95% and things

I notice that the deadline for the submission to the UK government’s IPCC inquiry has passed. I didn’t submit anything; partly because I just forgot, and partly because I’m still just an anonymous blogger and doing so would seem to be taking myself a little too seriously. I have, however, discovered that a number of other people have made submissions. It pains me to link to the blog of the most unpleasant, non-anonymous person I’ve encountered online, but it does contain links to some of the other submissions. I notice, there, that our Scottish friend has submitted evidence. A quick perusal of this would seem to confirm my suspicion that a number of people (myself included) wasted a great deal of time discussing Murry Salby’s ideas with him.

There do, however, seem to be some submissions that are rather critical of the IPCC’s statement

It is extremely likely that more than half of the observed increase in global average surface temperature from 1951 to 2010 was caused by the anthropogenic increase in greenhouse gas concentrations and other anthropogenic forcings together. The best estimate of the human-induced contribution to warming is similar to the observed warming over this period.

This statement is basically saying that it is extremely likely (95% – 100% probability) that most of the warming since 1951 is anthropogenic. As far as I understand it, this is not claiming that 95% of the warming is anthropogenic, but is an attribution statement (i.e., almost all the warming can be attributed to anthropogenic influences). Judith Curry has, however, apparently suggested that the IPCC is deliberately trying to mislead the public with this statement. My view, on the other hand, is that the reason the IPCC has said this is because it is extremely likely (virtually certain, I would say) that most of the warming since 1951 is anthropogenic.

Why? Well, we’re fairly certain that it’s not the Sun. It seems extremely unlikely that the geothermal flux has suddenly increased. The only other non-anthropogenic option is internal variability, but that can only move energy around. It can’t create energy, so can’t explain the overall warming. Additionally, as I explain here, if the rise in surface temperature was due to internal variability, we’d expect it to be accompanied by a negative energy imbalance (i.e., we should be losing more energy than we’re gaining), not a positive energy imbalance. Also, the heat content of the land and atmosphere is relatively low, so if the surface warming was simply associated with some natural cycle, the energy associated with this surface warming should have been lost relatively quickly (months or a few years).

What’s more, the known anthropogenic forcings are consistent with the overall warming. Ultimately, we have anthropogenic influences that can largely explain the overall warming and natural/internal variability processes that cannot, and hence we are virtually certain that the warming since 1951 is almost all anthropogenic. A challenge I have for those who criticise the IPCC statement is to come up with a non-anthropogenic mechanism, that’s physically motivated (i.e., not simply curve-fitting), that can largely explain the warming, and that doesn’t violate the laws of physics. If you can do that, maybe you have a point. If you can’t, you probably don’t.

I thought I’d finish this post by discussing the figure below, which is – I believe – associated with the IPCC’s attribution statement. The figure shows the observed surface warming (black), the contribution due to greenhouse gases (green), other anthropogenic influences (yellow), natural warming (purple), internal variability (no colour because it is too small to see), and the net anthropogenic contribution (orange). Each contribution has an associated uncertainty range.

IPCC AR5 attribution figure (Fig 10.5 in AR5 WG1)

IPCC AR5 attribution figure (Fig 10.5 in AR5 WG1)

This figure has been criticised by some because they claim that the overall anthropogenic contribution (orange) has much smaller error bars than would be the case if you summed (in quadrature I assume) the errors associated with all the anthropogenic contributions. This has been covered by Gavin Schmidt, but I thought I would add my thoughts here.

What those who criticise this graph fail to realise (which is quite remarkable, given who some of them are) is that the net anthropogenic contribution (orange) is not based on the anthropogenic contributions only. It’s determined by considering the anthropogenic and natural/internal variability contributions. I’ll see if I can explain this (and I’ll hope to get it roughly correct). If we consider the anthropogenic contributions only, then it is possible that they could provide much more warming than is observed. However, when the natural/internal variability contributions are also considered, it seems unlikely – if this were the case – that these could provide sufficient cooling to then match the observed warming. Similarly, considering anthropogenic contributions only, it is possible that they could have produced much less warming than is observed. It is, however, unlikely that the natural/internal variability contribution could then provide sufficient warming to explain the observed warming. Hence, by combining all the possible contributions, one can constrain the anthropogenic influence much more tightly than if one were to consider the anthropogenic contribution alone.

Assuming I’m right (and, as usual, am happy to be convinced that I’m not) I’m quite amazed that those who criticise this figure have not recognised that attribution is determined by considering all possible contributions, not just the anthropogenic ones. I must admit, that it took me a while to work out what the figure was actually saying but, having done so, it doesn’t seem all that complicated (assuming I’m roughly correct that is) and am surprised that others haven’t managed to do the same. That would, in my opinion, be true scepticism. In the spirit of scepticism, I’m happy to consider that I might be wrong and am happy to be convinced – by someone who can actually construct a coherent argument – that I am.

This entry was posted in Climate change, Climate sensitivity, Global warming, IPCC, Judith Curry and tagged , , , , , , , , , . Bookmark the permalink.

112 Responses to 95% and things

  1. Rachel says:

    It pains me to see you link to an unpleasant, non-anonymous person online. Most of his comments here have been personal attacks directed at you. Couldn’t you just link to Richard Tol’s and Scotty’s submissions separately?

  2. You should not be too modest … the last 4 pages were copy and pasted from this website (although a lot shorter when I started)!! Also, I mentioned ENSO as evidence that regionally CO2 does change in response to temperature.

    However, the main reason I came here was on a completely different subject and that is I’m searching for a list of the most widely read “convinced” blogs.

  3. > our Scottish friend has submitted

    You have to admit, its impressively barking.

  4. Hi re: “unpleasant, non-anonymous “, could you direct me to the comments that concern you and whilst I don’t know the person I will try to see whether there is anything I can do.

  5. ScotScep

    the last 4 pages were copy and pasted from this website (although a lot shorter when I started)

    Sorry, what do you mean? From my site?

    the main reason I came here was on a completely different subject and that is I’m searching for a list of the most widely read “convinced” blogs.

    Why, and why did that mean you had to actually comment here? Are you wanting some information?

  6. Rachel,

    Couldn’t you just link to Richard Tol’s and Scotty’s submissions separately?

    Possibly, but it just seemed easier and more apt to simply link to the post I did. He may add more, so it may end up being a resource of sorts.

  7. ScotScep,

    could you direct me to the comments that concern you

    No, I’d really rather not. And, to be honest, I’m not really looking for anything to be done. People get to “own” their behaviour as far as I’m concerned.

  8. WottsUpWithThat .. if you’ve got something to say, why not just write it down and send it to the committee anyway?

  9. ScotScep,

    I suspect that they will have received submissions saying most of what I would have to say from people much more credible than I am (although, I guess, if everyone holds that view, maybe they haven’t).

  10. andthentheresphysics says:
    People get to “own” their behaviour as far as I’m concerned.

    In this kind of area it is easy to become detached from how comments are felt by those on the other end so sometimes people are not aware of their behaviour and perhaps a word from a sceptic might tone down the language.

  11. sometimes people are not aware of their behaviour and perhaps a word from a sceptic might tone down the language.

    Possibly, but in this case if that were true this person must have virtually no self-awareness. I’m fairly certain that they’re saying precisely what they want to say and intending it to mean precisely what it appears to mean. Anyway, it’s not really that important and I’d really rather that this thread didn’t become dominated by that particular aspect of this post.

  12. I wonder what the point of the inquiry is. A review of a review. Which will itself be reviewed, perhaps. Is this anything other than giving MPs a chance to chatter?

  13. William, I have no idea and partly wondered the same thing. My cynical side (most of me at the moment, to be honest) assumed it’s been motivated by those influences who think that the IPCC should be disbanded/completely changed and have convinced some (who may have not needed much convincing) that an inquiry is needed to as to hasten its demise.

  14. ScotScep
    the last 4 pages were copy and pasted from this website (although a lot shorter when I started)
    andthentheresphysics says:
    Sorry, what do you mean? From my site?

