roughly 400 times stronger.
    It may still be true that the observed rate of cooling was highly unlikely and that it was difficult to present physical mechanisms to explain quantitatively the rate of cooling,”

    I understood the ‘highly unlikely’ element was the difficulty in explaining a 1W/m2 increase in emissions when surface temperature did not show a proportional increase. Even though the increase in emissions was a small proportion of the total and E == T^4.
    Increased rates of cooling from a cooling object is the qualitatively unphysical problem, which is why I seem to remember other phase change options being explored.
    Agreed that to be definitive you have to get into the devil of the detail, especially for quantitative results.

  55. Marco says:

    David:
    “On that last point, here is an interesting graph of how far the magnetic north pole has moved over the last 150years, and – just for fun- I’ve put Hadcrut 4 alongside”

    You will then enjoy this paper:
    http://dx.doi.org/10.6000/1927-5129.2012.08.01.13
    In which the geomagnetic North pole is confused with the geological North pole, and as a result all their calculations using the obliquity of the earth are complete nonsense.

    I am surprised no US congressman has cited this paper in support of his denial.

  56. Marco,
    That link doesn’t appear to work and I can’t work out the correct one.


  57. -1=e^ipi says:
    @ WebHubbleTelescope –
    A problem with your CSALT approach is that it doesn’t allow for a delayed response of climate to changes in forcing. ”

    Imaginary Number guy, I think you have deeper problems of your own to clear up before I am going to discuss this further with you. You are claiming to have done a multiple regression analysis, yet you don’t have a blog or any graphical artifacts to provide evidence that you have actually accomplished anything.

    I asked you this before and you said:

    “@ WebHubbleTelescope – Thanks for the advice. I am aware that what I write is a bit sloppy and could use visuals, sorry about that. I might get a blog / youtube channel or something in the distant future when I have time and have a better understanding of things. Now I am on the verge of homelessness so I have other priorities.”

    Does that explain your predicament, or maybe it is this:


    Sorry, mean to say single not since. I might need to see a neurologist about these spelling mistakes, it might be due to my brain injury. :/

    You might want to come clean on what your situation is because I really have no idea where you are going with this and you seem to be all over the map. [Mod : redacted]

  58. In reply to my comment on February 28, 2015 at 11:52 am, WebHubTelescope said on February 28, 2015 at 2:46 pm”

    “K&A, After looking at the Steinman paper some more with comments from Mann at the RealClimate site, I can see that they are using simulations to match the variability in AMO and PMO, and not necessarily the AMO or PMO data by itself.
    ……
    I still don’t understand why they avoid using the ENSO SOI to match the intra-decadal natural variability.
    ……
    It always puzzles me why they don’t want to use the simplest approaches.”

    This is just a guess, but it may have something to do with the usually good idea to keep pushing the envelope to generalize things more and more. (This drive to generalize of course characterizes much of what mathematicians and scientists do.) Since a general result of course has the advantage of giving a readymade answer to any particular thing that could be found that is covered by the generalization, this approach they use might be better able to give a readymade global answer in some way to the ever increasing talk of multidecadal internal variability and the ever increasing set of results and commentary by researchers with respect to oceans other than the Pacific, these being most importantly recent results involving the Atlantic and Southern Oceans, and also future findings including any possible future confirmation of the mechanism proposed by Tung and Chen. (By the way, wasn’t Mann the first one to use “AMO”?) (Again, this is all just a guess as to why they use the more complex approach.)

    Anyway, here is what I had in mind when I spoke of recent results and commentary by researchers involving the Atlantic and Southern Oceans:

    In addition to the Tung and Chen paper and those two studies finding more heat than expected in the Atlantic and Southern Oceans, this paper covered the Atlantic, Southern, Indian Oceans as well as the Pacific:

    “”Surface warming hiatus caused by increased heat uptake across multiple ocean basins”
    http://onlinelibrary.wiley.com/doi/10.1002/2014GL061456/abstract

    Quote (this is the abstract):

    “The first decade of the 21st century was characterized by a hiatus in global surface warming. Using ocean model hindcasts and reanalyses we show that heat uptake between the 1990s and 2000s increased by [0.7+0.3Wm^{-2}]. Approximately 30% of the increase is associated with colder sea surface temperatures in the eastern Pacific. Other basins contribute via reduced heat loss to the atmosphere, in particular, the Southern and subtropical Indian Oceans (30%) and the subpolar North Atlantic (40%). A different mechanism is important at longer timescales (1960s-present) over which the Southern Annular Mode trended upward. In this period, increased ocean heat uptake has largely arisen from reduced heat loss associated with reduced winds over the Agulhas Return Current and southward displacement of Southern Ocean westerlies.”

