RealClimate logo

How not to attribute climate change

Filed under: — rasmus @ 10 October 2006

In an earlier post, we discussed a review article by Frohlich et al. on solar activity and its relationship with our climate. We thought that paper was quite sound. This September saw a new article in the Geophysical Research Letters with the title «Phenomenological solar signature in 400 years of reconstructed Northern Hemisphere temperature record» by Scafetta & West (henceforth referred to as SW). This article has now been cited by US Senator James Inhofe in a senate hearing that took place on 25 September 2006 . SW find that solar forcing accounts for ~50% of 20C warming, but this conclusion relies on some rather primitive correlations and is sensitive to assumptions (see recent post by Gavin on attribution). We said before that peer review is a necessary but not sufficient condition. So what wrong with it…?

Recent Climate Change attribution (from Wikipedia)

The greatest flaw, I think, lies in how their novel scale-by-scale transfer sensitivity model (they call it “SbS-TCSM”) is constructed. Coefficients, that they call transfer functions, are estimated by taking the difference between the mean temperature of the 18th and 17th centuries, and then dividing this by the difference in the averages of the total solar irradiances for the corresponding centuries. Thus:

Z = [ T(18th C.) – T(17th C.) ] / [ I(18th C.) – I(17th C.) ]

Here T(.) is the temperature average for the century while I(.) is the irradiance average. If the two terms, I(18th C.) & I(17th C.), in the denominator have very similar values, then the problem is ill-conditioned: small variations in the input values lead to large changes in the answers; which implies very large
error bounds. In my physics undergraduate course, we learned that one should stay away from analyses based on the difference between two large but almost equal numbers, especially when their accuracy is not exceptional. And using differences of two large and similar figures in a denominator is asking for trouble.

So when SW repeated the exercise for the differences between the 19th and 17th centuries, and for three different estimates of the total solar irradiance, the results gave a wide range of different values for the transfer functions: from 0.20 to 0.57! The problem is really that SW assume that all climatic fluctuations in the 17th to the 19th centuries to solar activity, and hence neglect factors (natural forcings) such as landscape changes (that the North America and Europe underwent large-scale de-forestation), volcanism (see IPCC TAR Fig 6-8), and internal variations due to chaotic dynamics. It is, however, possible to select two intervals over which the average total solar irradiance is the same but not so for the temperature. When the difference in the denominator of their equation is small (the changes in the total solar irradiance are small), then the model blows up because other factors also affect the temperature (i.e. the difference in temperature is not zero). Thus their model is likely to exaggerate the importance the solar activity.

To show that the equation is close to blowing up (being “ill-defined”) their exercise can be repeated for the differences between 19th and 18th centuries (which was not done in the SW paper). A simple calculation for the 19th and 18th centuries is quickly and easily done using results from their table 1 and figures 1-2: A back-of-the envelope calculation based on the 19th and 18th centuries suggests that the transfer functions now would yield an increase of almost 1K for the period 1900-2000, most of which should have been realized by 1980! One problem seems to be that now the reconstruction based on solar activity increases faster than the actual temperature proxy. That would be difficult to explain physically (without invoking a negative forcing).

Modified SW2006b figure 2

The SW paper does discuss effects of changes in land-use, but only to argue that the recently observed warming in the Northern Hemisphere may be over-estimated due to e.g. heat-island effects. SW fails to mention effects that may counter-act warming trends, such as irrigation, better shielding of the thermometers, and increased aerosol loadings, in addition to forgetting the fact that forests were cut down on a large scale in both Europe and North America in the earlier centuries. Another weakness is that the SW analysis relies on just one paleoclimatic temperature reconstruction, but using other reconstructions is likely to yield other results.

Looking at the SW curves in more detail (their Fig. 2), one of the most pronounced changes in their solar-based temperature predictions is a cooling at the beginning of the record (before 1650), but a corresponding drop is not seen in the temperature curve before 1650. It is of roughly similar magnitude as the increase between 1900 and 1950, but it is not discussed in the paper. As in their earlier papers, the solar-based reconstructions are not in phase with the proxy data. However, SW argue that by using different data for the solar irradiance, the peaks in 1940 (SW claim it is in 1950) and 1960 would be in better agreement. So why not show it then? Why use lesser data?

The curves in Figure 2 (Fig. 2 here shows the essential details of their figure) of the SW paper suggests that their reconstruction increases from -0.4 to 0K between 1900 and 2000, whereas the the proxy data for the temperature from Moberg et al. (2005) changes from -0.4 to more than +0.6K (by rough eye-balling). One statement made both in the abstract of the SW paper and the Discussion and Conclusions (and cited in the senate hearing) is that «the sun might have contributed to approximately 50% of the total global surface since 1900 [Scafetta and West, 2006 – an earlier paper this year])». But the figure in the SW paper would suggest at the most 40%! So why quote another figure? The older Scafetta and West (2006) paper which they cite is discussed here (also published in Geophysical Research Letters), and I’m not convinced that the figures from that paper are correct either.

There are some reasons to think that solar activity may have played some role in the past (at least before 1940), but I must admit, I’m far from convinced by this paper because of the method adopted. It is no coincidende why regression is a more widely used approach, especially in cases where many factors may play a role. The proper way to address this question, I think, would be to identify all the physical relationships, and if possible set up the equation with right dimensions and with all appropriate non-linear terms, and then apply a regression analysis (eg. used in “finger print” methods). Last week, we discussed the importance of a physical model in making attributions because statistical correlations are incapable of distinguishing between forcings with similar trends. Here is an example of a paper that has exactly that problem.

There is also a new paper out on the relationship between galactic cosmic rays and low clouds by Svensmark. We will write a post on this paper shortly.

210 Responses to “How not to attribute climate change”

  1. 51
    Nicola Scafetta, PhD says:

    dear Urs Neu,

    if you have doubts, the best thing is that you repeat the calculations by using the following roles:

    1) use the data as they are, you cannot apriori assume some data are wrong.
    2) the climate sensitivity to solar changes should be calculated by going as far as possible in the past by using as much as possible data.
    3) it is better to use an algorithm that covers the three centuries before the industrialization (before 1900) but in a way that the 19th century is only partialy covered.

    I would also suggest that everybody reads very carefully a paper before criticizing it. Many doubts can be easily solved by simply reading carefully and trying to understand what an author has written and how he is reasoning. Sometime it is also necessary to read carefully the references to understand why an author is writing what he writes. In general, it is better not to be arrogant in science.

    Is it possible, for example, that nobody here has noticed the long comment from #21?

    Ferdinand Engelbeen has read my paper very carefully and he has immediately understood that the comments of Rasmus are wrong.

    Why doesn’t Rasmus want to unswer Dr. Engelbeen’s critique to his own criticism?

    Dr. Engelbeen has also correctly noticed that the volcano model predicted signals are too large when compared with the surface temperature data. If you look at Figure 5B in the recent paper by Foukal et al. on Nature, where they claim the Sun does not have any effect on climate, you will be able to see huge volcano spikes that really do not have anything to do with the temperature data. This suggests that the IPCC models, these authors have used, might be badly wrong because also the other forcings might be mistakly modelled.

    Why doesn’t Rasmus write a nice criticism on this point?
    Urs Neu, any comment on this?

    [Response: I would suggest that you heed your own advice and read the referenced papers carefully. Foukal et al did not claim to show that solar forcing has no effect on climate, merely that there is no positive evidence for a longer term forcing larger than seen over the 11 year cycle. And just to show that we are fair about criticising work, the averaging of different paleo-reconstructions in their Fig 5A is a very odd procedure and not one I would approve of. You should also be aware that there are different volcanic resconstructions around (Crowley, Amman) and it makes some difference which ones you use. In particular, estimates for the really large eruptions (i.e. Tambora) are very uncertain. For the period from the Maunder minimum to a century later though (which is the period we looked at), there are no obvious discrepencies between the solar+volcanic forced changes and the reconstructions. Assuming that the change is all solar would lead to a doubling of the solar impact, thus including long term volcanic effects is important. They might well be uncertain – but that is not a reason to ignore them. -gavin]

  2. 52
    Nicola Scafetta, PhD says:


    About Foukal et al., they have simply ignored a lot of litterature that says otherwise. It is true that the amplitude of the slow variation of the solar activity are uncertain, that is why I used 3 different TSI reconstructions in my paper (even if Rasmus has missed this fact). However, this uncertenty does not imply that the slow solar component is missing. Only that its amplitudes are uncertain, that is all.
    They have not even cited the ACRIM satellite reconstruction that shows a significant slow variation of the temperature since 1978 (They have shown only the PMOD composite, and I can ensure you those people know quite well about the existence of the ACRIM composite).

    You write:”For the period from the Maunder minimum to a century later though (which is the period we looked at), there are no obvious discrepencies between the solar+volcanic forced changes and the reconstructions.”

    I am sorry but you should look more carefully figure 5B in Foukal et al.. You will see that around 1810 the two models predict a huge cold spike with a cooling at -0.4K. The average temperature is around 0.0K, and no temperature reconstruction shown in Fugure 5A shows a cooling at -0.4K, some of them go to -0.1K. So, there is at least a -0.3K error that is sufficiently big to say the IPCC model used by Foukal et al. is wrong.

    Also the other volcano spikes at 1175, 1250, 1450, 1960 are not recovered at all by any temperature reconstruction. In 1250 the difference between temperature and model recontruction is as large as -0.6K!!!!

    I just wanted to show that it is very easy to criticize traditional climate model studies, that is all.

    About my calculations I had to compromize among many things.
    One thing is that from 1600 to 1900 there might be some little cooling because of volcano and deforestation but on the other side there was a little warming because of antropogenic added GHG. So, because climate models are not perfect yet I cannot use them to predict these minor effects and I supposed the two things compensate each other from 1600 to 1900, and just with this hypothesis I found a good correspondence between solar signal and temperature data. I do not claim my calculation are perfect, but my findings seem to reproduce temperature patterns much better than model simulations.

    [Response: Thanks. But read again. I said Maunder minimum to a century later, and in our papers we used the periods 1680-1700 and 1780-1800 precisely to avoid the Tambora uncertainty (which you highlight). I am a little bothered by the description of the Foukal analysis as being with an ‘IPCC’ model. What does that mean? IPCC doesn’t develop models nor does it do original research. In any case, the ‘model’ being used in Foukal is simply an energy balance model of the most basic type. Certainly nothing like as sophisticated as the GCM models being used in the detection and attribution sections of IPCC AR4 for instance. However, the response to Tambora (1815) is completely a function of the specified forcing, which is indeed uncertain. It doesn’t prove the model wrong – though it certainly highlights an inconsistency – most likely with the forcing though. As to whether you’re results fit better than ‘tradiational’ analysis, I’m afraid I must disagree. Crowley (2000) does a very good job (again with an EBM though), as does Gerber et al (2003). Full GCMs are only now being run over the these periods, but the biggest problems are in specifying the forcings prior to the 20th Century. During the 20th Century of course, the GCMs do a much better job (as seen in the figure above), or in for instance, and they have attributions to solar of much much less than you suggest (but see my previous post on attribution) for details on that. – gavin]

    [Response:Just for the record, Nicola, I didn’t miss the fact that you used three different TSI reconstructions. It’s pretty clear from your Figure 1b. This aspect is not the so central to my criticism, although the fact that these different estimates gave such large spread in the coefficient estimates is a bit concerning (don’t you think? so wouldn’t that be a good reason for exploring different temperature proxies too? I didn’t miss that you relied on just one temperature reconstruction too, which perhaps would give you so crazy results (?) that you would realise your method is not sound) . I think you are trying to jump too quickly to conclusions before you have all the facts. -rasmus]

  3. 53
    Mark A. York says:

    New Hampshire aquires Tennessee’s current climate by 2070.