    It was a comment I posted … I literally copy and pasted it from your website. Kind of ironic!

  15. ScotScep, thanks. That’s quite good to know. Would be rather disappointed to discover that anything I’d written could have been modified to end up saying what you’ve said in the last 4 pages of your submission.

  16. Rachel says:

    I must say I don’t envy the people who have to read all those submissions. What a tedious task!

    Scotty, where is the comment you have copy and pasted? I’m curious.

  17. klem says:

    “It is extremely likely that more than half of the observed increase in global average surface temperature from 1951 to 2010 was caused by the anthropogenic increase in greenhouse gas concentrations and other anthropogenic forcings together.”

    Um, if more than half of the warming since 1951 is due to observed anthropogenic increase in greenhouse gas concentrations and other anthropogenic forcings, then wouldn’t that also mean that 17 years of observed near-zero warming are also caused by the same influences?

  18. klem, I wondered if someone would suggest that. That basic answer is no. What this particular study shows is that the surface temperature has risen by about 0.6oC since 1951, and most of this can be attributed to anthropogenic influences. That doesn’t immediately imply that every rise in atmospheric CO2 has to be accompanied by a corresponding rise in surface temperatures. It might be true if the only influence was anthropogenic, but – as the figure clearly shows – internal variability and other natural factors can also influence the surface temperatures and can produce both warming and cooling. Consequently, the surface temperature shows variability. The current “hiatus” (during which surface temperatures have likely still risen by 0.1oC per decade) is likely a consequence of internal variability reducing the warming trend (as it can quite clearly do). Therefore, there is some expectation that without this natural influence, surface temperatures would be higher today than they actually are (i.e., anthropogenic influences could have actually contributed to more than 100% of the warming since 1951, with internal variability/natural influences having a cooling effect).

  19. I’m sure one of the crew here can help me with this. If the temp increase is pretty well certainly the influence of humans, why did it initially manifest as atmospheric increase, then change to an oceanic increase leading to this apparent hiatus in atmospheric increase? I understand the increased temp is being absorbed by the oceans, if so, why did it not do this before, why over the last 16 years? It appears logical that natural variations are affecting temperatures up as well as down, but the change in direction of heat absorption is puzzling. Thanks in advance.

  20. Gareth, it didn’t really do what you think it did (at least, I think that’s a fair statement). If the system is out of equilibrium (i.e., we have an energy imbalance) then the whole climate system needs to gain energy so as to retain equilibrium. About 93% of the energy, however, goes into the ocean and a few percent is associated with heating the surface. However, this distribution of the energy isn’t fixed and the fraction going into the ocean can varying depending on ENSO and other ocean cycles. Wind driven gyres – for example – can act to transport water from the surface to depth, taking the energy with it and effectively moving this energy to the deeper parts of the ocean and increasing the fraction of the energy that the ocean absorbs (i.e., the surface remains cool and absorbs more energy simply because it radiates/conducts less into the atmosphere).

    So, imagine that rather than the ocean absorbing 93% of the excess energy, it changes to 95%. This can change the fraction heating the surface by 50%. Similarly, if the fraction going into the ocean reduces to 91%, this can increase the fraction heating the surface by 50%. Hence small changes in the rate at which the ocean is taking up energy can have a big effect on the rate of surface warming. That’s essentially – I think – what the internal variability bar in the above figure is indicating.

    I should say that I may not have explained that as clearly as I could have and others may well be able to do a better job, but I think it is roughly correct and hopefully answers your question.

  21. BG says:

    I am not at all sure I would interpret the graph as you have described.

  22. BG, how would you interpret it then?

  23. BG says:

    Maybe this is what you were saying, but it did not seem to be to me:

    OBS = (net)ANT + NAT + IV

    (net) ANT = GHG + OA

    = OBS – NAT – IV

  24. BG, not really. I’m replying on my phone so will respond in more detail later.

  25. BG, let me see if I can explain what I was meaning. If you were to consider the anthropogenic contributions only, then the uncertainty range is quite large. Probably an uncertainty of around +-0.5oC. If this is the 2σ range, than that would suggest that there’s a 2.5 percent chance than anthropogenic influence could have provided less than 0.1oC of warming. If, however, you determine the uncertainties on the natural/IV contributions it’s about 0.15oC. Hence, if natural/IV are to provide the rest of the warming (0.5oC) that would be a 6σ chance. That’s so unlikely, that you can rule out that anthropogenic influences provide so little of the warming (i.e., the actual chance of anthropogenic influences providing less than 0.1oC of warming is miniscule).

    Essentially you can combine the likelihood of the warming provided by anthropogenic forcings with the likelihood that natural/IV can provide the remaining warming/cooling to much more tightly constrain the likely anthropogenic contribution and because natural/IV can only provide a small fraction of the warming/cooling, that means the anthropogenic influence dominates.

  26. Frank says:

    There was an 0.4 degC global warming between 1920 and 1940. The same type of studies that attribute warming in the second half of the 20th century to man (mainly CO2 from burning fossil fuels) say that none of the forcings you cite were responsible for this warming – it was mostly natural variation. This is substantial evidence that natural variability can be bigger than 0.1 degC. The natural variation between the MWP, the LIA and the temperatures before the mid 20th century is likely to be even bigger. Chose changes haven’t (to my knowledge) been attributed to any of the forcings you cite, so they could be natural variability. ENSO is a form of natural variability that produced an 0.25 degC warming in 1997-8 and 10 degC local changes in SSTs. We know there are other oscillation (PDO, AMO) like ENSO with longer periods, but we haven’t observed them for long enough to know what amplitude of temperature change accompanies them (perhaps the 0.4 degC of warming that took place in 1920-1940 and 1975-1995 was associated with the PDO). A change in the rate of upwelling of cold water of South America is partially responsible for this change. Then we have the pause since 2000 which appears to be natural variation that has so far negated perhaps 0.3 degC of GHG-mediated warming. The greatest potential source of natural variability appears to reside in the MOC, the formation of deep water in the polar regions which results in upwelling of cold water in much warmer areas. Climate scientists speculate that a shutdown in part of the MOC caused by release of freshwater from melting ice caps cause the Younger Dryas which cause a sudden 15 degC drop in Greenland temperature and perhaps 5 degC globally. Natural variability in the MOC obviously has some potential to cause 0.5 degC changes in global temperature and have no reliable information on this subject. The top 50 m of ocean (that portion responds to seasonal changes in solar radiation and is in near equilibrium with the atmosphere) has a very small heat capacity compared to the much colder deeper ocean. Minor changes in heat transport between these two reservoirs can have a dramatic effect on SSTs, but still cause changes in the temperature of the deeper ocean that are difficult to detect. (The average depth of the oceans is 2000 m, so this reservoir is about 40 times bigger than the mixed layer and atmosphere. An 0.4 degC cooling of the surface requires only an 0.01 degC warming of the deeper ocean.) Then there is the earth’s albedo which reflects more than 100 W/m2 of incoming solar radiation back to space. The current forcing from CO2 is only about 2% as big as the effect of albedo, so small natural variation in albedo could have a much bigger effect than 0.1 degC.

  27. Frank,

    ENSO is a form of natural variability that produced an 0.25 degC warming in 1997-8

    Sure, all of what you say could produce some amount of surface warming. The main issue is, however, that the heat capacity of the land and atmosphere is low enough that the energy associated with this warming should be lost quickly (months, a few years). Furthermore, we have a positive energy imbalance (i.e., the climate system is gaining energy). Everything you describe should be associated with a negative energy imbalance (i.e., if some natural process were to produce a rise in surface temperatures, that should push surface temperatures above equilibrium and produce an energy excess and a negative energy imbalance). So, if what you suggest can explain all of what we observe, fine. I really don’t think it can.

