Climate Science Glossary

Term Lookup

Enter a term in the search box to find its definition.


Use the controls in the far right panel to increase or decrease the number of terms automatically displayed (or to completely turn that feature off).

Term Lookup


All IPCC definitions taken from Climate Change 2007: The Physical Science Basis. Working Group I Contribution to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change, Annex I, Glossary, pp. 941-954. Cambridge University Press.

Home Arguments Software Resources Comments The Consensus Project Translations About Support

Bluesky Facebook LinkedIn Mastodon MeWe

Twitter YouTube RSS Posts RSS Comments Email Subscribe

Climate's changed before
It's the sun
It's not bad
There is no consensus
It's cooling
Models are unreliable
Temp record is unreliable
Animals and plants can adapt
It hasn't warmed since 1998
Antarctica is gaining ice
View All Arguments...

New? Register here
Forgot your password?

Latest Posts


Scafetta's Widget Problems

Posted on 24 February 2012 by dana1981, Dikran Marsupial

We have previously examined the work of Nicola Scafetta, a climate "skeptic" and solar-climate researcher at Duke University.  Scafetta's pet hypothesis is that astronomical cycles are somehow responsible for most of the observed global warming over the past century; a concept we have termed "climastrology," because Scafetta has proposed no plausible physical mechanism through which the orbital cycles of various planets should exert so much influence over the climate on Earth.

In recent papers, Scafetta has put forth predictions as to how the average global surface temperature will change in the future.  He has also now created a widget to compare his prediction to the IPCC projections and the monthly observed global surface temperatures.  However, as we will discuss here, there are problems with both the widget itself, and the research on which it is based.

Extreme Curve Fitting

The widget is based on Scafetta (2011), which is very similar to a paper we previously examined, Loehle and Scafetta 2011 (LS11).  The latter created a very simple climate model using two cycles (of 60- and 20-year periods) plus a linear warming trend, and adjusted the parameters in their model to fit the observed temperature data.  As we showed, this simple model does not accurately hindcast past temperature changes (Figure 1), and thus there is little reason to expect it to accurately predict future temperature changes.  It was merely an excercise in curve fitting, matching up a model with the temperature data without any physical constraints.

L&S failure

Figure 1: The LS11 Case 2 model projected backwards in time (red), compared to the Moberg et al. (2005) millennial northern hemisphere temperature reconstruction (blue) and the Loehle (2008) millennial global temperature reconstruction (green).

In his newer paper, Scafetta has taken this curve fitting process several steps further yet.  As in LS11 he uses a model with 60- and 20-year cycles and a linear warming trend, but now he has also added a 10.44- and 9.07-year cycle, as well as a quadratic term. 

With so many parameters in his model (each astronomical cycle has a time and amplitude variable, in addition to the quadratic, linear, and constant parameters), we're reminded of Barry Bickmore's examination of Roy Spencer's efforts to attribute global warming to internal natural cycles.  Bickmore referenced a quotation from the famous mathematician, John von Neumann (h/t Tim Lambert).

With four parameters I can fit an elephant, and with five I can make him wiggle his trunk.

With so many parameters to work with, it's not surprising that Scafetta is able to match his model well with the temperature record.  However, as with LS11, Scafetta (2011) has no hindcasting power.  It is specifically fit to the 1850-2011 temperature record, and thus is not intended to apply to past temperatures.  But if it can't hindcast past temperatures, then why should we expect it to accurately forecast future temperatures?

In fact, not even Scafetta expects his model to create accurate temperature predictions.  He applies a different formula to predict post-2000 temperatures than he does to fit the 1850-2000 temperatures.  His argument for changing the model is essentially that solar and volcanic effects contributed to his quadratic and linear warming trends during the 1850-2000 period, and he does not expect those contributions to continue into the future.  Thus for the post-2000 period, he rather arbitrarily eliminates the quadratic term, and simly models a small linear warming trend of approximately 0.09°C per decade into the future, or approximately 0.9°C warming from 2000 to 2100.

However, there is little purpose in creating a model if we simply discard its predictions and replace them with our own rather arbitrary adjustments.  Scafetta's astronomical cycles do not explain the warming trend from 1850-2000, because they are cyclical.  This is illustrated in Figure 2, which removes the quadratic and linear terms from Scafetta's model.  The influence of the remaining cycles results in nearly zero long-term temperature trend.

scafetta astronomical only

Figure 2: Running the Scafetta (2011) model with the quadratic and linear warming factors removed, leaving just the astronomical cycles.