  4. 54
    Mark A. York says:

    As the biologist and journalist in the fray here, I’ve found it’s all in the forcings. Trying to deny the greater effect of increasing CO2 is a fool’s errand, but there are many complex attempts I’ll give them that much.

    One thing puzzles me though, and that’s how can one get a Ph.D in physics and apply this level of falty logic? I don’t mean to be abusive which, frankly happens to me all too often for my liking. I just don’t get the logic path here outside of blatant denialism such as Milloy uses.

  5. 55
    Urs Neu says:

    Dear Nicola Scafetta

    Neither you nor the comments of F. Engelbeen (Nr. 21) do answer my questions. I�m sorry, my doubts are not solved by reading your paper very carefully. Just pointing to your paper does not help to answer a question which is not covered in the paper.

    It was you, not me, who said in comment 23: The sun-spot measurements from 1611 and 1650 are probably quite poor, so the TSI reconstructions during that time are poor because poor are the data.
    By this argument you tried to explain that TSI does not match the temperature data before 1650. If it is o.k. to use this data in your study, why isn�t it o.k. to compare the TSI and the temperature data?

    You say that the 19th century should only partially be covered. However, you calculate the difference between the 17th and the full 19th century and you take the unweighted mean between the two periods, thus you fully use the 19th century. But if you really want to use the 19th century only partially you should use the period 1750-1850.

    Also if reading very carefully, your paper does not explain what is wrong with using the periods 1650-1750 and 1750-1850, nor do the comments of F. Engelbeen. You also could use 1600-1700 instead of 1650-1750, according to your rules, the result is the same. Thus with the periods 1600-1700 and 1750-1850 all your rules are fulfilled:
    – Use of the data as they are
    – Going as far back as possible
    – Use the 19th century only partially
    – The relative error is similar to your periods, so this is no reason not to choose it.
    The only reason not to choose this period is that the results are different, that means the solar signal does only explain half of the longterm trend during the preindustrial period (which astonishingly well matches the results of Gavin).
    However, if the match of the solar signal to the temperature data is used as a selection criterion in your study (as you stated), you cannot, of course, present this match as a result of your study. If you have to cherry-pick your data period to get to your result, your method is not robust and the result is not meaningful at all.

    And a last comment on the TSI satellite data after 1978:
    You are ignoring two things:
    1. There is a third independent composite by Dewitte et al. (Dewitte et al., 2005: Measurement and uncertainty of the long-term total solar irradiance trend. Solar Physics, 224, 209-216) which shows a similar result as the PMOD reconstruction (no significant trend of TSI since 1978).
    2. there are two reasons why the trend in the ACRIM composite (by Willson and Mordinov) likely is an artifact:
    a) the positive trend is not due to a long-term increase, but the result of a short episode of increase (1989-1992) found in the data of one satellite (Nimbus 7). This increase has not been measured by the other satellite measuring at this period (ERBS);
    b) other indicators of solar activity, which are closely correlated to TSI (sunspot number, faculae, geomagnetic activity) show no trend in that period, either.

    If there are two composites with a similar result (no trend) and a third with only a probably artificial trend, it is much more likely that there is no trend in TSI since 1978.

  6. 56


    Getting a PhD does not change you from an ordinary person into a demi-god, who can spout the truth effortlessly. You still retain the same old prejudices.

    One of those, which is common to all people including myself, is that disaster is not just around the corner. We all believe that people who predict it are just Chicken Littles. That is why there is no progress being made in curbing greenhouse gases. And that is why the skeptics have been so successful. Not because they are telling people what they want to hear, but because they are telling people what they already believe!

    Most of what appears in scientific journals is produced and peer reviewed by PhDs, but that is not science. Science, which has been so successsful in replacing religion as the standard paradigm, is the distillation of those papers in journals which have been proved true. Science is true by definition, but papers in journals are just the thoughts of people. The history of science is littered with cases of mistaken beliefs, but these play no part in the science that is now accepted. It is also littered with breakthroughs that were not believed at the time and are now standard science.

    Mankind and scientists have not changed, and mistaken beliefs are still widely held by scientists today. The Younger Dryas being caused by the outflow from Lake Agassiz stopping the THC is probably the best example of that, since it is now being recognised as false but most earth scientists still adhere to it.

  7. 57
    Nicola Scafetta, PhD says:

    Reply to Gavin and Urs,

    (Rasmus continues to not understand that the reason why I used the Moberg temperature signal is that Moberg data are the novelty and that in a letter I cannot write too many things! Of course by using Mann and Jones’ hockey stick record I would get different results, but recently this record has been seriously criticized.)

    Gavin, your comments just confirm that the model have serious problems that are in part due to uncertainty in the data. This is a huge huge problem, because a model cannot be really tested if the data are uncertain!!! Moreover, the model should be tested on a long period of time, several centuries, to be really significant; when this is done the models give results that are at odd with the data as Foukal’s model show. Finally, there are different secular temperature reconstructions, Mann and Jones’ reconstruction is very different from Moderg’s one. If a model fit Mann’s reconstruction, it will not fit Moberg’s one. So, if your model looks ok with Mann’s data, if Moberg’s data are correct your model is badly wrong (and viceversa)!!!!

    About IPCC, Foukal el al writes: “We use the upwelling-diffusion energy balance climate model used by IPCC(Ref 63-67) for these model simulations”. So I was referring to this.
    Moreover, paradoxically Foukal et al. show that by using a low-frequency component of their model gives better results. So the logical conclusion is that the literature that they reference claiming that there in no variation in solar output and/or that the models predict a low sensitivity to solar change is likely wrong!!!

    About the paper by Hansen you reference, you have to look carefully their pictures. In Fig 6 you will find that their simulations simply cross the data. For example they do not recover the warming during 1940 and the volcano signals such as during 1961, 1982, 1992 are clearly overestimated, the temperature record does not present those deep spikes. Moreover, they are partitioning the forcings in a way that would underestimate the solar impact on climate. For example they are using the measured CO2 as an external forcing, without considering that CO2 concentration is also the results of several carbon cycles natural mechanisms that might in part respond to solar variation too. So, part of the solar signal might be embedded in what Hansen consider “other forcings” such as GHG forcing. These things are explained in my paper. Just, read it carefully.

    [Response: Nicola, with all due respect, take a step back here. Your arguments will be more persuasive if you focus on one aspect at a time. However, are you seriously arguing that the increase in CO2 over the 20th Century is a feedback to solar forcing? If so, I would suggest you work out what the rough temperature-CO2 relationship is over glacial to interglacial time scales (Petit et al etc) (or even over the last 1000 years – Gerber et al, 2003) and estimate how big an effect a ~1 deg C rise would have on CO2. Then compare that with the known emissions, carbon isotope data and increases in CO2 in the ocean and biosphere. So putting that aside, your criticism of the Hansen paper (which I am a co-author on) is that the ensemble mean results show clearer volcanic peaks than observed. That is actually to be expected since the ensembles average over the ‘weather’ noise, and are therefore closer to the forcings. The real world is of course only one realisation, and if you look at individual realisations, say around 1963 (Mt. Agung), you don’t see any obvious disconnect – the cooling from Pinatubo (1991) is also well matched. ENSO is a confounding factor of course because the tropical variability in the model is not correlated with the observed record (and is too weak in any case), but in the real world was coincident with both El Chichon and Pinatubo. In any case, I will guarantee that the GISS model match to the observed out-performs any statistical or physical model that relies purely on natural forcings.

    Going back to Foukal, you misrepresent their position completely. They do not argue that ‘there is no variation in solar output’, nor that ‘models predict a small sensitivity to solar’. Our model is pretty much as sensitive to solar as it is to GHGs, and that is seen in most other cases too. Foukal point out correctly that the arguments used for supporting a long term solar component – other than that related to sunspot/faculae – have fallen down. In particular, the cycling/non-cycling sun-like stars argument used in Lean 1994 has not survived more comprehensive sampling of sun-like stars. That is not to say that there is no long-term component, just that there isn’t any evidence of it’s magnitude. It’s therefore a tricky thing to have to rely on. And now going back to your paper, the basic point is that single forcing attribution studies cannot distinguish between different forcings that have correlated behaviour. Thus when there is ‘constructive’ interference (like at the Maunder Minimum, or the 20th Century), any such method will over-attribute the response. Surely this is something we can all agree on? – gavin]

    About the comments from Urs,
    The climate sensitivity to solar variations cannot be ½ (as Urs claims) or 2 times (as Rasmus claims) larger that what I estimated for the simply reason that there would be no match between the data patterns.
    About ACRIM data, Urs should read carefully the relative papers. He would realize that the quality of Nimbus 7 data is much much higher than the Erbs one. And that Nimbus patterns during the ACRIM gap (1988-1992) do not fit the sunspot numbers but they do fit the magnetic record. This is why there is a controversy between the two groups.
    Moreover, ACRIM composite uses the data as they are published, while the other two composites alter the data with hypothetical models that might have large errors and might be wrong.

    I am sorry but I think that Foukal el al’s paper is very biased in the references they decided to use and dismiss all debates and, paradoxically, they arrive at the conclusion that the side of the debate they represent is likely wrong!!!

  8. 58
    Nicola Scafetta, PhD says:

    the things look more complex to me than what you say.

    1) Hansen’s model does not reproduce the warming around 1940.
    The warming from 1900 to 1940 occurred at an higher rate than Hansen’s model predicts. If you try to adjust a little bit the model sensitivity to CO2 forcing to better match the warming from 1900 to 1940, the model would not match the data from 1950 to 2006 any more.
    Just try to fit the temperature data and the simulations from 1900 to 1940 and compare the two slopes in your figure 6, you will see that there is a significant difference. [edit]

    [Response: Climate sensitivity is not an adjustable parameter (though it does vary over models). You could get a slightly better fit by playing with the forcings, solar or land use maybe, or it could be due to internal variability. How could you tell? – gavin]

    2) the model should include carbon cycle mechanisms and methan cycle mechanisms to be realistic, and it does not. The bacteria produce a lot of CO2 and CH4 and the ocean exchange a lot of gases with the athmosphere. These mechanisms are temperature dependent. Moreover, the water vapor feedbacks and cloud cover are not very well modelled yet.