  28. John Mashey says:

    The Holocene almost certainly has not had a purely-“natural” climate for 8,000 years, although pre-1AD the changes were fairly slow (in CO2 and CH4 buildup0 above the natural decline expected). Sometime after 1 CE, there got to be enough humans that pandemics could drop CO2 (by reforestration), and wars drop CH4 even faster (as per destruction of rice paddies by Mongols), so that one can see effects on multi-decadal/century scale.
    Here’s Law Dome CO2, 1 CE – 1900 CE.
    Roman era: CO2 higher than it “should have been”
    pandemics, drop CO2.
    lack of pandemics: rise into MWP, thus not entirely natural
    smaller pandemics: CO2 jiggles
    Unique ~9ppm CO2 drop 1525 CD – 1600CE: the biggest pandemic in human history, 50Mperson die-off in Americas from European disease, massive regrowth in high-biomass areas, followed by (natural) volcanoes and Maunder Minimum.

    See Ruddiman, Kutzbach, Vavrus (2011). or more comprehensively, Bill's brand-new book, Earth Transformed,. on which gave just gave the Tyndall Lecture @ AGU Thursday. I recommend this highly: while there is much random noise in the climate system at various time-scales, climate has not been entirely natural for milennia, not just since Industrial Revolution.

  29. John Mashey says:

    Argh, sorry, Rachel/ATTP: can you fix the URLs in previous?

  30. John, I’ve tried to. Let me know if I fixed it how you’d intended it to be.

  31. John Mashey says:

    Yes, that’s right, thanks much.

  32. Tom Curtis says:

    Running through Frank’s comments:
    1) The “0.4 C” rise in temperature was not between 1920 and 1940, but between 1910 and 1940. If, instead of cherry picking dates to include the bottom of a sharp, and fairly obvious trough, we just take a 50 year interval, the rise is 0.3 C, significantly less than the 0.5 C rise in the later half of the twentieth century.

    2) The IPCC do not claim that “… none of the forcings [Anders] cite[s] were responsible for this warming”, and nor do they say that most of the warming was natural. They say that a natural cause for that warming cannot be excluded, and indicate that a combination of natural and anthropogenic forcings are the best explanation of the warming. This is, therefore, not “substantial evidence that natural variability can be bigger than 0.1 degC”.

    3) The changes between the MWP and the LIA are probably greater than 0.2 C, and are likely to be around 0.4 C, but occur over a much longer period than the 60 odd years of the equivalent warming since 1950. The radical difference in rates does not suggest an equivalence of cause.

    4) Frank cites several purportedly natural oscillations* but fails to note that for those with a longer period, the late twentieth century warming cited by the IPCC is from 1951, ie, following closely a mid century peak in those signals, to now, again following the peak. Consequently those oscillations are likely to have had near zero effect on the overall rise, even if they may have slowed or accelerated the warming in individual decades during that time.

    5) Frank also cites ENSO, failing to note that ENSO has recently shown a strong positive trend in the SOI (ie, towards cooler conditions) and that that trend peaked in 2011 with the strongest SOI in over 130 years of record; a fact somewhat obscured in temperature based indices by the obscuring trend of global warming. Consequently ENSO is at best neutral with respect to 1951, and is likely a slight cooling influence.

    * It is not clear that the dominant signal of the AMO in the Twentieth century is natural, rather than the product of mostly anthropogenic twentieth century forcing of the North Atlantic.

  33. BG says:


    Speaking of communications, particularly with the general public, did you happen to read comment #5 at the RC link you provided?

  34. BG, I didn’t, until you pointed it out. What might be of interest is that I did not really understand Gavin’s post myself and most of what I’ve said in this post was based on me thinking about the figure until it appeared to make sense. So, I don’t know if I’ve done it more clearly than Gavin or even that my explanation is correct (I always worry that I do get things wrong, this is not really my field). So, that comment may have some merit and I may well make similar mistakes here, to be honest.

  35. Rachel says:

    That’s why I don’t read the RC website. I don’t understand most of it.

  36. verytallguy says:


    a confession. I read the RC post, and I’ve read ATTP above, and I still don’t understand where the ANT figure comes from, and how it can possibly have lower error bounds than GHG+OA.

    ATTP/others – can anyone explain this simply?

  37. Rachel says:

    I’m lost too, VTG.

  38. VTG and Rachel, this is the moment where I hope I’m not completely wrong and that someone like Richard Betts/Gavin Schmidt doesn’t read this and go “what an idiot”.

    Anway, here’s an attempt to explain it. I’ll also simplify a little to hopefully clarify. Let’s imagine that we have only two contributions, anthropogenic and natural/IV. We also have the observed warming. Let’s also imagine that the anthropogenic contribution is very uncertain (as it kind of is) while the natural/IV contribution and the observed warming are very certain. Therefore, we know – quite precisely – what the actual warming is. We also know – quite precisely – what the natural/IV contribution is. If the natural/IV contribution is small and we know what it is (quite precisely) then we know that most of the warming has to be anthropogenic even if – when considered by itself – the anthropogenic contribution is quite uncertain. I know that the actual observed warming and natural/IV contribution are not quite as certain as I’ve implied here, but that was – hopefully – to make it clearer.

    In other words, given that we know what the actual warming is and how much can be natural/IV, we can constrain how much must be anthropogenic. In the ideal scenario where we know exactly what the actual warming is and exactly what the natural/IV contribution was, then we’d know, exactly, what the anthropogenic contribution would have to be.

    In fact, if you consider the figure, the uncertainty is more similar to the combined natural/IV uncertainties because (I think) given that we know that contribution more precisely than the anthropogenic contribution the, uncertainty in the final anthropogenic influence is determined more by the uncertainty in the natural/IV contribution than by the uncertainty in the anthropogenic contribution. That’s not quite right, but roughly correct.

    Anyway, I hope that’s correct, and I hope it makes sense.

  39. BG says:

    VTG and Rachel,

    I’m with you on that. I have not yet had time to go to SkS to see if they address it.

  40. RB says:

    I think the hint might lie in Gavin’s alluding to “degeneracy” and what the AR5 says about degeneracy:

    Attribution studies must compromise between estimating
    responses to different forcings separately, which allows for the possibility of different errors affecting different responses (errors in aerosol forcing which do not affect the response to greenhouse gases, for example), and estimating responses to combined forcings,
    which typically gives smaller uncertainties because it avoids the issue of ‘degeneracy’: if two responses have very similar shapes in space and time, then it may be
    impossible to estimate the magnitude of both from a single set of observations because amplification of one may be almost exactly compensated for byamplification or diminution of the other (Allen et al., 2006).

    Pls to forgive formatting errors. The way I read it is this: if you assume that you are burning x amount of coal, it produces y contribution to temperature by mostly cooling anthro (OA) and corresponding (scaled) amount of warming due to (GHG). So, independently, the cooling and warming influences have wide uncertainty because of uncertainty in the amount of coal burnt. But combined together, specifying one specifies the other also. Normally when you have independent random variables, you would combine their uncertainty variance as sum of variances of each together, but when they are correlated, you wouldn’t do so.

    That explains why the spread in GHG and OA is not the same as spread in ANT, but you still have to how the spread in ANT was calculated as per Gavin’s link on Pg 10-20

    a total anthropogenic contribution to warming can be much more robustly
    constrained by a regression of observed temperature changes onto the simulated responses to all anthropogenic forcings and natural forcings (Figure 10.4; Ribes and Terray, 2013; Gillett et al., 2013).

    …which I haven’t done so yet.

  41. BG, that’s put me in my place 🙂

    RB, I suspect you’re correct that an element I haven’t mentioned/considered is how the errors are actually combined. If they are correlated than it is not a simple summation in quadrature.

    I believe the last bit that you quote is essentially what I’m trying to say here. If you know the range of anthropogenic and natural forcings then, together, they can constrain the anthropogenic influence more tightly than if you considered only the anthropogenic component.