    “Heat uptake by several oceans derives pause says study: Major new research explains how increased heat retention by a number of oceans has driven the Pacific Ocean to maintain the so called pause in global warming.”
    http://www.reportingclimatescience.com/news-stories/article/heat-uptake-by-several-oceans-drives-pause-says-study.html

    Quote (including a number of comments by the lead author):

    “”This is a big debate,” lead researcher professor Sybren Drijfhout of the University of Southampton told reportingclimatescience.com. “The Pacific is not doing this alone,” he continued. “Many climate scientists are more or less convinced that the hiatus is due to natural variability. We see cold air temperatures over the Pacific but is this the only driver or does the Pacific need other ocean basins?”
    ……
    “We have the hiatus and we can explain it through reduced heat loss from the oceans which is manifested as reduced atmospheric temperatures in the Pacific but we needed a driver outside of the Pacific that allows the Pacific to maintain this atmospheric cooling and that is what we have found,” Drijfhout explained to reportingclimatescience.com.
    ……
    The team…….includes scientists from the University of Southampton, the UK National Oceanography Centre and the European Centre for Medium Range Weather Forecasts……..
    ……
    The researchers also identified a second mechanism that operates over longer time-scales. This has been associated with reduced winds over the Agulhas Return Current in the Indian Ocean and an upward drift in the Southern Annular Mode (SAM), also known as the Antarctic Oscillation (AAO).
    ……
    “This is a longer term separate mechanism that is growing with time,” Drijfhout explained. “It is a long term upward trend that becomes stronger and stronger every decade. We think it is being caused by the ozone hole interacting with the upper atmosphere and with the Southern Ocean so that the westerlies are shifting southwards and increasing the pumping of cold water from below to the surface,” he said.
    ……
    The research team did not use conventional climate models for this study. They used state of the art stand-alone ocean-only models developed by the UK National Oceanography Centre, according to Drijfhout. These models were fed with historic (re-analysed) data on the air temperatures and wind directions over the oceans which provided the inputs to drive the model to calculate changes in the ocean heat flux – the net amount of heat the oceans absorb and give up.”

    Wow. Interesting, in my view.

    Finally, here is some commentary by many researchers on the Tung and Chen paper – I quote two of them, and note that one of them is Forster, who as we all know used a method with the perfectly legal mathematical feature in question to do a more general or global study:

    “expert reaction to new research on ocean heat sinks and global average temperatures”
    http://www.sciencemediacentre.org/expert-reaction-to-new-research-on-ocean-heat-sinks-and-global-average-temperatures/

    Quote (I note especially the one on “oceanographic sense”):

    Prof Andrew Watson FRS, Royal Society Research Professor at the University of Exeter, said:

    “It will be very interesting to see whether their finding that during the last decade the heat has penetrated to depth mostly in the Southern and Atlantic Oceans stands up. It does make oceanographic sense however, because we know these are the major sites for deeper water formation – water from the surface Pacific does not penetrate nearly so deeply into the ocean.”

    Prof Piers Forster, Professor of Climate Change at the University of Leeds, said:

    “The hiatus has a really interesting morphology. This paper suggests that heat disappearing into the depths of the Atlantic and Southern oceans are the dominant cause. Their ideas seem fine but I’m also convinced there is more going on: the El Nino and relative cooler European and Asian winters remain important aspects to understand.
    ……..
    Most importantly, this paper is another nail in the coffin of the idea that the hiatus is evidence that our projections of long term climate change need revising down. Variability in the ocean will not affect long-term climate trends but may mean we have a period of accelerated warming to look forward to.”

  59. izen
    This is a wavelet scalogram of SOI I produced recently:

    The x-axis is months after 1880. In trying to model SOI as a stationary yet quasi-periodic process. I did find a clear indication that something changed at around 1980 (the 1000 month mark), which broke the stationarity of the model. You can see this in the wavelet scalogram.

    There is evidence for a Pacific climate shift around 1976-1977 that Trenberth has alluded to.

    Having said that, I am still trying to understand how non-linearities play into this. I am taking the tact that ENSO is what it is. I am not trying to predict anything with it, because ENSO is difficult to project into the future.