    [Response: Over the 20th Century we know what CO2 and CH4 changes were so you don’t need to. Understanding the feedbacks is of course useful and should also be done (and it is underway), but a simple back of the envelope calculation demonstrates clearly they can not have had a significant effect over the 20th C compared to the rise due to industrial emissions. Water vapour feedbacks are very robust across different models – cloud feedbacks less so of course. See Brian Soden’s posting and paper for more details on that. ]

    3) I am not claiming that “all” CO2 increase comes from solar increase, but if only 5% of that increase is due to the sun increase, this component should be included in the sun contributions, not in the anthropogenic ones. The same should be done with the other forcings. The models are not capable to disintangle these things.

    [Response: Give me a quantitative estimate based on more than your feelings! That would be a good contribution, but just saying it’s plausible doesn’t make it significant (and if it’s even as large as 5% of the CO2 rise (~ 5ppm), it’s completely trivial in terms of the forcing – < 0.1 W/m2. ]

    4) the millenaria records of CO2 should be interpreted correctly, These data (Petit et al.) are moving averages on several centuries and large faster fluctuations would be cut off. So it is wrong to compare the Petit’s CO2 data with the measured athmospheric CO2 data of the last century

    [Response: Then look at the higher resolution data from Law Dome or Siple… ]

    5)About Foukal’s paper, you are correct. However, in my paper I have argued that if the long term of the solar variability falls down and the Moberg temperature data are correct, the actual models are very wrong because they will never be able to reproduce the millenaria cycle presented in the Moberg data without a strong climate sensitivity to solar cicle. Foukal too argue that the models might be missing a lot of sun-climate interactions.

    [Response: But this depends on a big ‘if’ – (if Moberg is correct). You cannot use that to argue that solar must be stronger – it is simply underdetermined (i.e. a dozen things could be tweaked to get a better agreement). But if we look at the 20th Century, we have much better data, and your claim of a 50% attribution to solar just doesn’t fit once you take into account all the other forcings. ]

    To respond to Rasmus’ [edit] criticism, I am simply showing that it is easy to criticize traditional model studies too.

    I am not accusing anybody. I am just saying that these things are complex, and that nobody can claim that everything is clear and already understood. So everybody (me first of all, but also you, Rasmus, Hansen etc.) should be humble in these things.
    Don’t you agree?

    [Response: Of course. All the more reason to be clear about errors and assumptions when giving the ‘headline’ numbers…. ]

    Moreover, the climate model’s guys, when they write new proposals, always argue that their models are imperfect and that they need more money to update them. Is it true, or isn’t? :)

  9. 59
    Nicola Scafetta, PhD says:

    thanks for your reply.

    However, I think that there are problems in your model.

    If we look at the 1900-1940 period the temperature increase rate is almost 0.011K/y, while your simulations seem to me to give a rate not higher than 0.0035K/y. This is one third smaller than observed. This is a significant difference, sufficient to claim the model is wrong and/or severely incomplete.

    Your model also shows an increase of the temperature from 1940 to 1960, while the temperature is decreasing during such a period. Also in 1890 there is again a large volcano cooling not seen in the data. So for 80 years (on 125 years) your model does not seem to reproduce the data well.

    Your model also uses the TSI data by Lean 2000. The most recent data (referenced by Foukal et al) shows a lower TSI variability. Thus by using the latest solar data your model will further lose the match with the temperature data, in particular during 1900-1950 when solar activity increases.

    You claim that you can adjust the things by adjusting solar and land change. The problem is that if you adjust the forcing, in particular the solar one, during this period you might not get the good fit during the second half of the century. So you might need to do a large correction to the model and adjust the CO2 forcing as well.

    Using the argument â??there might be some additional internal variabilityâ?? is convincing but does not help you, because you are just stating that the actual model is wrong and/or incomplete. GCM are supposed to reproduce internal variability, right?

    It is true that you can simply add CO2 and CH4 measured values as a forcing, but if you do so, you cannot interpret them as anthropogenic forcing because their concentration is modulated by CO2 and CH4 cycle mechanisms that might in part be modulated by the Sun variability. The CO2 and CH4 cycle mechanisms are not a little thing, IPCC 2001 chap. 3 claims that the GHG human emission rate is approximately 2 times larger than the observed rate during the last decades, so if the Sun drives a little bit the CO2 and CH4 cycle mechanisms (which in this case are absorbing large amount of CO2 and CH4 from the atmosphere) it might leave a signal in the CO2 and CH4 record as well.
    (I cannot estimate this amount; it is the model guys’ job to write a model with “all” mechanisms included and do the calculations and control that everything looks good on simulations covering several centuries ).

    I just suspect that the models are overestimating the anthropogenic contribution and that there is a larger solar effect. The reason is because the solar pattern seems to mimic quite well the pattern in the temperature for all 4 centuries I have analyzed. My simple reconstruction does not cross the data as your simulation seems to do from 1880 to 1960, but reproduces quite well the large patterns of the Moberg temperature for several centuries. This might be a coincidence, I agree, but it might also suggest that Moberg’s reconstruction might be a reasonable one and that the Sun is an important contribution to climate change.

    It is true that my findings are based on Moberg temperature reconstruction, but Mann’s reconstruction has been seriously discredited. The warm Middle Age period and the cool periods during the 16th and 17th centuries seem to be well established historical facts (Vikings could not live in Greenland and could not navigate in the northern Atlantic if there in the Middle age the temperature was low, even today Greenland temperature is too cold). This historical record is a strong evidence that would support Moberg’s reconstruction with a large millenarian cycle. Solar data seems to have such a millenarian cycle that could drive a large internal variability in the ocean circulation, for example. If so, the actual models are in serious trouble because none of them is able to reproduce this temperature patter, and they might need a much stronger climate sensitivity to solar cycle that might include a lot of things in addition to the simple TSI forcing.

    This is just a paragraph taken from Wikipedia about the medieval maximum and little ice age
    The Medieval Warm Period partially coincides in time with the peak in solar activity named the Medieval Maximum (AD 1100-1250).

    In Chesapeake Bay, Maryland, researchers found large temperature excursions during the Little Ice Age (~AD 1400-1850) and the Medieval Warm Period (~AD 800-1300) possibly related to changes in the strength of North Atlantic thermohaline circulation.[8] Sediments in Piermont Marsh of the lower Hudson Valley show a dry Medieval Warm period from AD 800-1300.[9]

    Prolonged droughts affected many parts of the western United States and especially eastern California and the western Great Basin.[7] Alaska experienced three time intervals of comparable warmth: 1-300, 850-1200, and post-1800 AD. [10]

    A radiocarbon-dated box core in the Sargasso Sea shows that sea surface temperature was approximately 1°C cooler than today approximately 400 years ago (the Little Ice Age) and 1700 years ago, and approximately 1°C warmer than today 1000 years ago (the Medieval Warm Period).[11].

    The climate in equatorial east Africa has alternated between drier than today, and relatively wet. The drier climate took place during the Medieval Warm Period (~AD 1000-1270).[12]

    An ice core from the eastern Bransfield Basin, Antarctic Peninsula, clearly identifies events of the Little Ice Age and Medieval Warm Period.[13] The core clearly shows a distinctly cold period about AD 1000-1100, nicely illustrating the fact that “MWP” is a moveable term, and that during the “warm” period there were, regionally, periods of both warmth and cold.

    Corals in the tropical Pacific ocean suggest that relatively cool, dry conditions may have persisted early in the millennium, consistent with a La Niña-like configuration of the ENSO patterns.[14] Although there is an extreme scarcity of data from Australia (for both the Medieval Warm Period and Little Ice Age) evidence from wave built shingle terraces for a permanently full Lake Eyre during the ninth and tenth centuries is consistent with this La Niña-like configuration, though of itself inadequate to show how lake levels varied from year to year or what climatic conditions elsewhere in Australia were like.

    Adhikari and Kumon (2001) in investigating sediments in Lake Nakatsuna in central Japan have verified there the existence of both the Medieval Warm period and the Little Ice Age.[15]


    Of couse, it is a “if” argument. But everything right now suggests that it might not be unreasonable. So, we should look at the data and solve their ambiguities first. The data will solve many of these problems. But until the secular data are ambigous, I do not think that it is safe to claim that the debate is over. don’t you agree?

    [Response: Now you are confusing me. We were talking about recent centuries, and you bring in a whole lot of random references to the Medieval Warm Period? Since neither your analysis, nor my model have directly assessed that, I will put that aside (though I would recommend reading Bradley et al (2003).). With respect to your assessment of our model simulations, you are advised to go directly to the source ( to calculate the trends of any particular period. For 1900-1940 the mean trend is 0.19 deg C/40 years, compared to 0.33 deg C/40 years in the observed data (land/ocean index). It’s definitely not an exact match, and examination of the regional patterns show distinct differences. Some part of the difference is undoubtedly due to the intrinsic variability which cannot be coherently captured in a climate model, while uncertainties in forcings, and model inadequacies probably also play an unquantified role. Your comments regarding the response to Krakatoa are less well supported – look at figure 7 and the associated discussion. Met stations clearly record a cooling over land associated with Krakatoa of the same magnitude as we predict – however the SST data do not show this. I would be more inclined to question the SST data in this case (since it is a reconstruction rather than pure observations). Look, I do not claim that climate models are perfect – far from it – but their matches to observed data at the large scale are impressive – Pinatubo, last 30 years, response to ENSO, NAO response, sea ice response, ozone hole response etc. They are, and will remain, the only way to quantitatively compare the myriad different influences and pathways in the climate system in physically consistent ways. Some of the forcings used are more certain than others. For instance, CO2 forcing can be calculated from highly accurate line-by-line codes and is implemented in GCMs with less than 10% error (it can’t be ‘adjusted’ in any signficant way) However, aerosol forcings are highly uncertain – our best guesses for those could well be significantly off – the same for solar. But when we take the best guesses (independently derived) for each of these items, we do get a reasonable, though not perfect, match to the obs. We don’t go back and adjust the forcings to ‘improve’ the match, since that would assume that the model was perfect (which we know it isn’t), so the comparison to the observed data is a valid test. And in that comparison, the model does a better job than any statistical model based on solar alone (or CO2 alone for that matter). Why don’t you suggest a test, and we’ll put your statistical model up against the GCM output and we’ll see who has the best match against the 20th C data. -gavin]

  10. 60
    Hank Roberts says:

    I notice the ozone holes are at the maximum — more UV light. Does that change climate sensitivity, by boosting plankton primary productivity and the plankton boosting ocean circulation and so gas exchange into the ocean, compared to any other time in the past when there was no ozone hole?

    Serious question, I’m still as an amateur wrestling with understanding what ‘climate sensitivity’ takes into account. I guess the question is, if all else was held the same — if we had our fossil fuel industry but had not invented the chlorofluorocarbons and equivalents so hadn’t lost so much of the ozone layer for so long — would that change climate sensitivity?