  42. RB says:

    I agree it’s on the lines you described. Let’s say for instance, we have OBS which are tightly constrained. Let’s say that you have NAT+Int.Variability which have relatively smaller spread. Then, let’s say (for argument sake) that NAT and INT. VAR are independent and equal in uncertainty for temp contribution. Then you would add their variances so that their uncertainty is only 1.4 times either of their uncertainty which in any case dominates OBS uncertainty (the observations uncertainty is 2 standard deviations instead of 1 for the others). Then, at a minimum, the uncertainty in ANT is not more than this uncertainty in NAT and INT. VAR together which we know is relatively smaller than either GHG or OA separately. So, at a minimum we know that uncertainty in ANT has to be lower than combining GHG/OA. But there are some additional ways to reduce this uncertainty further which again must be on the lines of reducing “degrees of freedom” by exploiting redundancies.

  43. RB, I agree. That’s pretty much how I had interpreted the reason for the uncertainty in ANT being much smaller than one would expect based on the uncertainties in GHG and OA alone.

  44. RB says:

    I suspect the uncertainty in NAT/INT_VAR is mostly uncertainty in the larger of the two i.e., INT_VAR. So, NAT can perhaps be visualized as one embodiment of Internal Variability on our Earth. (That may not be necessarily justified by physics depending on where the internal variability arises from). And that correspondingly is perhaps essentially the uncertainty in GHG.

  45. RB says:

    er… “perhaps essentially the uncertainty in ANT”

  46. Rachel says:

    Thank you, Anders. Your explanation makes sense to me.

  47. BG says:


    No intent to insult. 😉

    I’m not an expert on the propagation of errors, but I do know from experience with measuring cross-sections of aerosols, and using that to try and predict signals of unknowns from signals of knowns, that the error propagation gets out of hand very quickly, i.e., very big uncertainties of the end product, despite very reasonable uncertainties on the measurables.

    And explaining these things to the non-expert is as much an art form as it is science. Part of what my job has been for the last 25 yrs.

  48. BG, no insult taken 🙂

    You’re certainly correct, error propagation is tricky to do properly and even harder to explain.

  49. verytallguy says:


    thanks for the renewed explanation. Smug greenie that I am, I was contemplating this during my cycle home, thus both thinking and acting on climate change at the same time. I think I reached the same explanation you did:

    The temperature rise itself (T) is measured to low error. This is split thus:

    GHG uncertainty is calculated from first principles, and has wide error bars
    Other anthro, OA (mainly aerosols) likewise.
    And we define total antho (ANT) = (GHG+OA)
    Natural forcing (NAT) and internal variability (IV) are also calculated from first principles, and have low error bars

    T= (GHG + OA) +(NAT+IV)


    (GHG+OA) = T – (NAT+IV) so
    ANT= T-(NAT+IV)

    And as everything on the right is well constrained, ergo the left must be likewise. About right?

    RB then suggests inherent cancelling out ie coal burned produces GHGs and aerosols, so the net uncertainty is reduced. Not sure about that. Need a longer cycle 😉

  50. VTG, I think that’s essentially correct. What I think RB was suggesting is that when you propagate errors you need to consider whether the different contributions are correlated or not. I don’t actually know how to do that, but I did once write a linear regression code for determining the trends and uncertainties in the surface temperature anomalies. I got the right trends, but the wrong errors because I wasn’t taking into account that the data points are correlated. I tried working out how to do it properly, but then just decided that using the Skeptical Science trend calculator was easier 🙂 .

  51. Rachel says:

    VTG, that is very nice, thanks. I had rather given up on following this thread too closely but I think I might be back in business.

    Isn’t it great to be able to cycle home? Give me a bicycle and somewhere to ride it and I’m happy.

  52. Mooloo says:

    The only other non-anthropogenic option is internal variability, but that can only move energy around. It can’t create energy, so can’t explain the overall warming.

    And yet temperature has changed in the past, over the long term and the short term. By large amounts often!

    Something other than anthropogenic influences must be at work in the system. We can debate their size, but to exclude them completely defies the historical record of ice-ages and sudden retreats.

  53. Mooloo, can you please point out where anyone has said that only anthropogenic influences are included here. The figure very clearly includes anthropogenic and natural.

  54. verytallguy says:


    Give me a bicycle and somewhere to ride it and I’m happy.

    Absolutely. Promoting cycling is one of the many win-wins in energy & climate policy, happier, healthier and lower carbon.

  55. RB says:

    Two separate things.

    First …
    I was suggesting a mechanism (of correlation between GHG and OA) by which error bars in (ANT) could be less than error bars in (GHG) and (OA) separately.

    Second …
    Separately, we are left with your
    ANT= T-(NAT+IV)
    But error bars for (IV) and (NAT) do not add up for (ANT). For that, we might guess that (NAT) trend uncertainty is contained by (IV) trend uncertainty as calculated by the modelers.

  56. BBD says:

    The PETM was caused by natural variability.

  57. RB

    I was suggesting a mechanism (of correlation between GHG and OA) by which error bars in (ANT) could be less than error bars in (GHG) and (OA) separately.

    Yes, that was what I understood you to be saying.

    But error bars for (IV) and (NAT) do not add up for (ANT). For that, we might guess that (NAT) trend uncertainty is contained by (IV) trend uncertainty as calculated by the modelers.

    I don’t quite understand what you’re getting at here. Is it simply that, from the figure, the ANT uncertainties are smaller than they would be if they were simply from NAT + IV. I think there is a factor that I think VTG’s explanation (and mine) don’t quite capture. I think that once one has determined the NAT + IV uncertainty one should still consider the likelihood of the GHG + OA forcings providing the remaining warming, which will slightly reduce the net ANT uncertainty (I think, at least).

  58. RB says:

    The confusion was mine – I was thinking of internal variability from model simulations (model ensemble). But NAT=solar+aerosols for e.g. and INT=ocean oscillations. I’ll have to think about how ANT error bar is not larger due to (NAT+IV); this must be due to being constrained by T in responses obtained from models. Since models don’t reproduce ocean oscillations and trends by latitude very well, I’ll check back on these error bars in AR6 🙂

  59. RB says:

    er… NAT=solar, volcanoes etc.

  60. Yes, I did wonder about the aerosols for a minute 🙂

  61. Tom Curtis says:

    I think part of the problem is that none of the factors are truly independent. Specifically, ΔGHG forcing covaries with Δ OA forcing given that rising CO2 emissions will be matched with rising sulfate emissions to a first a approximation. Further, Δ GHG, OA, and Nat response will covary because the size of all depends on TCR. Indeed, higher IV also implies higher TCR, but I am unsure of the scale of the correlation with the other three, or even if the relationship is linear. Finally, the uncertainty in OA forcing is larger than that for GHG, but OA forcing inversely with TCR. The more strongly negative OA forcing, the stronger TCR.

    The upshot of this is that none of the uncertainties can be derived from the others by addition in quadrature. More specifically, the higher Nat + IV, the smaller can be the reduction in Anthro and vice-versa so that the error in anthro is less than the error in Nat + IV added in quadrature. Beyond that, there is less uncertainty about GHG +OA than about each individually, both because the error in forcing of the two covaries, and because error in TCR means a higher estimate of GHG requires a lower (more negative) estimate of OA, thereby resulting in less uncertainty in Anthro than would be implied by adding in quadrature.

  62. Frank says:

    Our host replied: “Sure, all of what you say could produce some amount of surface warming. The main issue is, however, that the heat capacity of the land and atmosphere is low enough that the energy associated with this warming should be lost quickly (months, a few years).”

    The 1920-1940 warming, the 2000’s pause, the LIA and the MWP are not “coulds”, they represent natural variability far greater than 0.1 degC that DID occur and last for more than a decade. That variability can’t be (or hasn’t been) explained by standard forcings, so there is no reason to believe that the +/-0.1 degC shown on your bar graph represents anything like the full range of natural variability The variable exchange with the deep ocean and changing albedo are physical mechanisms that could produce natural variability much greater than 0.1 degC.

    Our host also replied: “Everything you describe should be associated with a negative energy imbalance.”