  60. The way I see WHT’s results is that he has searched for a model that correlates a couple of time series that represents external forcings and a couple of time series that describe the state of the Earth system. One of the variables is GMST, the others are picked out of a number of available variables as those that result in the best fit, when the total number of variables is kept low.

    That he finds a good fit with a small number of variables is of some interest, but it’s very difficult to say, what we should conclude of that. First of all we do not have proper statistical tests to tell, how significant the close agreement is. The second problem in interpreting the results is the generic problem of curve fitting: Finding correlations does not tell, what are the causal relationships. Is one of the variables causally caused by the phenomenon described by another, or vice versa, or perhaps both are caused by a further factor not directly described by the variables.

    El Niño / La Nina does certainly affect GMST, but what are all the factors behind El Niño / La Nina.

    LOD is related to the state of the atmosphere. Thus it’s natural that it’s correlated with GMST as well, but I cannot see that LOD itself would cause anything, the variations in it are too small. Thus LOD tells about other things, one of those other things is the state of the atmosphere, but there may be also something else that affects climate.

    Finding close correlations as WHT has found may lead to further ideas about the actual mechanisms, but it may also turn out that such findings lead nowhere further as has happened to so many other apparent successes in curve fitting. The variables of WHT are such that they might really tell about something significant (as opposed to some of the other attempts to explain climate by such orbital parameters that are virtually certain to tell almost nothing significant for the state of the Earth system).

    I’m not surprised that WHT’s findings have not raised much interest. Their significance is not likely enough for that, but it’s not excluded that they tell something worthwhile.

  61. K&A,
    I do see your point about generality, but my idea is that we are trying to explain specifically the variability in the GISS time series. That is what people see. So my goal is to get the primary explanatory factors for GISS which doesn’t generalize to other areas such as OHC, for example.

    These oceanic dipoles such as SOI from ENSO and the Southern Annular Mode from the Antarctic Oscillation are simply quasi-periodic exposures of heat and absence of heat (cold) to the atmosphere, as the sloshing of the ocean’s volume and thermocline bring heat to the surface or lower it.

    If I could collect all the major oceanic dipoles going back to 1880 and feed that into the analysis, I would. But as far as I can tell, only the ENSO dipole is revealing as a major factor.


  62. but I cannot see that LOD itself would cause anything, the variations in it are too small

    Common mistake. LOD doesn’t by itself cause anything but it is a measure of some other variability that is revealed by LOD. That is the nature of a proxy measurement.

    Pekka, I noticed that you just made this comment at Nick Stoke blog:


    Pekka Pirilä, March 1, 2015 at 6:39 AM

    My impression is that the discussion does not have any clear direction, because the neither the parameter being estimated not the noise model have been specified accurately enough to allow for unique answers.

    I can go further and wonder why Nick is trying to remove this “noise” at all with his filtering study. I really don’t consider any of this noise. Clearly the ENSO variations are providing the signal variability and people ought to just own up to it. Why filter when we already have a model for the “noise”?

  63. Marco says:

    ATTP, I don’t know why that doesn’t work – it is the doi the journal (a Predatory Publisher, but still) gives.

    Anyway, go here:
    http://www.lifescienceglobal.com/independent-journals/journal-of-basic-and-applied-sciences/volume-8-number-1#discoveryof

  64. Marco,
    Wow, that is a rather embarassing mistake.

    The motion of the North Pole, deduced from the geomagnetic polar shift data, is highly correlated with major earthquakes. This is an indication that the frequent occurrence of major earthquakes had increased earth’s obliquity and induced global warming and possibly emission of greenhouse gases. It was shown by a simple model developed here that seismic-induced oceanic force could enhance the obliquity leading to increased solar radiative flux on earth.

  65. Marco, That is very illustrative of the “battle” we are up against.


    The analysis showed that obliquity change due to North Pole shift and total solar irradiance accounted for 63.5% and 36.4% respectively, while CO2 changes accounted for 0.1% of the observed global warming.

    Part of the reason that I am pushing these simpler models is to provide ammo that the average person or the layman can better appreciate.

    So when a model like CSALT shows that log(CO2) provides over 90% correlation of the time series starting from 1880, and then increases to above 98% when SOI, dLOD, volcanic aerosols, and dTSI are added, certainly that is good ammo to counter these ridiculous assertions that CO2 is only accounting for 0.1% of the global warming.

    Consider also that I see no pushback to the assertion that aCO2 has only provided a forcing post 1950-1960, with no evidence that it had impact before that. That really seems nonsensical considering the information that we have available.