  11. 61
    rasmus says:

    The bottom line is: the method used in the paper is unsound because it neglects noise (influence from factors other than solar and GHG) and it may blow up if the temperature changes while there are unrelated changes in the temperature. -rasmus

  12. 62
    Nicola Scafetta, PhD says:

    Gavin, thanks again for the answer.

    Just a short reply.

    [edit – please no personal comments!]

    (About the comment by Hank, yes there are also the plankton and vegetation effects that are temperature and light sensitive that the models do not contain)

    Now a short reply to Gavin.

    I am talking of several centuries because that is what my paper addresses. If you read carefully my paper there I discussed too the 1000years case, even if very briefly.
    The warming that I associate to the solar activity since 1600 is approximately equal to the cooling from the Medieval Maximum to the 17th century. This cooling according to Moberg is about 0.6K and this could not be caused by humans. So, it must be natural and the sun is probably the major cause. Because there are some evidence that the sun in the middle age was as hot as today, this simply implies that the sun could induce another 0.6K since the little ice age in the 17th century.
    And this would confirm my calculations.

    [Response: False premise. You already assume that a) Moberg is correct, and b) no other factors are involved. We have already discussed evidence that volcanic activity could play a significant role, and absent better constraints on the forcings it is difficult to attribute this effect precisely – your assumption that it must all be solar, is an assumption, not a result. – gavin]

    The problem with the models is that all of them have an energy balance model inside, also the GISS one, The energy balance core is what drives the slow climate variability. The problem is that when the modern energy balance models, such as the Foukal’s one, are run on 1000 years they give a result that is compatible with the hockey stick reconstruction of Mann and Jones. They predict that the cooling from the medieval warming and the little ice age was approximately 0.2K. Moberg reconstruction suggests that such a cooling is 0.6/0.7K, that is three times larger: this is a lot. So, if Moberg is correct the core itself of all present climate model (that is their energy balance model), including the GISS one, is incorrect.

    [Response: The results from EBMs, and indeed GCMs, depend a great deal on the forcings. A mismatch between the paleo reconstruction and a model result can be due to a) an incorrect forcing, b) an incorrect reconstruction, or c) an incorrect climate sensitivity (or of course all three). There is independent evidence that the climate sensitivities are in the right ballpark, and that leaves the forcings or the reconstructions. Both have significant uncertainties and so absent other information you cannot conclude what the error is. Foukal may well be wrong about long-term solar impacts, Moberg may have over-estimated low frequency variability – my point is not that either one of these things is correct, but that neither you nor I can tell. Therefore conclusions that there is something wrong with energy balances models (which are just conservation of energy, which I doubt you want to challenge) are highly premature. ]

    Now, the problem is determining if Mann and Jones are right or if Moberg is right. [edit] On the contrary, there are overwhelming evidences from several independent studies that the middle age was quite warm and that the 16/17th centuries were quite cold, this would strongly support Moberg reconstruction and therefore the conclusion of above (and my estimates).

    [Response: You are grasping at straws. Medieval climate was much more complicated than simply ‘it was quite warm’ and it remains entirely unclear to what extent it was globally warm. It is completely possible that neither Mann and Jones nor Moberg are right – better data will help, but you must acknowledge the uncertainty there. And given that uncertainty, and the uncertainty in the forcings, the sensitivity cannot be usefully constrained from Medieval data. ]

    Now, let us go to your model.
    I thank you for having kindly acknowledged that the model is not “perfect” and that right now there is a serious discrepancy between the model prediction and the data from 1900 to 1960. This would suggest that the model has a serious problem. Be careful! Your model must match the data to be credible, while my simple analysis of the Solar effect alone does not need to match the data perfectly, in my paper I say that only approximately 50% of the warming since 1900 is related to the sun and from 1600 to 1900 the match is quite good, indeed!

    [Response: That is curious logic. An independent estimate matches the observed data better than your statistical model. That estimate has solar contributing less than 10% of the overall warming factors (0.3 W/m2 out of 3 or so). That is therefore evidence that your attribution of 50% is a better fit? How does that work? I should point out that your attribution makes the error I alluded to recently in not accounting for cooling effects – thus the warming forcings will add up to more than 100%. A better metric would be the ratio of GHG influences to solar and our modelling (as well as many others) demonstrate that ratio is much greater than 1. ]

    In any case, let us now “assume” that your model is perfect. My argument is that Hansen’s “interpretation” of his own model is badly wrong.

    The reason is that Hansen and many other model guys interpret the “measured” CO2 and CH4 concentration increase as “anthropogenic” forcing. This is in principle wrong because there are CO2 and CH4 natural cycle mechanisms that determine the concentration of these gases in the atmosphere and these mechanisms might be driven by the sun. The model do not contain these mechanisms, so the model does not have any way of determining how much of the observed CO2 and CH4 concentration come from humans and how much is a natural response to solar increase.

    Now you ask: can you give any estimate of the solar signal in the CH4 and CO2 record? I reply that I cannot because I need the CO2 and CH4 natural cycle mechanism models that I nor you do have. So the problem remains unsolved from a model point of view and you cannot used the findings of you model to criticize my findings which would take care of correclty considering the possible solar signal inside the CO2 and CH4 concentration records, that might be quite large.

    [Response: On the contrary, it is easy to constrain. Over very long time periods such that the carbon cycle is in equilibrium with the climate, one gets a sensitivity to global temperature of about 20 ppm CO2/ deg C, or 75 ppb CH4/deg C. On shorter timescales, the sensitivity for CO2 must be less (since there is no time for the deep ocean to come into balance), and variations over the last 1000 years or so (which are less than 10 ppm), indicate that even if Moberg is correct, the maximum sensitivity is around 15 ppm CO2/deg C. CH4 reacts faster, but even for short term excursions (such as the 8.2 kyr event) has a similar sensitivity. Now compare that to the unprecedented post-industrial rise: 100 ppm for CO2, over 1000 ppb for CH4. In both cases the potential for a temperature lead contribution is around 10% – way too small to effect the forcings substantially. The evidence that the current rises of both CO2 and CH4 are anthropogenic are overwhelming (from isotope data, O2 data, ocean data, emission inventories etc.). If your result relies on reducing the impact of CO2 and CH4 as forcings, then we might as well stop here. ]

  13. 63
    Urs Neu says:


    Your argumentation gets more and more contradictory.

    It is not me who claims that the sensitivity to solar variations is only half of what you say. It is your method with your rules that produces that result.

    And you reject this result only because it does not match your claim that solar variations are the only relevant forcing factor. This is not convincing at all. If you want to show that there is a match, you can not use this as a precondition. This is basic logic. You persistently ignore that point.

    You write to Gavin: … because the solar pattern seems to mimic quite well the pattern in the temperature for all 4 centuries I have analyzed.
    The solar pattern only mimics the temperature pattern so well because you reject all the other results which do not show this.
    However, I agree that on the multidecadal scale (Dalton Minimum etc.) both patterns are similar. But there isn’t any evidence at all that the long-term trend from the 17th to the 19th century can be solely explained by solar forcing. There are such large differences in the trends over these centuries both for temperature and for TSI (the only consistency is that the trends are all positive) that any attempt to analyse the influence from these data is hopeless. If you use only half the Zs value, the multidecadal pattern will still match, but not the long-term trend.

    You claim that the model Gavin used does not match exactly the temperature curve. That is only because Gavin included all the different results of the model (the ensemble) in his result. If he would do the same as you did in your study and exclude all the results that do not match, the match of the model would look much better (as he has explained). If you want to compare the model to your study, please include all the results and not just the results that match your hypothesis.

  14. 64
    Coby says:


    As an exercise and self test, I’ll take an amatuer’s stab at your question. The phrase “climate sensitivity” absent any particular context seems to generally mean the temperature change the climate will undergo on the short term given a doubling of atmospheric CO2 concentrations and all other forcings held constant, I will presume that is what you also mean.

    Calculations of the climate’s sensitivity to 2xCO2 includes the radiative effect of CO2, the radiative effect of H2O vapour that increases as a feedback and the albedo changes that result from loss of sea ice. It does not include potential long term feedbacks such as ice sheet melting or carbon cycle feedbacks.

    As such it is a theoretical property of the climate as a whole. Any change in the climate as a whole will therefore potentially impact the climate sensitivity value. For example, in a climate where there is no sea ice, this feedback would be absent and climate sensitivity would be less. An earth with continents in different locations or more or less land would also respond differently and have a different sensitivity value. Thus it is possible that any alteration of the atmosphere such as ozone depletion, will alter the climate’s sensitivity to CO2 doubling.

    That said, I don’t know if the climate sensitivty studies try to isolate and remove the other currently understood anthropogenic forcings to get a more natural or theoretically pure value, ie the value of 2xCO2 in an otherwise natural world.

    I rely on any of the forum’s experts to correct me where required.

  15. 65
    Nicola Scafetta, PhD says:

    I think that there is some problem of communication here.

    I am not starting from a false premise.

    I am explaining the meaning of my paper.

    In my paper I start with a hypothesis and I have a goal.

    This is the hypothesis and goal:
    “Let us use the Moberg’s data and let us deduce its main consequence.”

    So you cannot argue that I am starting from a false premise. I do not care if the premise is false or true. I just do that hypothesis, that is all. And I claim that in accordance with this hypothesis, we have to conclude that the climate sensitivity to solar change is likely much stronger than what present models claim.
    This is all.

    I have also no problem in acknowledging that if the Medieval warming did not exist, and if Mann and Jones data are closer to the truth, the sun might have had a little impact on climate and that present models well reproduce the Hockey Stick pattern these data show. So, in this case, the actual models do not need great corrections and are already quite good.
    This is all.

    However, I notice that the hypothesis I did is not totally unrealistic because today we have several evidences that the medieval period was quite hot and the little ice age what quite cold. And other reconstructions of the temperature suggest a large millenarian cycle more or less compatible with the Moberâ??s reconstruction. Perhaps, all this is wrong, or perhaps not. We do not know yet.
    This is all.

    Finally I attempt a suggestion that perhaps one solution to the problem that the solar impact on climate is underestimated by models might be because EBM and GCM, like GISS, do not contain CO2 and CH4 cycle mechanisms that might be partially effected by the Sun, and other mechanisms are missing or uncertain (water vapor, cloud cover, vegetation, bacteria respiration, UV radiation, cosmic ray effects etc.). When all these mechanisms will be included in the model, perhaps we will understand better climate and the causes of climate change.
    This is all.

    So, what is wrong with this?
    Do you want that I do not have to do a study by doing an innocent hypothesis and deduce its consequences?

    [Response:My greatest objection is your choice of methods – they are likely to give you spurious results, and therefore I do not trust the conclusions. The merit of any analysis or experiment depends on the objectivity of their setup, even if you were to start with a valid hypothesis. -rasmus]

  16. 66
    Nicola Scafetta, PhD says:

    I have a final comment and I ask you to give a fair response.