    The changing convective heat flux to the deep ocean can chaotically effect temperature for a long period of time without changing the balance between incoming and outgoing energy. Even if other forms of natural variation did change produce an imbalance, we can’t detect and quantify the current imbalance with today’s instruments; they aren’t accurate enough. The only reason we are reasonably confident that an imbalance probably exists comes from rising ocean temperature (heat content) during the ARGO period. (The pre-ARGO data was limited and needed post-acquisition adjustments comparable to the observed warming signal.)

  63. verytallguy says:


    LIA and the MWP are not “coulds”, they represent natural variability far greater than 0.1 degC that DID occur and last for more than a decade.

    Any chance of a cite for that Frank? My understanding is that neither is really considered a global phenomenon? But I can’t remember the details.

  64. Tom Curtis says:

    Frank, in fact models run with known natural forcings during the LIA and MWP produce the LIA and MWP. Further, the early twentieth century warming is explained by known forcings except for a dip around 1910, and a peak around 1940, both of which are about 0.1 C, and last for only a decade. Each of your specific claims on which your argument depends, therefore, is false.

  65. Anders mentioned correlations earlier, which my open source R code handles using the generalized least squares algorithm with the argument “corr=corARMA(p=1,q=1)”. These settings use ARMA(1,1) noise, where p=1,q=0 would use AR(1) noise, etc. The outputted summary lists the AIC and BIC, where AIC = Akaike Information Criterion. The p and q values which minimize the AIC (or BIC) likely provide a decent estimate of the noise, at least for these parameterizations.

    The SkS calculator is easier, but only if you want to examine one of the datasets they provide.

    Amusingly, yet another unskeptical claim of a ‘pause’ since 2000 has appeared. The SkS trend calculator can calculate the trend and uncertainty from 2000 to 2013, and for an equal timespan before 2000:

    HadCRUT4 1987-2000 trend: 0.177 ±0.196 °C/decade
    HadCRUT4 2000-2013 trend: 0.042 ±0.161 °C/decade

    The uncertainties overlap, showing that there hasn’t been a statistically significant change in the warming rate. In both cases, the error bars are larger than the observed warming. All this shows is that contrarians compulsively focus on timespans that are too short. Scientists use longer timespans to study the climate because those error bars are smaller. For instance, calculate the trend and uncertainty over both of these timespans at once to reveal the significant surface warming trend:

    HadCRUT4 1987-2013 trend: 0.149 ±0.065 °C/decade

  66. Frank says:

    Reply to Tom Curtis:

    1) I didn’t cherry-pick the starting point for the 1920-1940 warming period, which you say began in 1910. The total rise from 1910 to 1940 is about 0.5 degC, but I chose not to count the recovery from the trough in 1910 and started around 1920 when the temperature was representative of the past 30 years. No matter when you start, this magnitude of this example of natural variably is far greater than 0.1 degC.

    2) In AR4, the IPCC asserted that most of the warming in the SECOND HALF of the 20th century was due to man BECAUSE several of the studies attributing warming to human forcing found that warming before 1950 could not be attributed to known forcings and therefore represented natural variability. The authors of the attribution chapter of AR4 chose to ignore the pre-1950 finds, but you can check the key attribution references in that chapter.

    3) I’m glad you agree that the natural variation association with the MWP, LIA and recovery from the LIA represent natural variation that is much greater than 0.1 degC. FIgure 6.10 of AR 4 WG1, suggests that centennial variability could be as big as 1 degC. The rate of change isn’t important to my argument, but changes of several tenths of a degC appear to have occurred over a few decades many times in the past. !920-1940 doesn’t appear to be unusual in this respect.

    4) I didn’t make any claims about how much or how little natural variability contributed to warming in the second half of the 20th century. I simply said that 0.1 degC is an absurdly small estimate.

    5) I chose ENSO (particularly the 1997-8 El Nino) as another example of natural variability greater than 0.1 degC. This form of natural variability is clearly linked to variation in the upwelling of cold water from the deep ocean. Longer variations in upwelling could produce bigger and longer lasting variations.

    None of your comments suggest that my main point was flawed: Natural variability is much bigger than 0.1 degC.

  67. Frank,

    To avoid strawman like arguments, I’m certainly not suggesting – and I don’t think anyone else is – that natural influences cannot change the Earth’s surface temperature. That would be absurd.

    The LIA is associated with a period when the TSI was particularly low – almost 1Wm-2 than the solar cycle average today. As VTG has pointed out, it is not clear that the MWP was global. The period 1920 – 1940 is associated with a change in average solar forcing of about 0.1Wm-2 and a change in anthropogenic forcing of almost 0.2 Wm-2. So, as Tom points out, most of the periods you mention can be explained by known forcings with the exception of the dip around 1910 and the peak around 1940, both of which are relatively short and could be consistent with internal variability at a level similar to that suggested by the Figure in this post.

  68. Frank,

    I’m not convinced this statement – referring to the period 1920-1940 – is correct

    No matter when you start, this magnitude of this example of natural variably is far greater than 0.1 degC.

    The change in solar forcing for the period 1920-1940 is around 0.1 Wm-2. The change in anthropogenic forcing (ΔF = 5.35 ln (C/Co)) is something like 0.15Wm-2. Therefore, anthropogenic influences likely contributed to more than half of the warming for the period 1920-1940.

  69. BBD says:

    PAGES 2k Consortium (2013) Continental-scale temperature variability during the past two millennia

    Past global climate changes had strong regional expression. To elucidate their spatio-temporal pattern, we reconstructed past temperatures for seven continental-scale regions during the past one to two millennia. The most coherent feature in nearly all of the regional temperature reconstructions is a long-term cooling trend, which ended late in the nineteenth century. At multi-decadal to centennial scales, temperature variability shows distinctly different regional patterns, with more similarity within each hemisphere than between them. There were no globally synchronous multi-decadal warm or cold intervals that define a worldwide Medieval Warm Period or Little Ice Age, but all reconstructions show generally cold conditions between ad 1580 and 1880, punctuated in some regions by warm decades during the eighteenth century. The transition to these colder conditions occurred earlier in the Arctic, Europe and Asia than in North America or the Southern Hemisphere regions. Recent warming reversed the long-term cooling; during the period ad 1971–2000, the area-weighted average reconstructed temperature was higher than any other time in nearly 1,400 years.

  70. BBD says:

    @ Tom Curtis

    Yes, that’s my take, for what little it’s worth.

  71. BBD says:

    @ Frank

    C20th natural variability ~0.3C. Source SkS article; Huber & Knutti (2011)

  72. Tom Curtis says:

    Anders, natural variability can easily be greater than 0.1 C; but natural variability includes variability in natural forcings. The natural forcings are accounted for, and are small taken over the course of the twentieth century. Internal variability can exceed 0.1 C, as for example in the El Nino of 1998, but the only clear examples of this are from ENSO, whose variability occurs over a time span of one or two years. Further, ENSO is known to be a neutral, or more likely negative influence on temperature over the period from 1870-2012. (Of course, it could turn around and become a very slight positive influence within a year because of the short characteristic timescale of ENSO fluctuations.) Frank trades on keeping these distinctions confused. On confusing “not most” with “not at all” for anthropogenic influence for the early Twentieth Century. On confusing the clear short term influence of ENSO with a hypothetical long term influence that is not proven, and whose best conjectured examples are known to be neutral or negative in influence for the period 1951 to now. And on confusing natural fluctuations in known forcings with fluctuations due to internal variability – the later confusion being necessary because the natural fluctuation in forcings since 1951 are known to be small, and slightly negative.

    He will not examine the evidence for each part of his claim independently, for thus examined it will be shown to not support his conclusion. Therefore he always crowds these disparate phenomenon together as though they were all just the same thing.

    And on a side note, changes in volcanic activity were the major change in natural forcings in the MWP (reduced volcanic activity), LIA (much increased volcanic activity), and early twentieth century (greatly reduced volcanic activity after approx 1910).