  66. but I cannot see that LOD itself would cause anything, the variations in it are too small

    Common mistake. LOD doesn’t by itself cause anything but it is a measure of some other variability that is revealed by LOD. That is the nature of a proxy measurement.

    On this, at least, WHT seems to be repeating exactly what I wrote, although the above could be interpreted to imply something else.

  67. Pingback: Some thoughts on internal variability | …and Then There's Physics

  68. Here is the chain of correlation.

    1. Clearly ENSO has an impact on climate variability — El Ninos, La Ninas and all that, with the kicker that SOI matches GISS fuctuations as I show in the chart.
    2. Wind speed and atmospheric circulation variations as manifested by atmospheric angular momentum (AAM) changes track ENSO SOI as well. This is a wind / pressure relationship.
    3. AAM can be sensitively measured by keeping track of LOD. This shows up very clearly in the LOD fine structure. This means that LOD can be used as a temperature proxy instead of ENSO SOI.
    4. A longer-term multidecadal structure also appears in the LOD.

    From observations 1,2,3, one could conclude from 4 that the longer-term LOD variations would also have an impact on the earth’s temperature. The converse doesn’t make as much sense. Otherwise one would have to explain why fine structure in the LOD associates with global temperature changes but not the long-term structure. Changes in kinetic energy, which is what LOD measures, always have an association with temperature changes in a free energy formulation.

  69. Kevin ONeill says:

    Pekka – That’s not how I read WHT’s response. LOD is the second largest Pareto component in the CSALT model.

    I think what WHT is telling you is that contrary to “… I cannot see that LOD itself would cause anything, the variations in it are too small…” even small variations in a proxy can be indicators of significant changes in the parameter of interest. The analysis shows that for the GISS record, LOD is a more significant factor than anything not named CO2.

  70. I am still on my kick of exploring simple models and using first-order approximations to understand climate change.

    Consider that thermodynamics is a simple model of physics. No one seems to question basic thermo. Yet, CSALT also started out as a free energy formulation of potentially significant thermodynamic climate parameters (my first post on CSALT, scroll down to Foundation of the CSALT model).

    Should this be questioned as to its applicability? Ponder that.

  71. -1=e^ipi says:

    @WebHubbleTelescope –

    “Imaginary Number guy, I think you have deeper problems of your own to clear up before I am going to discuss this further with you. You might want to come clean on what your situation is because I really have no idea where you are going with this and you seem to be all over the map.”

    Okay. I got the message. I apologize for wasting your time. I’ll try not bother you in the future. I wish you luck in your endeavors.

  72. Kevin,

    Read again, what I wrote in my earlier comment:

    I wrote “LOD tells about other things”, WHT wrote “.. it is a measure of some other variability”. What’s the difference?

  73. Marco says:

    WHT, as I noted I am surprised no US congressman cited this paper. Perhaps it was too obviously stupid in its mistake.

  74. BBD says:

    Pekka

    Thanks for the explanations yesterday. For some reason (lateness of the hour; alcohol) I managed not to understand what you were saying at all at the time. Apologies.

  75. Arthur Smith says:

    -1 – I wanted to thank you for the reference to van Hateren’s paper. I’ve been looking into “impulse response”-type formulations for a while, and I’ve actually been working on a blog post for several weeks on how they work and why they are probably a useful technique; also how they relate to energy balance etc. I hope to get it up shortly but it’s been slow going. My original prompting to look into this was to try to figure out how one would properly fix Monckton & friends’ “irreducibly simple” model to actually be a reasonable approach. A lot of people in climate seem to have done something along these lines but I haven’t found any good summaries or reviews. So it’s nice to have at least another citation on the list.

    For some input on regression issues you might want to post a question on the “Open Mind” blog at https://tamino.wordpress.com

  76. -1=e^ipi says:

    @ Arthur – You are welcome. I appreciate the links you have provided as well.

    I’ve made quite a bit of progress in terms of constructing regression models to test different impulse functions during my free time. Though I’m starting to get a bit more skeptical of the Van Hateren approach. I’m finding that the resulting ECS can be fairly sensitive to the choice of decay times, even with a large number of decay times; so whatever you choose will be subject to specification error. Dealing with all the non-linearities is difficult as well. Also, you must ensure that all of your coefficients for your exponentials are positive, otherwise you get a non-sense impulse response function.