In short, the quadratic and linear terms explain virtually all of the warming trend, and at most the astronomical cycles could explain the temperature variations.  Additionally, Scafetta's model without the quadtraic term is not a better fit to the data than when the quadratic term retained post-2000 (Figure 3).

Scafetta with and without quad term

Figure 3: Scafetta model with (red) and without (blue) the quadratic term vs. NCDC annual surface temperature (green) from 2000 to 2011.

Since Scafetta's 1850-2000 model fits the data better than his post-2000 adjustment, he has no justification for making this adjustment.  Running the 1850-2000 model to 2100 nearly doubles the surface warming prediction from 0.9 to 1.6°C between 2000 and 2100.


The quadratic and linear terms in Scafetta's model must be associated with an external radiative forcing, and as we have previously shown, greenhouse gases are the dominant contributors to that radiative forcing and thus the consequent temperature change (Figure 4).


Figure 4: Percent contributions of various effects to the observed global surface warming over the past 100-150 years according to Tett et al. 2000 (T00, dark blue), Meehl et al. 2004 (M04, red), Stone et al. 2007 (S07, green), Lean and Rind 2008 (LR08, purple), Stott et al. 2010 (S10, gray), Huber and Knutti 2011 (HR11, light blue), and Gillett et al. 2012 (G12, orange).

Scafetta tries to downplay the role of greenhouse gases by making passing comments about galactic cosmic rays and other effects which he asserts climate scientists have underestimated.  However, Scafetta has not supported these assertions in his research.  All he has done is create a simple model in which short-term temperature variations are attributed to various cycles and the long-term global warming trend is attributed to quadratic and linear terms.  Based on the body of scientific evidence, these two terms are mainly due to greenhouse gases (Figure 4).

Scafetta also attributes his two new hypothesized cycles to Jupiter-Saturn and Sun-Moon tidal cycles, respectively, but as in his previous paper with Loehle (in which the 60- and 20-year cycles are also attributed to Saturn and Jupiter), he does not propose a physical mechanism through which these cycles could impact temperatures on Earth.  Thus we are again forced to describe this hypothesis as "climastrology," and since these cycles only explain temperature variability (not long-term trends), they give us no reason to doubt the anthropogenic global warming theory.

Since he has diregarded his own model in predicting post-2000 temperature changes, we have little reason to believe that prediction will be accurate.  Scafetta gives a range of possible surface warming trends from 2000 to 2100 (0.66 to 1.3°C), but does not explain how greenhouse gas emissions changes will play into this range.  The IPCC issues temperature projections based on specific emissions scenarios, but because Scafetta's model has no physical basis, he cannot do the same.  Thus his prediction is based on little more than his interpretation of the body of climate science literature (i.e. overemphasizing the role of cosmic rays on climate), and his reading of the scientific literature is itself quite problematic.

Selective Reading

Since Scafetta's hypothesized astronomical cycles have no impact on the long-term warming trend, his prediction that future global warming will be minimal is based entirely on his interpretation of the scientific literature.  For example, he twice mentions that galactic cosmic rays may play a significant role in influencing global temperatures, but fails to mention any of the dozens of papers which have found little if any correlation between cosmic rays and cloud cover on Earth.  Scafetta also fails to mention that solar magnetic field, which influences the amount of cosmic rays reaching Earth, has no long-term trend over the past ~60 years, nor does cosmic ray flux on Earth.  Thus how can cosmic rays possibly explain the rapid warming over that period?  This question remains unanswered.

Additionally, Scafetta references a number of papers which have received peer-reviewed responses, without even mentioning any of those papers:

Scafetta (2011) was published in a solar rather than climate journal (the Journal of Atmospheric and Solar-Terrestrial Physics), so perhaps the reviewers were unfamiliar with the body of climate science literature and thus allowed this selective reading to pass, but if so, the lack of reviewers informed about the body of climate science literature is itself a problem.  Regardless, Scafetta's justification in predicting a small amount of global warming over the next century is based on little more than this selective reading and disregarding of most of the body of climate science literature.

The Widget and IPCC

Scafetta has taken a graphic from his 2011 paper comparing his prediction and IPCC projections to the observational temperature data to create a 'widget' which can be updated to see which is closer to the data at any given time (Figure 5).

scafetta widget

Figure 5: Scafetta's Widget.

However, aside from the issues with Scafetta (2011) discussed above, this widget has some problems of its own.  First, the IPCC projections come from this figure, which only depicts the 1-sigma uncertainty range.  Most readers would interpret the green area in Scafetta's widget to be a region that the IPCC would confidently expect to contain observations, which isn't really captured by a 1-sigma interval, which would only cover 68.2% of the data (assuming a Gaussian distribution).  A 2-sigma envelope would cover about 95% of the observations, and if the observations lay outside that larger region it would be substantial cause for concern.  Thus it would be a more appropriate choice for Scafetta's green envelope.