    I have claimed above that perhaps a solar signature is likely hidden in the GHG signature and in other climate forcing as well.

    Rasmus have done a nice thing above.
    Look carefully the data reported by Rasmus in his figure above.
    In particular look at the ozone and GHG signature. I do not know where Rasmus took these data from. But let us assume they are correct.

    The Ozone signature seem to have great oscillations with 5 maxima in 100 years. This well fit the 22-year solar magnetic cycle.

    The GHG signature is more complex but if you see before 1950 there might be four oscillations in 40 years. This well fit the 11-year solar cycle. (Note that there is a time lag of 5 years, this might be OK, because the climate does not respond instantaneously to a solar change).

    Perhaps, this is only an illusion, but if not and if Rasmus data are correct these pattern might suggest that there might be a serious possibility that there is a solar signal hidden in the GHG and Ozone record. As I think.

    Do you have any comment?
    Do these oscillations have some other reasonable explanation?

    [Response: The figure is from Wikipedia, based on results from Meehl et al (2004). However, your question illustrates exactly why one has to be very careful when looking at noisy data. These curves are the smoothed model output, and as such contain ‘weather’ noise that is uncorrelated to the forcings. While you might think you can see a 22 year cycle in the GHG and ozone runs, I am 99% sure that there was no 22 year cycle in the forcings. There certainly isn’t such a cycle in the GHG concentrations, and while it’s conceivable that the ozone runs used an ozone field that was modulated by solar activity, I doubt that this was the case (that kind of simulation is only just being to be made). The ozone forcing most probably only has a trend due to industrial activity affecting tropopsheric ozone (a warming), and the depletion of stratospheric ozone due to CFCs from the 1980s onwards (a slight cooling). Therefore, what you perceive as a cycle, is simply decadal ‘noise’ and would be different in another set of realisations. But please be sure to understand me – I am not claiming that there is no possibility of solar feedbacks affecting the other ‘forcings’ (in fact we are working on ozone feedbacks quite extensively – though they are affected by the irradiance changes on 11 year timescales, not on 22-year timescales). However, these potential feedbacks can be shown, by looking at the pre-industrial for instance, to be much smaller than the anthropogenic contributions over the 20th C. Prior to that they would have been relatively more important and potentially detectable in the record, but one has to be very careful about extrapolating from much more uncertain forcings/response in the paleo-climate record to the much better characterised changes in recent decades. – gavin]

  17. 67
    Nicola Scafetta, PhD says:

    Thanks Rasmus,
    everybody here has understood what is your objection.

    I accept your point but I respectfully disagree with your interpretation of my paper.

    I told you that you should look at the large picture, and not focus on some details that you might misunderstand.

    Moreover, I notice that I am not alone. Here there are people who had some problem with your way to read and criticize my paper.

    Again, I tell you to read comment #21.

  18. 68
    Nicola Scafetta, PhD says:

    Ok Gavin,
    thanks for your explanation.

  19. 69
    Nicola Scafetta, PhD says:

    Ok Gavin,
    just a last comment

    I still notice that also the Meehl model, whose result is in the above figure has some problem from 1900 to 1950.

    The increase rate of the model simulation is significantly lower (~1/2) than the increase rate of the data during the same period.

    Yes, i agree that the Ozone should have a 11-year cycle.

    Well, I think we have discussed enough, good work with the ozone feedback (and do nor forget the second generation water feedbacks to it.)

    thanks for the discussion.

  20. 70
    muller.charles says:

    Just a question, before Nicola and Gavin make mincemeat of each other :D. GHG forcing is global (“well-mixed” in atmosphere). Could we imagine a different and regional modulation of solar forcing for past decades and centuries ? For example, Milankovic global solar forcing is quite small, but latitudinal and seasonal variations imply strong feedbacks. On a shorter time scale (decennal, pluridecennal), could it occur with modest variation in TSI related to atmosphere-ocean circulation ? Or is it a non-sense ?

  21. 71

    General comment about the foregoing discussion:

    Any method to detect the real climate sensitivity for the four main climate drivers (GHGs, aerosols, solar, volcanic) is dependent of several assumptions and constraints.

    The first constraint is the temperature profile of the past century. Any model, regardles its sensitivity, must fit the past century’s profile. But the problem is that many sets of sensitivities (different sensitivities for different forcings) can fit the past century’s profile, because if one forcing or sensitivity (like solar) is underestimated, another one (like CO2) may be overestimated and need to be adjusted downward. Because there is an overlap between the warming induced by solar and by GHGs, it is impossible to know which one is responsible for what part of the warming. Btw. sensitivity for CO2 in current models (~3 K/2xCO2) is much higher than what can deduced from IR absorption alone (~0.85 K/2xCO2), this includes all feedbacks, of which some (like clouds) are very uncertain…

    The second constraint is for the pre-industrial part: we have only reconstructions, which are quite different in amplitude for the same time frame. Some show small variations (~0.2 K), the newer ones show larger variations (~0.8 K). In the case of smaller variations, the attribution between solar and volcanic is about 50:50 for the pre-industrial period, as human made GHGs and aerosols had little influence. If the newer reconstructions are taken into account, then the ratio between solar and volcanic influences increases to 7:1. Not taken into account that volcanic may be overestimated…

    In general, the larger amplitude reconstructions have a better match with independent bore hole and ice core temperature reconstructions. The latter show a ~10 ppmv CO2/K relationship over all glacials/interglacials, including a change of ~10 ppmv between the MWP and LIA, thus caused by ~1 K cooling. A similar temperature change may be expected between the LIA and current, be it that the expected ~10 ppmv CO2 change is overwhelmed by human emissions.

    This is what Moberg, Esper, Luterbacher and others have concluded: if the variability in the pre-industrial past was larger, then the sensitivity of the climate to man-made emissions must be lower than currently implemented in the models.

    [Response: This is nonsense. The sensitivity implied by a particular temperature history is not determined by the amplitude of that history. It is determined by the covariance between the temperature history and the estimated radiative forcing. A large amplitude that is uncorrelated with the forcing would indicate a sensitivity of zero. You might learn something from the appendix of this paper by Waple et al describing how to estimate sensitivity from forcings and their estimated responses. It is somewhat ironic that you cite both the Moberg et al and Esper et al temperature reconstructions in your efforts to argue for a greater role of solar forcing in past temperature variations. Indeed, they both have larger amplitudes than most other reconstructions as you say. But they also both happen to be essentially uncorrelated with each other at centennial timescales (see the Wikipedia plot), and Esper et al is actually negatively correlated with most reconstructions of Solar Irradiance. But perhaps you are advocating a negative sensitivity to solar forcing? Of course, this behavior has nothing to do with solar irradiance. It has to do with the fact that there is an enhanced sensitivity to volcanic forcing in the Esper et al series, owing to its bias towards the summer seasonal and extratropical continental centers. And it happens that the centennial timescale changes in the amplitude of explosive volcanic forcing happen to be negatively correlated with estimated solar irradiance variations over past centuries. This simply arises from chance, and the fact that there are very few realizations of the century-scale variability present in the two short forcing series. But it does serve to underscore how naive your interpretations are. You might benefit from a more thorough and careful reading of the literature. The GISS modeling work referred to elsewhere in this thread would be a good place to start. -mike]

    Thus we urgently need better reconstructions…

  22. 72


    In your comment you allude to a 20 ppmv/K sensitivity of climate to changes in CO2 in equilibrium.

    There is a ~10 ppmv/K change in CO2 as result of temperature changes in ice cores, but that is one-way, as temperature changes near always preceded the CO2 changes. As there is in general a huge overlap between temperature change and CO2 change during the ice age – interglacial and vv. transitions, one can think of a huge influence of CO2 on temperature, as a positive feedback.

    But in one particular case, there was no overlap, the end of the Eemian, where CO2 levels stayed steady high, while temperature (and CH4 levels) were already near their minimum. If the 20 ppmv/K holds, then the subsequent decrease of CO2 levels with ~40 ppmv would have induced a ~2 K temperature drop, which is not visible in the temperature proxy of the ice core. See here

    [Response: Wrong way around. ~20 ppmv/K sensitivity of CO2 to global temperature change (assuming ~5 deg C global cooling at the LGM, and roughly 100 ppm decrease in CO2). You get pretty much the same thing with a regression over the whole Vostok timeseries if you make some reasonable assumption about the relationship of vostok T to global T. The relationship the other way is related to the cliamte sensitivity and is complicated by the other forcings (ice sheets, other GHGs etc.). You can’t derive that purely from the Vostok curves. – gavin]

  23. 73
    Nicola Scafetta, PhD says:


    Gavin and I are not making mincemeat of each other :)

    We are just discussing from two different prospectives, the theoretical model one and the empirical one.

    These problems are very complex.

    To answer your question, for what I know,
    there are regions of the Earth were the 11-solar cycle is quite evident and other regions where it is weak or it seems that it is not present. Several studies claim that on average the earth the solar cycle has amplitude of approximately 0.1K. (Three times larger than what present models predict).

    The regional pattern might depend on the cloud pattern formations, or I do not know.

    If you are interested in this topic you might read the Book by Hoyt and Schatten “The role of the sun in climate change”

    [Response:I’d of course recommend my own book: Benestad, R.E. (2002) Solar Activity and Earth’s Climate, Praxis-Springer, Berlin and Heidelberg, 287pp, ISBN: 3-540-43302-3 (available at My book provides takes a more critical stance… -rasmus]

  24. 74
    Nicola Scafetta, PhD says:

    Just a final response to Urs Neu #63,

    As I tried to explain above, the calculations are done by taking in consideration a lot of alternative facts that you are not considering.

    It is true that by doing them as you want you might get a lower value than mine (1/2 you claim), and by doing the calculations as Rasmus wants you might get a value larger (2 times, perhaps).

    The problem is that in both cases the match with the data from 1600 to 1900 is lost. If I plot the data in such a way that the signal and the temperature have the same value during the 17th century and I do the calculations as you want, the signal will be too low during the 19th century (-0.2K in 1900). If I do the calculations as Rasmus wants to do them, the signal will be too high (+0.4K in 1900).

    The hypothesis that I did was that the warming during the pre-industrial centuries, before 1900 is given by the sun increase. So I need to find values that match the data during all three centuries. Moreover to minimize the error I need to compare a period with a minimum value and a period with a maximum value. With your calculation the relative error is 2-3 times larger than mine. With Rasmusâ?? method the relative error is 3-4 times larger than mine. My choice fits better all these properties.

    But there is another deeper problem with your calculations. Your value might be seriously in conflict with other findings and expectations.

    You have to understand that I am looking for the climate sensitivity to slow secular change, with your method you say that you find one half of what I found. Well let us seen the data from Lean1995, I found for the sensitivity something between 0.17 to0.23, you would find something between 0.9 to 0.12.