  73. Tom, yes I agree.

    And on a side note, changes in volcanic activity were the major change in natural forcings in the MWP (reduced volcanic activity), LIA (much increased volcanic activity), and early twentieth century (greatly reduced volcanic activity after approx 1910).

    That I did not know and I realise that I did not mention volcanoes in my post, but – I believe – that their influence is not that significant post 1950 (apart from certain events).

  74. Tom Curtis says:

    Anders, you should check out figure 1 of FAQ 5.1 in AR5 WG1. There is a strong volcanic influence in the late twentieth century, which ran counter to the effect of anthropogenic factors. It has quietened down since the mid 1990s, however, so its effect on the 1951-2012 trend is probably minor.

  75. Tom, yes I was meaning over the entire period – although I could have said it more clearly. I’m aware that there are periods when the volcanic influence was not insignificant. Interesting figure though. I hadn’t read that part of the WG1 yet.

  76. BBD says:


    The best review of the MCA/LIA I have found is Diaz et al. (2011). The expert view seems to be that explosive volcanism and solar variability were instrumental in triggering persistent changes in ocean and atmospheric circulation. It’s a very clear, engaging paper – not in the least hard work – and as a review, full of useful links to the wider literature. Recommended for the keep file.

  77. RB says:

    I think you realize that the IPCC attribution refers to the 1951-2010 period?

  78. Anders, having noticed Frank’s comments above and been watching how things proceed, I’m afraid I can’t just stand by any longer. Still, I’m not sure whether I’m gonna do the right thing now.

    Trouble is, Frank used to be my “discussion mate” elsewhere (I try to leave it as anonymous as I possibly can) for more then 3 years now. For a long time, I thought I can convey some of the mainstream science to his “lukewarm” mind. Unfortunately, and despite two personally pleasant encounters, I’ver never managed to convince him that mainstream science is just that, mainstream science (he isn’t an atmospheric scientist by training). Worse, he seemed to have become more and more immune to my comments over time. In fact, he openly distrusts me. So at some stage, I simply gave up and stopped any further technical discussion. I’ll stick to it also in this particular case, as I pointed out to him a million times already that there are well known “external” reasons for the early 20th century temperature increase (as well as LIA etc), namely volcanic recovery and solar+early anthropogenic forcing change. Note that it isn’t a perfect consensus statement. However, it is widely supported in the literature (sth which Frank fails to mention at all) and in my opinion the most likely explanation which fits very well with the current AR5 wisdom. Nowhere did I ever stated that natural variability doesn’t exist (quite to the contrary). Thanks Tom for picking up all the important bits already.

    What’s the main reason why I deem this comment necessary? Given that Frank’s comments are usually quite technical, the lurkers get certainly confused. This is particularly true if people express their opinions such as if they’d speak for someone else. Note the rather dubious use of “we” to elevate an opinion. While I can’t be bothered to counter these frustratingly circular arguments anymore (believe me, I spent ages to explain the mainstream view and my opinion alike), I sometimes feel the lurkers simply deserve a bit more balanced statements. Neither did I ever want to convince him, nor is it ever going to happen (no, he is not gonna be the next BBD, unfortunately). For the time being, I therefore decided to leave a short comment without further discussion (where we used to have our “debates”) in cases things are too obviously wrong on his part. Not sure this is the right way to go, but it doesn’t take much time and seems to be working given that I keep answering questions from lurkers. Any thoughts of you guys as to what’s the right thing to do (without shifting the thread off-topic)? Ties in with the recent communication threads I guess …

    P.S.: I’m afraid I can’t contribute much in the next days as I am on vacation. Will try anyways …

  79. Karsten, excellent comment. Thanks. Enjoy the vacation. I’m starting to think I need one myself 🙂

  80. dana1981 says:

    You’re correct on the attribution statement. It’s based on very sound science in general.

    There’s also a new paper on this by 3 scientists including Myles Allen, showing that using various different models of internal variability, the IPCC statement remains correct.

  81. Thanks Dana, and yet two of the submissions that I have seen to the UK government’s IPCC inquiry have suggested (claimed?) that this attribution statement is misleading. One of these, I would argue, from someone who is sufficiently qualified to have checked the validity of this statement for themselves.

  82. AnOilMan says:

    BBD: You neglect what many of the implications were for the ‘medieval warm period’.

    The United States Midwest suffered a mega drought. I’m sure drying up the Colorado river will have no serious consequences.

    But Greenland was indeed nice…

  83. RB says:

    As Tom Curtis described regarding El Nino, the IPCC attribution statement applies to the long-period trends over the 1951-2010 period wherein short-term decadal oscillations may not contribute to trend estimates. As a sanity check, assuming zero internal variability over 1951-2010 is consistent with TCR as calculated over this period. From Nic Lewis’ article on Climate Audit as quoted below, the most-likely TCR calculated assuming zero internal variability over 1951-2010 is 1.4C and consistent with TCR calculated from the most recent decade (the 2000s) in Otto et al. over the hiatus period. (There will be substantial error bars around this most-likely TCR estimate though due to uncertainty in aerosol forcing.)

    When instead basing the estimate on the linear trend increase in observed total warming of 0.64°C over 1951–2010 per Jones et al. (2013) – the study cited in the section to which the SPM refers – (the estimated contribution from internal variability being zero) and the linear trend increase in total forcing per AR5 of 1.73 Wm-2, the implied TCR is also 1.4°C.

  84. It’s so adorable when contrarians lecture scientists about natural variability, as though scientists haven’t been discovering and naming different modes of natural variability for centuries.

    Ironically, if no natural changes in the Earth’s temperatures have occurred, then negative feedbacks dominate, cancelling out any forcing, causing climate sensitivity to be exactly zero. This is not the position of mainstream science, which has concluded since 1979 that the equilibrium Charney sensitivity is very likely above 1.5C per doubled CO2.

    However, the Sky Dragon Slayers seem to think that climate sensitivity is exactly zero. Maybe they just need to be told about the LIA and the MWP, and they’ll realize that the existence of natural variability means that the climate sensitivity can’t be exactly zero. Hope springs eternal.

  85. BBD says:


    BBD: You neglect what many of the implications were for the ‘medieval warm period’.

    First line of PAGES-2k abstract:

    Past global climate changes had strong regional expression.

    Which I think covers the megadroughts in the South Western US. If you take a look at the Diaz et al. paper you will find a discussion of the mechanisms now thought to be responsible for this and other regional climates associated with the MCA/LIA.

  86. BBD says:


    Interesting 😉

    Enjoy your break and Christmas etc.

  87. Frank says:

    Tom, Andthenthereisphysics and others: This reference demonstrates that warming prior to 1950 represents natural variability (including solar), not anthropogenic forced change. The total dynamic range of the change during this period is 0.5 degC. This paper was cited in the attribution chapter of AR4 WG1.

    Nozawa, T., T. Nagashima, H. Shiogama, and S. A. Crooks (2005), Detecting natural influence on surface air temperature change in the early twentieth century, Geophys. Res. Lett., 32, L20719, doi:10.1029/ 2005GL023540.
    We analyze surface air temperature datasets simulated by a coupled climate model forced with different external forcings, to diagnose the relative importance of these forcings to the observed warming in the early 20th century. The geographical distribution of linear temperature trends in the simulations forced only by natural contributions (volcanic eruptions and solar variability) shows better agreement with observed trends than that does the simulations forced only by well-mixed greenhouse gases. Using an optimal fingerprinting technique we robustly detect a significant natural contribution to the early 20th century warming. In addition, the amplitude of our simulated natural signal is consistent with the observations. Over the same period, however, we could not detect a greenhouse gas signal in the observed surface temperature in the presence of the external natural forcings. Hence our analysis suggests that external natural factors caused more warming in the early 20th century than anthropogenic factors.