Second, while the IPCC envelope (Scafetta's green) is based on annual data, in his widget Scafetta plots monthly data, which has greater variability and thus is much more likely to fall outside of the envelope. 

Third, Scafetta has used HadCRUT3 data, which has a known cool bias and which will shortly be replaced by HadCRUT4

Fourth, although the widget itself only shows post-2000 data, Scafetta has used a 1900-2000 baseline.  The choice of baseline is not an important issue when discussing trends, which are independent of the baseline choice.  However, when only comparing the data graphically, as Scafetta's widget does, the choice of baselines can deceive the eye.  For example, if Scafetta's widget were to use the 1980-1999 baseline (as in the IPCC report), the apparent discrepancy between the data and green envelope would be reduced.  If Scafetta wants to use the 1900-2000 baseline, an argument can certainly be made to do so, but he should also update the uncertainty of the model projections to account for the change in baseline (which was not done, and as a result likely further biases the test against the IPCC models).  If a longer baseline is used, this will necessarily increase the variance in the 1980–1999 period, making the uncertainty envelope wider in 2000 than depicted using the original baseline.

We have made these adjustments to Scafetta's widget in Figure 6, showing both 1-sigma and 2-sigma IPCC uncertainties, using annual NCDC temperature data, and using the 1980-1999 baseline on which the IPCC projections are based.

scafetta widget revised

Figure 6: IPCC AR4 envelope of model runs with 1-sigma (grey) and 2-sigma (black) uncertainty range, NCDC annual surface temperature (red), Scafetta (2011) prediction (green), and multi-model mean for IPCC Scenario A2 (blue).

As Figure 6 shows, while the NCDC data falls outside the 1-sigma envelope in 2008 and 2011, it has not fallen outside the 2-sigma envelope.  Thus there is no evidence of a significant inconsistency between the IPCC models and the observations.

Wonky Widget

In summary, the basis of Scafetta's widget (Scafetta 2011) has a number of flaws.  Scafetta's model is little more than an extreme example of curve fitting without a sound physical basis.  The proposed astronomical cycles only explain the variability in the temperature data, not the long-term trend.  The trend is explained by quadratic and linear terms, which he arbitrarily adjusts starting in 2000, based on a very selective reading of the climate science literature.  And his model does not fit the post-2000 data any better with the arbitrary adjustment.

The widget itself is similarly problematic.  It only displays the IPCC 1-sigma uncertainty range when a 2-sigma range would be more appropriate.  It uses monthly data, which has more variability than the annual data in the IPCC figure.  It also uses HadCRUT3 data, which is biased low and will soon be superceded with HadCRUT4.  And the choice of baseline in the figure exaggerates the visual difference between the data and IPCC projections.

While it is commendable that Scafetta is putting his money where his mouth is with this specific temperature prediction, there is little if any reason to expect the prediction to have any accuracy, and the widget itself should be revised to address the issues discussed above, as in Figure 6.

0 0

Printable Version  |  Link to this page


Prev  1  2  

Comments 51 to 70 out of 70:

  1. >> [JH]You have astutely avoided telling us what you are attempting to acheive in your serial and rambling postings. I came here to learn and to discuss. Learning is an engaging process. I believe I speak for most humans when I say that learning is weak if done by rote. You have to lay your doubts out on the table. My communication skills are rather limited, so perhaps I ramble more than makes you comfortable. I thought I was following the Comments Policy that I have seen pointed out to others.
    0 0
    Moderator Response: [JH] I just wanted to make sure that you were not here to make mischief. My aplogioes for being gruff. Please proceed.
  2. DM, since this is authentic climastrology, perhaps we can persuade Tamino to waste his time by performing said statistical analysis for Volker. I'm sure Volker would welcome the chance to have an independent statistical analysis of his theory. I'm certain that if Tamino found robust significance across a variety of tests, he would help publish the findings. I do wonder, though, if Volker would abandon the theory if no significance was found.
    0 0
  3. It's a good thing if people can come here and engage. There is a wide gap in understanding among most of the people out there. To close off this forum to most non-scientists and to people who don't already understand and obviously can't agree would represent a missed opportunity.
    0 0
  4. I was constructing a response to a comment which made unsupported assertions, which has gone as I previewed, but for the benefit of readers, here's RealClimate's take on comparisons between AR4 models and data. Give me these over climastrological cycles any day! RealClimate 2011 update to model-data comparisons
    0 0
  5. Note, the figure shown by skywatcher @54 is essentially the same as Figure 6 in the post above (with a few minor differences).
    0 0
  6. Dikran Marsupial wtote at 07:12 AM on 2 March, 2012: “Volker, the scientific component of my objection was that you had not demonstrated that the coherence was anything more than temporary and actually existed throughout the satellite records.” I have demonstrated, and AFAIK as the _first_ researcher in climate, that there is coherence between the main frequency of the season cleaned sea level oscillations of the satellite records with the frequency of the heliocentric synodic tide function of the couple of Mercury/Earth over the whole time interval of the satellite measurements as a fact. I have no idea what mechanism is involved in that coherence; the fact is that the phase coherence shows high sea levels if the solar tide function shows spring tides and vice versa. And the high sea level peaks are also visible in global temperatures of the UAH satellite measurements as a higher temperature. If you call this a not more demonstration, that is your freedom; but I think it is worth to discuss it in the science community because a possible mechanism can bring more light into the basic question of the cause of the global warming. “I also pointed out that you had not performed a statstical analysis to see if a temporary coherence is surprising. Those are both valid scientific points, which you have refused to address.” I have refused your claim because I not agree with you on that analysis. A trained eye is able to see in a graph if there is a correlation or not. I have no possibility and no software to calculate a correlation coefficient. But each interested guy can perform such coefficient, because all date are public. It is not necessary that I must perform that. An other point is that the statistic tool is not an adequate tool to get a significance here; a significance is given better by a great number of comparisons of well known temperature reconstructions or proxies with the all the relevant solar tide functions, like A. Moberg et al, Bond et al, Zorita, R. Edwards, Mangini, G. Patzelt, and 10+ more, inclusive hadcrut3 and UAH. But as you know, there are differences between the proxies, maybe from different chemical processes in stalagmites and tree rings. I have done some hundred comparisons and have found by empery the strength of the tide functions of the several couples. “There is also the point that, like Scafetta, you need a plausible physical mechanism that can explain the strength of the effect, not just the correlation/coherence.” No, you are wrong. To realize a precise climate forecasting tool for the next millennium I do not need a mechanism, I do need only the NASA ephemerides and the empirical fitted tide function strengths. But nevertheless I have some hints about the physical background, which is not the gravitation of Sir Newton, but the density of the volumes. Rest snipped. V.
    0 0

    [DB] ...

    DNFTT, people.

  7. Volker,
    I do not need a mechanism...
    To be taken seriously, yes, you do.
    0 0
  8. "... which is not the gravitation of Sir Newton, but the density of the volumes."
    Oh my! And next up from Volker, how white dwarf and neutron stars hundreds of light years away, with extreme densities (but somehow nothing to do with their gravity), control the future of the Solar System ... Sphaerica's point #57 resoundingly seconded!
    0 0
  9. "... which is not the gravitation of Sir Newton, but the density of the volumes." I recall from the Cosmos TV series when Carl Sagan noted that the gravitational attraction of the obstetrician was larger than that of Jupiter, Saturn, or Mars - simply because (s)he was much closer. Thus throwing some severe doubt on astrology... Volker - Quite seriously, if you cannot point to a physical mechanism (a testable one, with some evidence), you are simply engaging in climastrology.
    0 0
  10. Hmm, when someone postulates that their forecasting tools are going to better than physical models, then I sense the possibility for a bet. So when do you think your "model" will deliver a climate that is outside the predictions of climate models?
    0 0
  11. Volker I also don't like repetition, so I will just address a couple of points. As KR, Scaddenp, Spherica and Skywatcher suggests, if you want your theory to be taken seriously, then yes, you do need a plausible physical mechanism. "I have refused your claim because I not agree with you on that analysis. A trained eye is able to see in a graph if there is a correlation or not. I have no possibility and no software to calculate a correlation coefficient. But each interested guy can perform such coefficient, because all date are public. It is not necessary that I must perform that. " This is wrong in almost every respect. Firstly we have statistics simply because the eye is unreliable, and an eye that has been trained to see correllations doubly so (as it will see spurious correllations even more readily). Secondly, there is a piece of statistical software called R which is in the public domain, so lack of software is no excuse for not performing the statistical analysis. If you don't understand statistics well enough to perform the analysis, then you need to be more receptive to the advice of those who do. Just measuring the correllation coefficient is easy, what you really need to do is to show that after having investigated the amount of data that you have looked at that it would be unusual to have found a match giving a coefficient as high as that which you observed. This is the test you need to perform to show that the coherence is nothing more than a random artefact. Your last point is the most egregious. It is your theory, this the onus is on you to perform the analysis to suggest that the theory is valid and to defend it against specific scientific criticisms. That is the way science works, whether you like it or not.
    0 0
  12. Density of the volumes? Boy, that ought to be some seriously interesting concept. Perhaps it's closely related to the density of the masses...
    0 0
  13. Sphaerica says at 10:43 AM on 2 March, 2012 Volker, I do not need a mechanism... To be taken seriously, yes, you do. Hi Sphaerica, My impression is, that a discussion about this point will be snipped by a moderator as OT. (-snipIn short: Astronomy is, as you may know the law of the Stars, while star comes from A_star_te or E_ster and means Venus). The laws are well known and as you also may know, there is no mechanism in the laws formulated by J. Kepler. Carl Sagan has argued: “ No mechanism was known, for example, for continental drift when it was proposed by Wegener. Nevertheless, we see that Wegener was right, and those who objected on the grounds of unavailable mechanism were wrong”.-) V.
    0 0