    The problem is that there are several independent studies, not only mine, that claim that the climate sensitivity to the 11-year solar cycle is 0.11. And the sensitivity to a 22-year cycle might be as large as 0.16. Theoretical estimate (Wigler) estimates that lower frequency sensitivity might be much larger (even 3-5 times for a smooth secular component). This is due to the thermal inertia from the ocean. I have argued in a previous paper that the climate sensitivity to a slow solar variation component might be as large as 0.21 and I cannot go too low of this value. (I wrote this in the paper if you look carefully)

    You understand that your numbers are too small and do not match these further condition.

    [Response:In this estimation, you divided a small amplitude ba an even smaller (the 22-year Hale cycle is not very strong, and not even discernable in the sunspot record, even though we have reasons to believe it exists since the magnetic fields flip), thus not a very reliable method. Furthermore, the phase information is ignored (the phases to not match), and the contention that the ~11 and ~22 year variability is linked to solar activity can only be built on faith… In other words, it’s not very scientific. -rasmus]

  25. 75
    Nicola Scafetta, PhD says:


    thanks, it seems you are the only one who understand what I try to say.

    So, what I say should not look so crazy, after all :)

  26. 76
    muller.charles says:

    Re #71 Mike answer

    It’s quite difficult to judge on Wikipedia plot, but Esper and Moberg reconstructions seem to have the same trends at centennial scale (even if Esper’s one have globally more amplitude than all others). Look at 1000-1500, period where the grah is more clear : upward and downward trends are quite similar. In fact, they diverge mainly for 1500-1600.

    And for 1600-1900, period of interest for Nicola’s paper, Moberg, Esper, Huang (and even Briffa) reconstructions seem to have a comparable slope. So, I think the main question of the debate is volcanic vs solar forcing. Since the beginning, Nicola explains the same thing : he chooses the solar forcing attribution that best fits the temperature data (Moberg) for the period. If volcanic forcing vs solar forcing is 10:90 (negligible), there’s no problem. If we are at 50:50, there’s a problem.

    More generally, the higher temp. variability a reconstruction shows, the higher sensitivity to natural forcings and / or the higher natural – chaotic climate variability we should expect. I think Ferdinand was OK on this point.

    [Response: You’re simply wrong on both counts. It doesn’t take a calculation to see that the two curves are nearly orthogonal over the first 600 years of overlap, at least. Tom Crowley (AGU Transactions, 2003) has shown that the Esper et al reconstruction yields a negative sensitivity to solar forcing using his estimated solar irradiance reconstruction, despite its very large amplitude. These issues just aren’t as simple as non-experts often like to think they are. -mike]

  27. 77

    Re #72 comment,

    Gavin, I misinterpreted your remark as the sensitivity of temperature for CO2. Indeed the ice cores show a remarkable (near) linear response of CO2 to temperature changes, be it overall ~8 ppmv/K for the 420,000 years Vostok ice core, where K more or less reflects the SH ocean temperature. As I suppose that the NH temperature variations are larger (more land, ice sheet building), this should reduce the ppmv/K ratio. Except if there are other influences to be taken into consideration. But no matter what the real ratio is, there is a quite good relationship between CO2 and temperature (at least in one direction)…

    Anyway, the change of ~12 ppmv during the MWP-LIA transition (seen in several Antarctic ice cores) points to an about 1 K cooling in that period. If the same co-variance still is applicable for the LIA-current temperature increase of ~1 K, then this increase in temperature can not be responsible for more than a 12 ppmv rise in CO2 levels. The rest of the observed increase (~70 ppmv in the Law Dome ice core) certainly is caused by burning fossil fuels.

    Thus even if solar was only/mainly responsible for the recent temperature increase, the total increase of CO2 (and CH4) is only marginally attributable to that change. Here I differ with Dr. Scafetta who supposes that solar variations may be more responsible for changes in CO2 and CH4 than can be deduced from historical data. For other responses (especially clouds), there is far more uncertainty which need to be cleared…

  28. 78
    Lawrence McLean says:

    Re #43, Thanks for that link, however, I would still like to see the infra-red absorbion data for CO2.

    On an unrelated subject, at work today, a colleague sent me an email and pronounced that man cannot be causing global warming and, that the whole issue is a load of bunk! His proof was the link: . He said I should get my “facts straight”. I have told him of this site (, however, he refuses to even look. What can you do with folks like that?



  29. 79
    Urs Neu says:


    I am afraid you will not get the point (or you do not want to).

    I perfectly understand what you did. You exclude any result which does not match what you predetermine at the beginning. You predetermine that there has to be a match and you predetermine that the sensitivity has to be around 0.20. It is no wonder that at the end there is a match and the sensitivity is 0.20.

    The problem is not what you did, but your conclusions:
    You write in the abstract:
    1. We find good correspondence between global temperature and solar induced temperature curves during the pre-industrial period.
    – This is in no way a finding but your presumption (if you only look for data which lead to this match, the finding is predetermined).

    2. The approach we propose (..) yields results proven to be almost independent on the secular TSI proxy reconstruction used.
    – This finding is also predetermined by your method because you look for a good match for each TSI reconstruction independently. If you adjust three curves to the same other curve, you will end with three similar curves in any case, this is no surprise.

    3. The sun might have contributed approximately 50% of the observed global warming since 1900.
    – This is a result you did not calculate but estimated from a graph with curves that show very questionnable things at their ends: The smooth Lean2005 curve (data is from Wang et al. 2005 according to your text) goes until about 2010 although Wang et al. 2005 only contain data until 1996. This is miraculous, although considering your 5y time lag. And smooth data is rather uncertain at the end if extended to the very end of the original data as you did.

    I don’t expect an answer again to my points above, but I really would be interested to know why your Lean2005 curve (Wang et al. 2005) contains data in the 17th century and clearly beyond the year 2000 although the Wang et al. data is from 1713-1996. And how you smoothed your curves until the end of the data period.
    Of course this explanation would have blown up the length of your paper, but it is relevant since your 50% approximation depends on the last years of your reconstructions.

  30. 80

    # Some minor points
    #57 Nicola Scafetta first writes that
    “Rasmus continues to not understand that the reason why I used the Moberg temperature signal is that Moberg data are the novelty and that in a letter I cannot write too many things!”
    meaning that he has chosen Moberg for being the latest millenial reconstruction available. That’s however not the case: The Hegerl, G. C., Crowley, T. J., Hyde, W. T. & Frame, D. J. Climate sensitivity constrained by temperature reconstructions over the past seven centuries. Nature 440, 1029-1032 (2006).
    #72 The millenial CO2 variations are quite difficult to estimate. Between different cores one finds differences of more than 10ppm for a specific century (most probably due to post-depositional effects and/or measuring process). If one would average over all records (disregarding the different quality of the data) one would get probably a millenial amplitude of less than 10ppm. The 12ppm/1K is the modelled value from Gerber et al.

  31. 81
    Nicola Scafetta, PhD says:

    A brief reply:

    To Rasmus: it is not only me who claim that there is a 11-year and 22-year and a long-range solar signature in the climate. There is a lot of litterature starting from 1800 to today that say the same thing, even if the things are very complex. Please, read some book such as that by Hoyt and Schatten. In the new IPCC 2007 there will be a paragraph by Lean where she talks also about a 11-year solar signature on climate of the size of 0.1K in average referencing a lot of authors, as I too found. (Science too starts with an act of faith, Rasmus! )

    [Response:You are right, the literature is littered with examples ‘demonstrating’ that there are ‘solar signals’ in the climate. Many of which were based on unconvincing analysis. The fact that we are still debating this centuries after, suggests that any effect is weak/obscure at best. Look at the northern lights in comparison: these phenomena are remote to most people, still there is little doubt that the frequency of northern lights are closely related to the level of solar activity; our climate, on the other hand, is intimately around us… The existence of variations with near 11-year periodicity is not by itself a proof of a solar influence – you may even find such ‘signals’ in computer simulations of Earth’s climate where the changes in the solar activity is not prescribed, as the dnyamics of Earth’s climate gives rise to intrinsic (chaotic) variations. -rasmus]

    To Urs: You have to see the good correspondence of the patterns between the curves. They correspond quite well and this suggests that the hyphotheses might be correct. This might be a coincidence, I agree. But the probability that it is a simple coincidence is quite small. Try to randomize the solar data and then put them on a linear slope equal to the actual linear slope found in the solar data and do again your calculation, then let me know if you find the same good pattern correspondence I found.
    About your question about Wang and Lean data, you should read more carefully my paper, you will see at the very end this sentence: “Acknowledgment. We thank J. Lean for having sent us the LATEST UPDATES of her TSI reconstructions.” Does this respond your question?

    To Ferdinand: I do not claim that “all” or even “most” of the observed CO2 during the last century is induced by the Sun. But let us do a very naive calculation.

    Law Dome ice core shows that there is a natural variability of 10 ppmv from the Middle age maximum to the little ice age of the 17th century. Let us suppose that this variation is induced by the decrease of the solar activity during that period. Let us suppose that the increase of solar activity from 1880 to 1975 has been of the same size at most of the previous decrease or 1/2 of it. This means that the solar increase from 1880 to 1975 might be responsible of 5-10 ppmv.

    Law Dome ice core shows a CO2 increase of 38.6 ppmv from 1880 to 1975.
    This means the sun might have contributes 13-26% of it. This might have a large effect on climate and this might be only the solar contribution via CO2, then there are other contributions from other sources.

    [Response: One last word. We have put more CO2 into the air over the industrial period than now remains there. Therefore the flux of CO2 must be into the other reservoirs (principally the biosphere and the deep ocean). There is no possibility of more CO2 coming back out because of highly speculative solar feedbacks. This is shown by actually measuring the increase of CO2 in the ocean (and it is increasing) and matching the isotopic data (both 14C and 13C) with the fossil fuel source. At most, changes in climate may be making it harder for the anthropogenic CO2 to get sequestered. But that is not good news! -gavin]

  32. 82
    Nicola Scafetta, PhD says:

    thanks but you should understand that a reconstruction should be “recent” but not “too recent” such as the Hegerl’s one. Scientific papers are not written few hours before they are published like newspaper articles!

    Note that Hegerl has not used Moberg’s one in his picture too, just referenced them!!!!!!!!!

    In any case I did not use it because I wanted to show the Medieval maximum that Hegerl’s reconstruction does not show (it starts in 1280).
    Moreover from 1600 to 2000, Moberg is a compromise between Hegerl and Esper.

    In any case the Hegerl climate model simulation (blue line in fig 1) has the same problem of the Foukal’s simulation. That is low secular variability of 0.2-0.3K from 1000 to 1900 and other problems. It mirrors the Mann and Jones data. Again, if the secular temperature variability is larger as Moberg claim, the model is in serious trouble and the climate sensitivity to solar variations must be significantly increased.

  33. 83
    Nicola Scafetta, PhD says:

    “We have put more CO2 into the air over the industrial period than now remains there. Therefore the flux of CO2 must be into the other reservoirs (principally the biosphere and the deep ocean). ”

    yes, this is what I was trying to tell you. There are carbon cycle mechanisms that are absorbing CO2 from the atmosphere into the ocean. And others that are producing is. The net flux is negative.