    The ‘best’ estimate I have been able to come up with for the ECS so far using the time series approach is a 95% confidence interval of [0.69,4.26] C with a central estimate of 1.71 C. And by best I mean has the least specification error and has proper evaluation of uncertainty (although it still has some specification error so the true 95% confidence interval should be larger than this). Right now my residual is very correlated with the multivariate ENSO index and the Atlantic multidecadal oscillation index. If I figure out a good way to account for these (while dealing with the obvious issue of reverse causality) then I should be able to reduce my unexplained variance by over half.

    I’ve spent some time looking through the literature on impulse response functions. It seems that Fortunat Joos is one of the more cited people who have looked at this. One thing I tried was I took the impulse response function from one of Joos’ papers http://www.climate.unibe.ch/~joos/IRF_Intercomparison/Protocol_CO2_impulse_response_modelcomparison_v1.0.pdf and tried to see what kind of functional forms could fit to the impulse response function reasonably well.

    A simple exponential response function (1 – exp(-A*t)) does not do very well. However, (1 – exp(-A*t^B)) and (1 – exp(-(A-B*exp(-Ct))t)) are fairly good.

  77. -1=e^ipi says:

    I’m starting to get the impression that there is a discrepancy between the empirical data and most computer models. Using time series analysis, I can’t seem to get high estimates of climate sensitivity no matter what I change. Most of my estimates for ECS fall in the low to mid 2’s, which seems to agree with most of the other literature on the topic (example: Van Hateren). But if the GCMs are consistently getting climate sensitivities over 3C, with best estimates of say 3.2 C, then that does make one question whether some of the assumptions used in the GCMs are valid and if there is a selectional bias of parameters.

  78. To verify the ECS of around 3C all you have to to do is apply a simple analysis trick. Use a time series that is land-only. The land-only BEST series is a good choice.

    This gives something that is close to a final ECS number because land does not have a significant heat sink that will squirrel away excess thermal energy.

  79. -1=e^ipi says:

    I have a question about that. Let’s say that you are correct about your claim with respect to land-only measurements. Since the Earth’s land has a stronger distribution to polar regions (lots of land is located at the northern mid-latitudes), won’t this bias any estimate of climate sensitivity upward due to polar amplification?

  80. The data is what it is and the models of the data reflect that situation. Only a few factors are needed to explain global temperature variability superimposed on the secular increase in temperature due to CO2 emissions:

    Say the land was your computer’s CPU and the ocean was the CPU’s heat sink. It doesn’t work very well if the heat sink is far away from the CPU.

  81. verytallguy says:

    -1

    There are essentially three ways of looking at sensitivity. As I understand it, broadly speaking

    1) The instrumental record gives the lowest range
    2) GCMs (and simpler physics based approaches) are somewhere in the middle
    3) Paleoclimate gives the highest eg http://pubs.giss.nasa.gov/abs/ro09210n.html … implies a warming of 2.2-4.8 K per doubling of atmospheric CO2 (h/t BBD)

    The reason the IPCC gives such a wide range is precisely because it’s difficult (impossible?) to reconcile these, and all are credible.

    See AR5 box 12.2

    The argument you use that GCMs are too sensitive can be used to argue that from paleoclimate, your estimates must be wrong.

  82. -1=e^ipi says:

    @ WebHubbleTelescope –

    I don’t think you answered my question. How are you dealing with polar amplification?

    @ VeryTallGuy –
    The higher paleoclimate estimates can be explained by a combination of two factors:

    1. The ECS is not as well defined as TCR or ESS, so the practical definition varies from paper to paper. Generally the ECS definition used in paleoclimate studies will involve a longer time scale and be closer to the ESS than an instrumental definition.

    2. Many paleoclimate studies often do not fully take into account other factors that are causing temperature changes (Milankovitch cycles, continental changes), thus some of this temperature change is attributed to CO2, leading to a higher estimate of climate sensitivity.

  83. Arthur Smith says:

    The other factor is that recent forcing values particularly for aerosols are quite uncertain – there is a huge range in the IPCC reports. Hansen et al (“Earth’s energy imbalance and implications”, Atmos. Chem. Phys., 11, 13421–13449, 2011, download from http://pubs.giss.nasa.gov/docs/2011/2011_Hansen_etal_1.pdf ) argue pretty strongly that typically used aerosol (negative) forcings are too small by about a factor of two. Any uncertainty in historical forcings feeds directly into uncertainty in sensitivity derived from observations.

  84. -1,

    Generally the ECS definition used in paleoclimate studies will involve a longer time scale and be closer to the ESS than an instrumental definition.