    [DB] Off-topic dissembling and evasion snipped.

  14. Why was Volker Doorman taken even halfway seriously? Look at his website; it's explicitly astrologically oriented with a new-age eastern mystical bent. There never *was* going to be a scientific argument from him.
    0 0
  15. You know why, Robert. Given the current state of understanding in the general public, every alternative theory, no matter how absurd, must be addressed. If no one challenged Volker, I guarantee that within five years my mother-in-law, bless her simple lifestyle, would be telling me, "but I heard that Mercury is actually causing global warming. See, it's a natural cycle." Fifty years from now, were she still alive, she'd be telling us how we need to blow up Mercury instead of charging more at the pump (government conspiracy!).
    0 0
  16. It has been a decade since Scafetta’s 2011 projections were made, so it is time to update the temperature data to compare it to his projections.

    In the figure below I have updated his graph by adding the global surface temperature yearly
    average anomalies from NOAA/GISS. I’d say the IPCC models are pretty good, especially since the last few years have been La Nina years.
    Scafetta projection, the black wavy line, misses the temperature change over the decade by a third of a degree. Recent temperatures of 2015 to 2022 are already comparable to his projection for 2100.

    (The NOAA/GISS data is referenced to the 1900-1999 average so I adjusted it to the 1960-1999 reference that Scafetta used. The NOAA/GISS data average for 1960-1999 is 0.135, so I
    subtracted that from the data before plotting it as the black dots.)


    Update of Scafetta's Widget

    0 0
    Moderator Response:

    [PS] If you want to put an image into a comment, you need to first put the image into some suitable image server (eg imgur) and then insert a link. SkS doesnt have ability to host images despite the illusion of it doing so.




    0 0
    Moderator Response:

    [PS] if you use a image service that just serves the image, then you can use insert image, make sure width is constrained to 500, and the image will display in the comment. google photos is bad for this as links are to the photo viewer, not the image.

  18. Here is the update to Scafetta's Widget.

    Scafetta widget


    0 0
    Moderator Response:

    [PS] I have updated comment to the actual image, and removed failed attempts You again linked to a page, rather than the image in that page. You want link that has .png, .jpg etc in it.

    Additional information on adding image or html to a comment can be found on the comments policy page (at the bottom). The link to the comment policy is just under the comment editor.

  19. sailingfree @68,

    Note that the data for global temperatre you plot is NOAA data but not GISS (LOTI) data. Both use the same raw data but process them differently. Thus in NOAA 2015 is warmer than 2017 by +0.02ºC while in GISS 2015 is cooler than 2017 by -0.02ºC.

    And both have published the 2022 figure which slots into the record above 2021, although not by much in NOAA.

    As you rightly say, the present La Niña is depressing the global average SAT, a La Niña which is expected to end through the coming year, and expected to end a lot more suddenly than previous strong La Niñas (like 1988, 1999 & 2008 which were more sudden transitioning into La Niña and more gradual transitioning out).

    MEI.v2 La Nina evolution

    0 0
  20. MA Roger @69,

    Thanks for clarification of the data. And thank you moderator [PS] for fixing the mess I made with the URL.

    0 0
    Moderator Response:

    [PS] You are welcome. Thanks for your work

Prev  1  2  

You need to be logged in to post a comment. Login via the left margin or if you're new, register here.

The Consensus Project Website


(free to republish)

© Copyright 2024 John Cook
Home | Translations | About Us | Privacy | Contact Us