    You model does not contain any of these mechanisms. These mechanism might be conditioned by the Sun.

    What I am telling is that without the increse of the solar activity during that period the flux from athmosphere into the ocean could be more negative than what it has been.

    Thus, you and your group cannot interpret the “measured” increase of GHG as an “anthropogenic” contribution. Above I have calculated that 13-26% of it might be from the Sun (at least from 1880 to 1975).

    Are you understanding now what is the problem with Hansen’s interpretation of his model?
    A significant part of the observed GHG increase might be due to solar increse.

    [Response: No. All of the extra CO2 in the atmosphere is anthropogenic. Possible feedbacks from the climate to CO2 absorbtion in the deep reservoirs may have made a few ppmv difference to current CO2 concentrations (so far). Your calculation of the potential for the solar input is woefully inadequate (you assume the result you want to prove, ignore timescales of response and exaggerate the importance of your preferred mechanism by ignoring all other potential effects). Basically you show that if no other effects occur than solar cannot explain the CO2 rise (which is obvious anyway), but if other effects are important (volcanos, GHGs) then your attribution of all the warming to solar cannot be valid and thus the CO2 ‘portion’ you come up with cannot possibly be that large. The only fair way to procede is to take account of ALL effects (with their uncertainties), and when you do, you find that solar is a small contributor relatively and a completely trivial contributor to any climate-CO2 feedbacks. – gavin]

  34. 84
    Nicola Scafetta, PhD says:

    to prove that “All of the extra CO2 in the atmosphere is anthropogenic” first you need the carbon cycle mechanisms and you do not have them yet.
    You are jumping to the conclusion you like without doing any calculation and ignoring the phenomenological evidences.

    I did a simple calculation on a time scale of several centuries, and only the Sun has such long range variability. Volcanoes do not have such long effects and in any case I have already show you that the temperature reconstructions almost do not show volcano signals, give again a look to the figure of Foukal, if you have forget it.

    Do you think that the decrease of CO2 from the medieval maximum to the little ice age has been “anthropogenic”?
    If it was not the Solar variability, what was its cause? (Do not be vague in your reply.)

    [Response: Nicola, think of a bathtub. I pour 2 buckets of water in, some of it spills over so that we end up with only 1 buckets worth of new water in the bath. All of the new water is from the 2 buckets I put in right? Now let’s imagine that the weight of the water or some other agency distorts the bath so that it stretches a little and less spills out. Now I have 1.5 buckets of new water in the bath. All of the new water is still from the two buckets I put in. It cannot be otherwise. Plus, in the CO2 case we have the isotope results and ocean measurements and O2 declines etc. which demonstrate conclusively that the extra CO2 is anthropogenic. Next point, changes in volcanic activity can affect decadal and century-scale temperatures due to the random occurence of eruptions of the right sort (though I don’t think you dispute that). Solar activity too can affect temperatures, as can internal variability or potentially, land use changes etc. All of these changes will lead to a change in climate that could affect CO2 sources/sinks. Right now, since we do not have accurate forcing functions for any of these factors going back to the medieval period it is difficult to say with any precision which one (or combination) caused the climate change and what effects that had on CO2. It could be all solar (though volcanoes certainly play a role in the late Maunder Minimum period), but even if it were, you cannot use that to prove that sensitivity to solar forcing is underestimated, for the very good reason that you don’t know what the solar forcing was in the first place. -gavin]

  35. 85
    Nicola Scafetta, PhD says:

    I understand that CO2 can be changed by multiple causes.

    However, in science one interprets the data with what
    is known, not with what is not known.

    Right now, we know that there is a solar cycle, that this cycle is parallel to the CO2 cycle and that the volcano signals (as interpreted by the models) seem to be almost invisible in the temperature data (Foukal picture).

    This leaves the solar cycle the only known cause of climate change on large scale, up to modern times where there is a documented antropogenic contribution.

    Until you identify other credible causes, you might suggest that other causes might exist and look for them, but you cannot prove yet that they really exist. Therefore you cannot dismiss an interpretation of the data based on what is known by using what you do not known yet.

    By the way, have you never thought the possibility that volcano activity might be some how influenced by solar variation too?
    I am not talking of the single eruption, but cluster of eruptions. and I see a correspondence between Mauder minimum, Dalton minimum and an increase of volcano activity.
    If so, volcano activity too might be a solar feedback (at least in part), who can tell?. :)

    (have you noticed that I talk about solar “variation” and not of solar “forcing”?)

    [Response: Solar ‘variation’ impacts on volcanoes? Please. Imagination is a good quality in a scientist but occasionally it needs to be exposed to the real world. -gavin]

  36. 86
    Marcel says:

    no one is denying that the climate is changing for the simple reason that it has always done so and always will continue to do so until this planet’s lifespan expires

    climate change is a natural phenomenon

    no amount of scaremongering can ever make it a man made phenomenon

    the mere fact that people advocate shoveling money to other countries to solve it proves it is about wealth redistribution

    we should focus on new technology and fighting pollution

  37. 87
    Dan says:

    re: 85. No, scientific inquiry and results published in peer-reviewed scientific literature have shown the recent global warming is primarily due to anthropogenic GHG emissions. Any other conclusion is simply denying science. No amount of denial can ever make the scientific evidence go away just because someone does not like the results. Read the information for yourself. Don’t let others tell you what to think or say.

  38. 88
    Blair Dowden says:

    Re #86: You have produced a remarkable number of logical fallacies in such a small amount of space.

    1) Fallacy of the single cause: Climate can change naturally, therefore it cannot also be changed by human effects.

    2) True but irrelevant: “no amount of scaremongering can ever make it a man made phenomenon.” Climate is determined by physical factors, not by statements that people make about it.

    3) Ad Hominem (questioning the motive rather than the facts): The fact that some people use the issue of climate change to pursue other agendas has no relevance to the accuracy of the science.

    4) Nice conclusion, but it has no relation to the other statements you made.

  39. 89
    Urs Neu says:


    Thanks for the information. This is the first paper where I have to search for the origin of the data in the acknowledgement, that s why I have missed it.

    As I have said before, I completely agree that there is some evidence for a corresponding pattern on the timescale of several decades up to one century (Maunder and Dalton Minimum). However, there is absolutely no evidence for the link in the long-term trend over the three centuries (17th to 19th) since the increase is nearly linear. The only correspondence is that both trends are positive.
    The correspondence of the medium-range pattern is even better with half of your sensitivity. The Dalton Minimum in the smoothed temperature curve (your Figure 1) has an amplitude of about 0.15K, while your reconstruction has an amplitude of about 0.3K. With half Zs this match would be much better. If you would have plotted the smooth temperature curve instead of the unsmoothed in Figure 2 (which would have been much more adequate for comparison), this would be obvious.
    Thus with half Zs the match would be better on the time scale, where a corresponding pattern is obvious, and would only be worse for the long-term trend, where the correspondence is only based on your assumption.

    Besides: It is not the first time that you do not follow your own arguments: On the one hand you tell Georg that he should understand that a reconstruction should be “recent” but not “too recent”, on the other hand you use very recent data, which is even not yet published (the Lean data of the acknowledgement).
    Please be consistent.

  40. 90

    # in 82 Sacfetta writes:
    “you should understand that a reconstruction should be “recent” but not “too recent” such as the Hegerl’s one”
    My point was mainly if you prefer Moberg for your analysis than please give rational arguments for it (and not “novelty”). Would you have taken Hegerl et al if your submissions were a month later? No, you prefer Moberg since it gives you a large signal (that’s not a crime as such but it should be honestly reported). It’s beyound me why none of your reviewers has asked you to check on the sensitivity to the selected reconstructions.
    Also some of the available reconstructions estimate the uncertainties in the reconstruction. Why did nobody ask you to do a proper error analysis in your paper? What I am missing is 1) sensitivity to different reconstructions. 2) Uncertainty/error analysis based on both the uncertainties in the temperature reconstruction and in the TSI (allready mentioned here by rasmus and urs). This is nearly no work due to the simplicity and partly linearity (Z function) of your empirical approach.
    By the way the reconstruction of Hegerl et al is presented in more detail in another paper (check on her webside: and in fact includes the MWP).
    Last point about hidden solar GHG feedbacks. On longer timesscales one could compare 14C reconstructed sunspot numbers (Solanki) with the Taylor Dome CO2 record. Whatever the smoothing in the ice exactly is and whatever inertia you assume between this proxy of solar activity and hypothesized CO2 feedbacks you will not find any similarity. Basically C02 shows a maximum of 270ppm at about 10k slight decrease to 8K (260ppm) and then a slow increase to 280ppm preindustrial. The reconstructed sunspot number shows in particular a minimum at about 7k, slow rise to a maximum at 4.5k and decrease again. Of course you might argue that there are also changes on Milankovitch timescales involved, but at least one can state that there is no hint on longer scales that your ideas about CO2/solar activity feedbacks were confirmed.
    # Rasmus
    I think your comments here deserve a comment to GRL.

  41. 91
    Nicola Scafetta, PhD says:

    Ok, I short reply:

    Urs: I have already explained you why I do not believe that the long sensitivity can be 1/2 of what I found. It will simply not match the data from 1600 to 1900 and will be incompatible with other measures. You should not see only one pattern, but several of them. The data are updated continously, and are available under request, normally a new paper with updated data is published every few years, not every month.

    Georg: I used Moberg because of the message it gives. That is, that there is a wide 1000 year cycle in the temperature against the Mann and Jones record. This is the “novelty,” not the fact that it is the most recent in time.

    A new reconstruction does not automatically prove that the previous one is wrong, in any case. The message of my paper is that if the temperature has a large 1000-year cycle, likely the climate models are seriously underestimating the solar effect on climate. This is all.

    If you change the temperature reconstruction you will get a greater or lower solar impact on climate according to the amplitude of the 1000 year temperature signal, of course. So, if you use Mann and Jones you will find a much lower value than mine which is more or less compatible to that the current models have, if you use Hegerl you will find a value lower than mine, then there is mine and finally if you use Ensper you will find a larger value than mine.

    About the CO2 effect, I did not do any real calculations, I simply suggest that the model should contain the CO2 and CH4 cycle mechanisms and many other mechanisms to be correctly interpreted. You comments on this are based on the assumption that the CO2 and CH4 cycle mechanisms are linear, while they might be strongly nonlinear. Until these mechanisms are understood the data cannot be interpreted correctly. This is all.

    I strongly believe that in 25 years the climate models will be much much better than today, and I strongly believe that Rasmus and Gavin are 100% in agreement with me on this (is it correct?).

  42. 92

    Re #71 (comment):

    Mike, if there is a larger variation in the pre-industrial past, no matter what the cause is, that implies that the current variation to a certain extent is more variable than is implemented in most GCMs. No matter if that was caused by solar, volcanic or other natural (internal) variations.