    This is not correct, I think. In most paleo studies you include changes in CO2 and albedo (essentially ice-albedo feedback) as being the external forcings. Hence, the estimate is the fast feedback response and therefore more reasonably regarded as the ECS than the ESS.

  85. verytallguy says:

    -1

    1) An assertion without reference, and you are flat wrong on this I’m afraid. The studies take the albedo effects from ice sheets and other slow feedbacks out before calculating the ECS. For instance, the AR4 description, section 9.6.3.2, of the methodology from the LGM, …When forced with changes in greenhouse gas concentrations and the extent and height of ice sheet boundary conditions…

    2) Another assertion without a reference. Citation please.

    Rather than have me and others do your research for you, if you actually look for a reference to back up your assertions before posting, you’ll find you post less but learn more.

  86. -1=e^ipi says:

    Perhaps I am ignorant with respect to the consistency of the ECS definition. Can you guys please enlighten me? My understanding is that ECS is the equilibrium change involving fast-feedbacks only. But how does one determine what is a fast-feedback and what is a slow feedback? How fast does a feedback need to be to be considered ‘fast’? In GCMs the definition is clearer because GCMs are an approximation of reality with a finite number of feedbacks that are easy to distinguish between fast and slow (and one can simply hold the ‘slow’ feedbacks constant to get ECS). However, if one is trying to obtain ECS from paleoclimate or instrumental data it seems less clear. Furthermore, there might even be a continuum of feedbacks with timescales ranging from fast to slow. Where do you draw the line?

    With respect to some paleoclimate estimates not taking specific factors into account, a good example is http://rsta.royalsocietypublishing.org/content/371/2001/20120294. The pleistocene estimate only considers greenhouse gas and ice-albedo forcings. The fact that Milankovitch cycles, which cause the ice ages, could act as a forcing that explains temperature changes over the pleistocene is ignored. The possibility that continental changes could explain changes during the cenozoic is ignored. This is also arguably a good example of paleoclimate ECS definitions involving much larger timescales than instrumental ECS definitions.

  87. verytallguy says:

    -1,

    very good questions.  

    I am not your research assistant. 

    If you don’t know the answers why make assertions which depend on the answers?

    Research more, cite more, assert less.

    Do report back.

  88. Meow says:

    The pleistocene estimate only considers greenhouse gas and ice-albedo forcings. The fact that Milankovitch cycles, which cause the ice ages, could act as a forcing that explains temperature changes over the pleistocene is ignored.

    Except that the global annual insolation variance over a Milankovitch cycle is tiny, as in < +- 0.1 W/m^2. (Try http://www.imcce.fr/Equipes/ASD/insola/earth/online/index.php , selecting "mean annual insolation", say for the last 1MY, and see). The Milankovitch theory is that orbital factors cause much larger changes in *high-latitude NH* insolation, which are amplified by feedbacks (such as ice-albedo and GHGs) into extensive global climate changes. So no, the paper does not ignore the Milankovitch forcing.

  89. BBD says:

    -1

    More from your favorite climate scientist:

    3.2 Fast-feedback climate sensitivity

    Fast-feedback climate sensitivity can be determined precisely from paleoclimate data for recent glacial-interglacial climate oscillations. This is possible because we can readily find times when Earth was in quasi-equilibrium with its ‘boundary forcings’. Boundary forcings are factors that affect the planet’s energy balance, such as solar irradiance, continental locations, ice sheet distribution, and atmospheric amount of long-lived GHGs (CO2, CH4 and N2O).

    Quasi-equilibrium means Earth is in radiation balance with space within a small fraction of 1 W/m2. For example, the mean planetary energy imbalance was small averaged over several millennia of the Last Glacial Maximum (LGM, which peaked about 20,000 years ago) or averaged over the Holocene (prior to the time of large human-made changes). This assertion is proven by considering the contrary: a sustained imbalance of 1 W/m2 would have melted all ice on Earth or changed ocean temperature a large amount, neither of which occurred.

    The altered boundary conditions that maintained the climate change between these two periods had to be changes on Earth’s surface and changes of long-lived atmospheric constituents, because the incoming solar energy does not change much in 20,000 years. Changes of long-lived GHGs are known accurately for the past 800,000 years from Antarctic ice core data (Luthi et al., 2008; Loulergue et al., 2008). Climate forcings due to GHG and surface albedo changes between the LGM and Holocene were approximately 3 and 3.5 W/m2, respectively, with largest uncertainty (±1 W/m2) in the surface change (ice sheet area, vegetation distribution, shoreline movement) due to uncertainty in ice sheet sizes (Hansen et al., 1984; Hewitt and Mitchell, 1997).