    Click to access NNS05.GRL.pdf

    Could variation in solar output explain the LIA? Of course. However, the uncertainty in reconstructing MWP/LIA temperature and solar output from proxies is huge. IMO (FWIW), we don’t know how much of that change is due to solar variation and how much might have other causes. Other opinions are equally valid, but calling this “settled science” seems dubious. Your figure for solar forcing (-1 W/m2) can be converted to a change in temperature (0.4-1.2 degC) by multiplying by ECS (1.5-4.5). Now include the uncertainty in the forcing (say -0.5 to -1.5 W/m2?). I’d be interested in a reference that shows how accurately we can translate changes Be10 and C14 proxies to changes in total solar output.

    Could volcanos have played a major role in the LIA? Unlikely. The transient signals from volcanos in the historical record are almost indistinguishable from background noise. Consensus estimates of aerosol forcing have dropped recently, so results from models with higher sensitivity to aerosols may over estimate the importance of volcanos in the LIA.

    (There is, of course, no doubt that the LIA existed and therefore must have been bounded by warmer periods, the MWP and the pre-GHG CWP. It doesn’t make any difference to my position whether the MWP was as warm (or global in extent) as the pre-GHG CWP; the three periods define climate change not driven by GHG’s or man.)

    The advantage of downplaying our limited understanding of natural variability (unforced variability) is that it allows scientists to make confident statements about man’s role in recent climate change. The disadvantage arrived with the recognition that there has been a recent unexpected hiatus in surface warming. If one considers the IPCC’s fundamental estimate of potential anthropogenic climate change – 70% likelihood that ECS lies between 1.5 (formerly 2.0) and 4.5 degC – there is plenty of room for natural (unforced) variability to have played a significant role since 1900, including the hiatus. If you consider only the narrower range of climate model output (and associated ECS), there is significant unanticipated warmth around 1940 and cold around 2010. Unfortunately, the physics modules of AOGCMs contain several dozen parameters that represent sub-grid processes and their output doesn’t include the large uncertainty arising from uncertainty in these parameters. (See perturbed-physics ensembles.) So I personally don’t trust the IPCC’s models to disclose the full range of possible pasts and futures that are consistent with our understanding of climate physics.

    As for “K.a.r.S.t.e.N’s” comment, I don’t recognize her name. I’ve seen others comment under the name “Frank” (a poor choice on my part), so she may or may not know me. My comments elsewhere have been technical, but don’t consistently support one point of view. I commented here because I thought the potential role of unforced natural variability had been underestimated.

  88. BBD says:

    And then there’s the end-Permian extinction. Or just hyperthermals in general, which are associated with CIEs. Arguments for low S to GHG forcing are contradicted by paleoclimate behaviour.

  89. It’s nice to see that we can agree with my second comment showing that natural changes in the Earth’s temperatures have occurred, which means that climate sensitivity can’t be exactly zero. Have you noticed that many WUWT readers don’t seem to understand this point? For some reason they seem to think that natural variability affects climate scientists like Randian garlic affects communist vampires. Odd, that.

    Hopefully we can agree that even though it’s difficult attribute a house fire in 1910 to a natural cause or arsonist, that doesn’t mean modern technology can’t help attribute modern house fires to natural causes or arsonists.

    Surely the words “recent unexpected hiatus in surface warming” are unintentionally jammed together, because if they were intended literally then my first comment seems to warrant a response. In it, I even linked to my open source R code, just in case real skeptics want to learn how to calculate uncertainties in the presence of correlated noise.

  90. BBD says:

    And as DS has pointed out, millennial climate variability would be impossible in a climate system insensitive to radiative perturbation. So unless we reject the current understanding of atmospheric physics, there is a problem.

  91. I just remembered that most contrarians seem to have problems with their computers not letting them click on links. I’m sorry for this oversight; here’s the code without any need to click.

    I calculated error bars on UAH trends. The black line on the second page shows the UAH trend ending in 2012, for different starting years. The error bars are shown in red; they’re 95% confidence uncertainty bounds. Note that error bars on longer trends are smaller than the large error bars on shorter trends.

    Anyone can reproduce my results by downloading the free “R” programming language used by professional statisticians. Then save this code as “significance.r”:

    # run using R CMD BATCH significance.r
    # outputs to Rplots.pdf and significance.r.Rout
    # load custom functions

    # for generalised least squares

    # options

    # read basin data
    indata = read.table(“greenland2013/GIS_climate.nasa.txt”,header=T)
    title=”Greenland mass”
    #indata = indata[which(indata$x>2002.0),]

    # remove mean
    indata$y = indata$y – mean(indata$y)

    n = length(indata$x)


    # fit model
    slope = fitsummmary$tTable[2,1]
    slopeerror = 2*fitsummmary$tTable[2,2]#2 sigma
    lowerbound=lowerbound – mean(lowerbound) + mean(fit$fit)
    upperbound=upperbound – mean(upperbound) + mean(fit$fit)
    text(midpoint,top,sprintf(“%+.3f+-%.3f %s/%s”,slope,slopeerror,yunits,xunits),cex=2,col=colfit)

    Just download temperature data (from WoodForTrees, Skeptical Science, Cowtan and Way, or any other climate data source). Then redirect the read.table command to that file, and save the data in this format:

    x y
    2003.04 1184.10
    2003.12 1006.97

    Then run it using the command “R CMD BATCH significance.r”

    Notes: The second page here uncommented the “which” command and used a for-loop to cycle through different starting years. Here’s another R script that automatically downloads the latest HadCRUT4 annual data.

  92. Tom Curtis says:


    “This reference demonstrates that warming prior to 1950 represents natural variability (including solar), not anthropogenic forced change.”

    “This paper:

    “Using an optimal fingerprinting technique we robustly detect a significant natural contribution to the early 20th century warming. In addition, the amplitude of our simulated natural signal is consistent with the observations. Over the same period, however, we could not detect a greenhouse gas signal in the observed surface temperature in the presence of the external natural forcings. Hence our analysis suggests that external natural factors caused more warming in the early 20th century than anthropogenic factors.”

    More warming from natural factors does not equal a trend caused solely by natural factors. Therefore it cannot, by itself quantify the potential variability from natural factors. You are taking a fact that everybody has acknowledged and treating it as if it somehow refutes the other position while neglecting the fact that the “natural trend” in the early twentieth century is almost certainly partly anthropogenic.

    You proceed similarly in your other criticisms, including a straw man characterization of the main stream position as that the exact forcings and causes of temperature changes over the past 1000 years are “settled science”.

  93. Reich.Eschhaus says:


    “The advantage of downplaying our limited understanding of natural variability (unforced variability) is that it allows scientists to make confident statements about man’s role in recent climate change. The disadvantage arrived with the recognition that there has been a recent unexpected hiatus in surface warming.”

    conspiracy thinking. Seen that before. Nothing new. Yawn! Everybody please ignore. DNFTT.

  94. Tom Curtis says:

    This howler is so bad I had to treat it separately.


    “The advantage of downplaying our limited understanding of natural variability (unforced variability) is that it allows scientists to make confident statements about man’s role in recent climate change.”

    Natural variability does not equal “unforced variability”. Conflating the two is a fundamental mistake which shows the person conflating them is at best ignorant on the topic they purport to discuss. Variations in solar forcing, ie, changes in the TSI, are entirely natural, but they are not unforced. Likewise, changes in the stratospheric optical thickness are (almost) entirely natural, being the result of volcanic eruptions. They are a forcing, and the resulting temperature changes are not unforced.

    In contrast, changes in the walker circulation due to changes in relative temperatures of the central tropical Pacific, and the Pacific Warm Pool (the driver of ENSO) are unforced, but they are also natural. This is probably also true of the PDO, and potentially also true of the AMO.

    Conflating forced and unforced natural variations is not science. It is gobbledygook.

    Franks argument proceeds thus:
    1) There have been certain large (primarily forced) natural variations in temperature in the past;
    2) Unforced variations in temperature are not well characterized by climate science;
    3) From (1) it follows that these unforced variations can be large;
    4) Therefore we can ignore the fact that natural forcings are known to be net negative from approximately 1951 and assume the warming since then may have been primarily due to unforced variations.