    In that way, I am in good company, see the discussion about a similar point of view by Esper, Wilson, Moberg, Luterbacher ea. at RC
    Most climate models use a climate sensitivity around 3 K/2xCO2. The Echo-G model uses a 2.1 K/2xCO2 sensitivity and is able to reproduce the Moberg reconstruction + instrumental trend. It also reproduces the Eemian-last glacial transition where CO2 levels remained high, while temperatures were near minimum (and ice sheet formation was near maximum).

    That means that models with high 2xCO2 sensitivity will show an overshoot for current temperatures, if the real variation in the past was larger. Or that models with low 2xCO2 sensitivity will show an underestimate for current temperatures, if the real variation in the past was lower.

    About the recontructions, there are large variations between them, partly because proxy data before 1600 are increasingly unreliable, according to the NAS panel. Further even the calibration period and scaling/regression methods play a huge role and can give a difference of 0.5 K in amplitude for the same reconstructions. See Esper, Frank, Wilson & Briffa.

    Further, reconstructions based only on tree rings may overestimate the influence of volcanic eruptions, as not only the temperature is reduced, but there is also a change in direct and diffuse incoming sunlight. If this is corrected, then there is no influence longer than a few years (up to ten years after Tambora) for the strongest eruptions, according to Alan Robock (warning: 36 Mb .ppt file!):

    When proxy records of Northern Hemisphere climate change are corrected for the diffuse effect, there is no impact on climate change for time scales longer than 20 years. However, it appears that there was a hemispheric cooling of about 0.6°C for a decade following the unknown volcanic eruption of 1809 and Tambora in 1815, and a cooling of 0.3°C for several years following the Krakatau eruption of 1883.

    Thus if volcanic eruptions only had a temperary and a limited influence on climate, most of the temperature variations are caused by solar changes (and maybe some internal variations).

    The reconstruction by Esper is only based on tree rings. If not corrected for the above influence, it will show too much cooling after major eruptions, while the reconstruction of Moberg has a reduced impact of tree rings vs. other proxies, thus is less influenced by them.

  43. 93
    Coby says:

    To take up the bathtub analogy, Gavin’s version misses Nicola’s point:

    The tube has to have a tap that is filling it and a drain that is draining it, roughly balanced. Now we dump in the buckets. Nicola’s argument is that at the same time the tap flow could be increasing. So now part of the 1 1/2 extra bucketfuls in the tub is indeed the “natural” tap water. This would be the solar feedback vs the anthro CO2.

    However this point is defeated by the isotope analysis of the atmospheric CO2. I don’t know the details, like the uncertainties, so there may indeed be room for a small increase in natural CO2, but it seems that it can’t be a significant amount (a point I believe Gavin already acknowledged up thread).

  44. 94
    Nicola Scafetta, PhD says:

    Yes, Ferdinand
    that is another reason I like Moberg’s reconstruction more than the others. He uses a wavelet filter to separate the tree ring proxies from the other proxies. There are a lot of mathematical/physical reasons to prove that this is the right approach.

    Moreover, I believe that Gavin is missing an important thing I told above. That is, his GISS model does not reproduce the temperature increase from 1880 to 1960. Only from 1960 to 2003 it is reasonable.

    I did the calculations exactly by using the GISS simulation and the HadCRUT3 global temperature data: this is the result.

    If I look at the period [1880:1910] I get:

    linear fit[1880:1910]=-0.0075K/y

    GISS simulation:
    linear fit[1880:1910]=+0.0032K/y

    So, the temperature DECREASES and the GISS model INCREASES!!!!

    If I look at the period [1900:1950] I get:

    linear fit[1900:1950]=+0.008K/y
    linear fit[1910:1940]=+0.015k/y

    GISS simulation:
    linear fit[1900:1950]=+0.00433K/y
    linear fit[1910:1940]=+0.00429k/y

    This means that in [1900:1950] the GISS rate increase is from 2 to 4 times LOWER than the observed temperature rate increase!!!!!!!!!

    Moreover if I look at the period [1940:1960] I get:

    linear fit[1940:1960]=-0.0081K/y

    Giss simulation:
    linear fit[1940:1960]=+0.0037K/y

    Finally, the temperature DECREASES and the GISS model INCREASES!!!!

    Conclusion: from 1880 to 1960 the data have a pattern of the type
    “Down-Up-Down” while GISS has a pattern “Up-Up-Up”. So, GISS simply crosses the data.

    Question for all of you, which forcing has a pattern “Down-Up-Down” in 1880-1960?
    Guess the answer:

    b)GHG+volcano+aerosol+(little)sun (=GISS estimates)

    [Response: [Can we reduce the number of exclamation points please? It’s the equivalent of you jumping on the table at a meeting….]. None of your analyses take account of the internal variability which is an important factor in the earlier years and would show that most of the differences in trends over short periods are not significant. Of course, if you have a scheme to detect the difference between intrinsic decadal variability and forced variability, we’d love to see it. Secondly, your comment about Moberg being ‘proven’ right because they use a wavelet filter demonstrates conclusively that you haven’t really thought about this. The big problem with Moberg (or with any scheme that separates out the low frequency component) is that it is very very difficult to calibrate that component against observed instrumental data while still holding enough data back to allow for validation of that low frequency component. Thus the component you are most interested in, is the one with the loosest calibration – that is just a fact of life. -gavin]

    [Response: Following up Gavin’s comment, it has indeed already been shown–based on experiments with synthetic proxy data derived from a long climate model simulation (see Figure 5 herein)–that the calibration method used by Moberg et al is prone to artificially inflating low-frequency variability. -mike]

  45. 95
    Nicola Scafetta, PhD says:


    short periods?!

    Giss model has simulated 123 years (from 1880 to 2003) and for the first 60 there is an evident problem that can be solved if the solar contribution is stressed against the antropogenic one.

    GCM are supposed to reproduce the “internal variability”!
    Or not? What kind of “circulation” do they contain?

    About Moberg I did not state that his approach is the “easiest” one!

    [Response: ??? Internal variability in each GCM simulation is uncorrelated to internal variability in the real world and so for short periods (a couple of decades), that variability makes any comparison of short term trends problematic. The forcing fields for the model are the best available to date. Should anyone come up with forcings that are shown to be more accurate, we will use them – however, the tendency with solar studies has been to reduce the magnitude, not increase it (and if you don’t like that, take it up with them, not us). -gavin]

  46. 96
    makarov says:

    An interesting discussion,ther seems to be an over reliance on the accuracy of interpretation of the volcanic signature viz a viz the solar signature.

    Looking at the review of the IPCC AR4 climate models review Stenchikov et al there seems to be some question on the accuracy of the models you use Gavin.

    Whilst the differentials and ranges in hindcast can be expected,the observations vs Giss was over estimated 115 times.No wonder you cannot see the inverse solar signal.

    As noted above, most earlier hindcasts of 20th century climate as well as current IPCC AR4 runs [Miller et al., 2006; Knutson et al., 2006] do not reproduce the observed trends over recent decades in the AO component of the circulation, and thus do not capture the intensification of warming trends that has been observed over Northern Europe and Asia. There are various possible explanations for this discrepancy, but it is interesting to speculate that it could indicate that the models employed may have a basic inadequacy that does not allow a sufficiently strong AO response to large-scale forcing, and that this inadequacy could also be reflected in the simulated response to volcanic aerosol loading.

    [Response: Err… yes. We’ve been discussing this for years: . I have no idea what your 115 times error refers to. – gavin]

  47. 97
    Urs Neu says:


    I would be really interested in a comment about the double amplitude of the Dalton Minimum in your reconstruction compared to the Moberg temperature. To compare amplitudes of two signals is much more meaningful than to compare trends.
    Why did you not show the smoothed curve of the Moberg temperature in Figure 2 (to compare apples with apples) if not to make comparison more difficult and hide otherwise obvious differences?

  48. 98

    Re #94 (comment):
    Mike, the (possible) overestimate of Moberg for the low-frequency signal is based on a model, which uses a small solar variability of 0.15 W/m2 over the 1650-2000 period. Von Storch ea. used a ~1 W/m2 solar variance for their model runs. As the Lean ea. 2000/2005 TSI estimates at the TOA are 2.5/1.25 W/m2 or at the surface 0.43/0.21 W/m2 resp., it seems to me that your model decreases solar variability, while VS inflates it.

    Thus are we not looking at a chicken-and-eg problem here? If solar is enhanced (by some factor like stratospheric and/or cloud responses), then the model fits the Moberg reconstruction, while with lower solar variability, the low-variability reconstructions are within the model margins…

    Re #95 (comment):
    Gavin, solar reconstructions have a lot of problems, as good as temperature reconstructions. But even if the secular trend of solar is reduced in more recent estimates, that only means that the pre-industrial fit of the temperature reconstructions needs a larger factor for solar…

  49. 99

    I’m just a humble observer but it seems to me that the main points Nicola and Ferdinand are making are quite reasonable.

    [Response:??? I must say I’m a bit surprised, as I think it’s fairly clear that the methods employed in the paper – the main criticism of this post – are unsuitable. Not only are the TSI and temperature reconstructions very uncertain (hence, probably quite noisy and containing large errors, as the choice of different curves give such large variations) , but also the fact that the temperature is affected by a multitude of factors. In addition, there are points which have been stretched beyond justification… No? – rasmus]

    Let’s put it this way: it is assumed that sensitivity to 2xCO2 is around +3 C. And we know that, since the late 19th century, CO2 has risen x1.35 and global temperature approximately +0.7 C. Well, if we come to the conclusion that half of that T rise was solar-induced or that the negative effect on T of sulphate aerosols over this period was much lower than previously estimated, we’ll have to revise our 2xCO2 sensitivity estimation downwards.

    (As a matter of fact, we might well have already experienced the equivalent of a 1.5xCO2 increase when we take into account the rest of AGHGs for a total T increase of only +0.7C, but let’s not complicate matters).

    [Response: Humble or not, you would still benefit from reading the articles under . Neither of your points are actually valid. – gavin]

  50. 100
    muller.charles says:

    As Mike told me in his answer, it’s sometimes difficult for non-expert to make his own opinion on a climatic issue, when there is a conflict. So, I’ve a very basic and non-expert question : is there any theoretical (non empirical, for the moment) reason to expect that climative sensitivity to 1 W/m2 solar forcing is the same that climate sensitivity to 1 W/m2 GHG forcing ? Or, another way for the same question, should we expect identical (in amplitude) feedbacks for the same (radiative) variation in TSI and well-mixed GHGs ? By advance, thanks.

    [Response: You don’t expect it to be completely the same since there are differences: GHGs cause stratospheric cooling, solar irradiance increases cause warming there – GHGs have a very even effect across latitudes, solar is stronger in the tropics. GHGs are stronger at night, solar obviously isn’t. However, in all modelling experiments so far done (and with many different models), the effects of equivalent GHG and solar forcings (defined as the forcing at the tropopause) are very similar (within about 10%). – gavin]