    Global mean temperature change between the LGM and Holocene has been estimated from paleo temperature data and from climate models constrained by paleo data. Shakun and Carlson (2010) obtain a data-based estimate of 4.9°C for the difference between the Altithermal (peak Holocene warmth, prior to the past century) and peak LGM conditions. They suggest that this estimate may be on the low side, mainly because they lack data in some regions where large temperature change is likely, but their record is affected by LGM cooling of 17°C on Greenland. A comprehensive multi-model study of Schneider von Deimling et al. (2006) finds a temperature difference of 5.8 ± 1.4°C between LGM and the Holocene, with this result including the effect of a prescribed LGM aerosol forcing of ‒1.2 W/m2. The appropriate temperature difference for our purposes is between average Holocene conditions and LGM conditions averaged over several millennia. We take 5 ± 1°C as our best estimate. Although the estimated uncertainty is necessarily partly subjective, we believe it is a generous (large) estimate for 1σ uncertainty.

    The empirical fast-feedback climate sensitivity that we infer from the LGM-Holocene comparison is thus 5°C/6.5 W/m2 ~ ¾ ± ¼ °C per W/m2 or 3 ± 1°C for doubled CO2. The fact that ice sheet and GHG boundary conditions are actually slow climate feedbacks is irrelevant for the purpose of evaluating the fast-feedback climate sensitivity.

    Hansen & Sato (2012)

  90. Willard says:

    Have you considered having a blog on that paper, BBD?

  91. Meow says:

    BBD: (Quoting your quote of Hansen & Sato 2012):

    The altered boundary conditions that maintained the climate change between these two periods had to be changes on Earth’s surface and changes of long-lived atmospheric constituents, because the incoming solar energy does not change much in 20,000 years.

    This needs unpacking. As I noted above, annual global insolation has changed only trivially over the last 1MY. However, high-latitude insolation changes greatly over 20KY (from 20KA to today 65N July-August insolation changed from ~375 W/m^2 to ~385 W/m^2, and it has been as high as ~409 W/m^2 in the Holocene). What Hansen is saying is that this localized forcing initiates the albedo and GHG changes that then drive global climate change. See the above-quoted paper at p26-27.

  92. BBD says:

    The fact that ice sheet and GHG boundary conditions are actually slow climate feedbacks is irrelevant for the purpose of evaluating the fast-feedback climate sensitivity.

  93. BBD says:

    Sorry, Meow, also this:

    The altered boundary conditions that maintained the climate change between these two periods had to be changes on Earth’s surface and changes of long-lived atmospheric constituents, because the incoming solar energy does not change much in 20,000 years.

  94. BBD says:

    Sorry about the repetition, W.

    It just keeps coming up.

  95. Steven Mosher says:

    BBD.

    I second the request for a blog post on HS 2012.

    In particular I have a question about this

    Henceforth, by fast-feedback climate sensitivity,
    Sff, we refer to the all fast-feedback sensitivity. Sff is
    thus the fast-feedback sensitivity that we estimated
    from empirical data to be
    0:75 +- 0:125
    C per W=m2
    ; (2)
    which is equivalent to 3 0.5C for doubled CO2.
    High precision is possible for fast-feedback climate
    sensitivity because GHG amount is known accurately,

    #####################################

    A couple points. First I like hansens approach because, as he argues, it includes all physics.. or in his words “the physics is exact”. Second, in a few ( from memory here so maybe wrong )
    of his papers he asserts that doubling c02 gets us 4 watts.

    with a lambda of .75, then of course the response to doubling c02 is 3C . But, If we use 3.71 W per doubling we get to something on the order of 2.8C per doubling. So my question is where does Hansen get 4 watts? is there something I am missing

  96. Willard says:

    > I second the request for a blog post on HS 2012.

    My suggestion was not for a blog post, but for a blog blog. Imagine the climate blogosphere with lots of little blogs doing one thing well… A unixian blogland, so to speak:

    This is the Unix philosophy: Write programs that do one thing and do it well. Write programs to work together. Write programs to handle text streams, because that is a universal interface.

    http://en.wikipedia.org/wiki/Unix_philosophy

    Willard talking about usability: I know, I know.