    Absent the conflation of “natural” with “unforced” it is a transparent non-sequitor.

    Absent clear acknowledgement by Frank that natural and unforced variations are not the same thing and that forced natural variations in the late twentieth century are at best neutral and more probably negative, it is plain that he is not here to discuss science but to push pseudo-science.

  95. Perhaps as addendum some sort of disclaimer: Given that I can’t check who Frank really is, I also can’t know for sure that it is the same Frank who commented here. In that particular case, I know for sure it’s him. The fact that the “new” Frank raised the same old talking points we used to be discussing, I might very well have jumped to unwarranted conclusions. Knowing that the Frank I know is a honest person, I have no reason to mistrust what “our” Frank is saying here. So I humbly apologize for sowing confusion. My bad! Frank, also apologies to you directly. I’m sorry! It doesn’t make your current statements more plausible, but there might be a lot more hope after all.

  96. Reich, if my earlier comment clears moderation then we’ll all be able to test Frank’s claim using open source code. You’ve previously asserted that contrarians know more than they let on, but I’ll defend Hanlon’s razor and the information deficit model to the dumb, naive, non-psychologist death. I refuse to believe that anyone who truly groks the Great Dying and the rate limits on adaptation via migration or evolution could keep spreading civilization-paralyzing misinformation.

    I suspect that Morton’s demon is far stronger than most people realize. For example, even Morton himself was later consumed by this demon in such a depressing way that I won’t link it.

    But I wouldn’t bother writing comments at WUWT because, having been banned once, it seems like my comments would never even appear in the first place.

  97. Reich.Eschhaus says:


    “You’ve previously asserted that contrarians know more than they let on, but I’ll defend Hanlon’s razor and the information deficit model to the dumb, naive, non-psychologist death. ”

    I would say that there are some that ‘know more than they let on’, and that there are others who follow the line in awe.

  98. I don’t know what that means, but I refuse to believe that some know more than they let on. Considering the stakes involved, that hypothetical informed contrarian (who I don’t believe exists) would have betrayed humanity. Even arsonists usually have a personal escape route, but knowingly spreading civilization-paralyzing misinformation has no plausible escape route. From my moral and pragmatic perspectives, the information deficit model seems to be correct.

    Even as their numbers dwindle, I’ll keep defending the morality of contrarians. There’s no shame in being insufficiently informed about a complex scientific topic, as long as one eventually stops spreading misinformation that threatens the future of our civilization.

    There are more enjoyable hobbies. Hobbies that don’t stain one’s legacy. Video games, reading, scuba diving, etc.

  99. OPatrick says:

    Psycopathy, or antisocial personality disorder, may be a factor in at least some ‘contrarian’ activity, I believe. This isn’t as dramatic as it sounds. Of course the Internet encourages and amplifies any such traits, so on-line personas may bear no relation to actual personality.

  100. Pingback: Putting the atomic bombs into perspective | And Then There's Physics

  101. Frank says:

    In my comments above, I haven’t always properly distinguished between the two types of natural variation: Natural forced variability – solar variability and volcanos, which we monitor accurately today and estimate from proxies in the past. Unforced variability – caused by chaotic or oscillator changes in ocean and air currents (which effect cloud cover and thereby albedo). ENSO, PDO, AMO, AO, the current pause and the 1940’s warm period are examples of unforced variability. Except for a few years after volcanic eruptions, forced natural variability has played a small role in the 20th century. If the LIA was caused by reduced output from the sun, that would be natural variability playing a much larger role.

  102. It’s so adorable when contrarians keep talking about a “current pause” despite the fact that there’s been no statistically significant change in the surface warming rate. If only I’d posted open source code so that actual skeptics could check their claims before spreading misinformation that stains their legacies and threatens the future of our civilization.

  103. Reich.Eschhaus says:

    “I don’t know what that means, but I refuse to believe that some know more than they let on.”

    I would say some do. It is an intent accelerated distribution of talking memes that come from somewhere. I am quite sure dishonesty is in play.

    I do not accuse anybody else

  104. Tom Curtis says:

    Frank, forced natural variability has not played a small role in twentieth century temperatures. There was marked increase in TSI from the early twentieth century to the end twentieth century, although the total forcing change is not large. More importantly, there was a very marked reduction in volcanism between 1918 and 1960 (see faq 5.1, figure 1). The result is a forced natural increase in GMST in the early twentieth century that is about equal to, or a bit stronger than the anthropogenic forcing over that period. In the late twentieth century, much of the natural forcing reversed itself. So while natural forcings have a minimum influence on centenial trends, and nearly none post 1950, they are a significant influence on the pattern of warming over the twentieth century.

  105. Rachel says:

    All written submissions to this are available online at http://www.parliament.uk.

    I just read the submission Myles Allen made and it sounded sensible to me. I hadn’t read the questions before and some are quite strange like how strong the economic argument for action on climate change is. Myles Allen points out that this is an odd thing to ask of a report on the physical science of climate change.

  106. Frank says:

    Tom: Thanks for the link to FAQ 5.1. As I read the data, significantly lower solar output (-1 W/m2) in the early 20th century has been reported by some authors and not by others. The text says (my emphasis): “A long-term increasing trend in solar activity in the early 20th century MAY have augmented the warming recorded during this interval, TOGETHER with internal variability, greenhouse gas increases and a hiatus in volcanism.” The same thing is true for the -1 W/m2 solar forcing our host mentioned above. In fact, it is difficult to measure solar output accurately from space. See the raw data here: http://sohowww.nascom.nasa.gov/hotshots/1999_12_20/SolIrradG.gif

    You are correct, a hiatus in volcanic activity did begun about the same time as the 1920-1940 warming I cited. Volcanic activity remained low until Agung (1963); much longer than the warming. That doesn’t mean much; the temp could have stabilized at a higher point in response to fewer volcanic aerosols. Treating each factor in isolation, however, means were don’t see the big picture from all factors combined.

    This figure from Roy Spenser (sorry about the biased source) indicates that none of the factors incorporated into the IPCC’s models (solar, volcanos, GHGs and aerosols) explain the 1920-1040 warming. It looks like unforced variability. The same presumably would be true of the current hiatus in warming.

    http://www.drroyspencer.com/2010/01/evidence-for-natural-climate-cycles-in-the-ipcc-climate-models-20th-century-temperature-reconstructions/ suggests that neither solar not volcanos explain the 1920-1040 warming. It looks like unforced variability.

  107. It looks like unforced variability.

    Quite possibly, I think. I believe one of the explanations for the recent slowdown is internal variability. I’m not sure how that influences our estimates of the long-term warming trend. It really shouldn’t.

  108. Tom Curtis says:

    Frank, Roy Spencer’s graph is a good one, except that it uses HadCRUT3 rather than HadCRUT4 (due to being nearly three years old). HadCRUT4 better accounted for the changes in different ways of measuring SST, with the result of a significant reduction in the peak in temperatures starting around 1940. The qualitative impact remains the same. That qualitative impact shows the multi-model trend accounting almost exactly (and certainly within error) for the observed trend from 1915 to 1935. That leaves two episodes unaccounted for by known forcings, ie, the trough from 1905-1915, and the peak from 1935-1945. These may well be accounted for by IV. They may also be accounted for as instrumental artifacts due to variation in coverage during the two world wars. They may also be accounted for by inadequate treatment of some forcing (particularly black carbon forcing with respect to the peak corresponding with WW2). So, this graph does not provide compelling evidence for a major factor in recent warming from IV. Rather, it suggests that significant effects from IV are short term at best, accounting for occasionally larger, but short term, perturbations from the forced temperature response than is typical from year to year.

  109. Pingback: Attribution | And Then There's Physics

  110. Pingback: Some comments on Judith Curry’s EPW testimony | And Then There's Physics

  111. Pingback: I also don’t get Judith’s logic | And Then There's Physics

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 )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.