  97. A blog blog has the same problems simple blogs have. They are mainly flowing information. Old blog posts are not read very much, and they get outdated.

    The Wiki approach is an alternative, but the all encompassing Wikipedia has its own practical limitations. I have been pondering on an similar maintained site that would contain systematically organized articles related to climate change and references to both scientific papers and interesting net resources.

    That kind of approach could be done on many levels including
    – a simple introduction to most important issues
    – a net based textbook
    – the ultimate replacement of IPCC reports

    Each of the levels requires different details of implementation, but the main idea of continuously maintained highly interlinked resource could be the same.

  98. verytallguy says:

    -1,

    looks like BBD is your research assistant. I hope you’re paying well 😉

  99. BBD says:

    Steven

    But, If we use 3.71 W per doubling we get to something on the order of 2.8C per doubling. So my question is where does Hansen get 4 watts? is there something I am missing

    Hansen includes CH4 and N2O as feedbacks to increasing CO2 forcing, which pushes the effective forcing from 2 x CO2 up to ~4W/m^2.

    The proportional forcing is discussed in more detail in Hansen et al. (2008):

    We use the GHG and sea level data to calculate climate forcing by GHGs and surface albedo change as in prior calculations [7], but with two refinements. First, we specify the
    N2O climate forcing as 12 percent of the sum of the CO2 and CH4 forcings, rather than the 15 percent estimated earlier [7] Because N2O data are not available for the entire record, and
    its forcing is small and highly correlated with CO2 and CH4, we take the GHG effective forcing as

    Fe (GHGs) = 1.12 [Fa(CO2) + 1.4 Fa(CH4)], (1)

    using published formulae for Fa of each gas [20]. The factor 1.4 accounts for the higher efficacy of CH4 relative to CO2, which is due mainly to the indirect effect of CH4 on tropospheric ozone and stratospheric water vapor [12]. The resulting GHG forcing between the LGM and late Holocene is 3 W/m2, apportioned as 75% CO2, 14% CH4 and 11% N2O.

  100. BBD says:

    Willard

    It (H&S12 micro-blog) is an interesting idea, but really all that’s required is that anyone interested enough to argue the toss RTFR. Unfortunately, you know how likely that is to happen…

    But really, I can’t read the stuff for them.

  101. -1=e^ipi says:

    Alright, I’m done commenting on this Blog. Too much censorship.

    Enjoy your echo chamber ATTP!

  102. -1,
    Whatever, but I have asked you to not make statements about scientists that appear defamatory, and you seem to be unable to avoid doing so.

  103. BBD says:

    Well I’m sure regulars will be happy for the Hansen paste-fest to come to an end 😉

  104. Willard says:

    > A blog blog has the same problems simple blogs have. They are mainly flowing information. Old blog posts are not read very much, and they get outdated.

    “Simple blogs” referred to blog posts, Pekka. Suppose you have a blog on HS08. All the posts relate to it. Each paragraph of HS08 gets its own post. Updating one post from the HS08 blog is easier than updating one blog post on the whole paper. This fact is known since the beginnings of exegesis.

    The same applies to wiki entries. The HS08 blog could be a wiki. It’s more a matter of information architecture than content management system. Take this:

    https://contrarianmatrix.wordpress.com/

    Is this a blog or a wiki?

  105. Rachel M says:

    I deleted your comments, -1, not AndThen. It’s not censorship since you are still free to express your views elsewhere.

    I have been a bit stricter lately and I don’t tolerate suggestions that scientists are deliberately manipulating the data to get more funding and/or make global warming look worse. Aside from being defamatory, it’s also provocative and not conducive to a harmonious comment thread. I’m also just a bit fed up with those sorts of accusations.

  106. Joshua says:

    Heh. “Censorship.”

    Ya gotta love “skeptics.”

  107. -1=e^ipi says:

    [Mod : Sorry, that’s crossed the line. You’re no longer welcome. And that’s from me – ATTP.]

  108. Willard says:

    You should give Judy’s a try, -1. She recently welcomed Swood:

    Actually I am enjoying swood1000’s contributions, and they make more sense to me than anything you are willard are criticizing him over.

    http://judithcurry.com/2015/03/03/ipcc-in-transition/#comment-680814

    Please join the “you make no sense” chorus over there.

  109. Steven Mosher says:

    Thanks BBD, that explains it.

Leave a Reply

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

WordPress.com Logo

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

Twitter picture

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

Facebook photo

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

Google+ photo

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

Connecting to %s