Open Mind

How to Fool Yourself

October 21, 2007 · 22 Comments

I posted some time ago about the recent paper by Lockwood and Frohlich (Proc. R. Soc. A, doi:10.1098/rspa.2007.1880), showing that recent trends in solar activity can’t explain modern global warming, because they’re going in the wrong direction. It so happens that Svensmark and Friis-Christensen have issued a reply to Lockwood and Frohlich. Notice I haven’t given a journal citation for their reply; that’s because it’s not published in the peer-reviewed literature.

Let’s be clear what Lockwood and Frohlich’s (hereafter L&F) claims are. They don’t say that solar variations have no impact on climate, in fact they say the opposite. They’re not talking about the solar cycle except to mention that surface temperature “does not respond to the solar cycle.” Rather they’re discussing the trends — changes on time scales longer than the solar cycle — and show convincingly that for explaining modern global warming, those trends are going in the wrong direction. And they don’t just discuss the radiative output of the sun (total solar irradiance, or TSI), they study the trends in cosmic rays as well.

Svensmark and Friis-Christensen (hereafter S&F) state in their first paragraph that they will “rebut their argument comprehensively.” They begin by attempting to establish a response to the solar cycle, in temperatures in the troposphere and ocean. So I have to wonder at the outset, why are they changing the subject? L&F’s thesis is the impact of trends, not the solar cycle, on surface temperature, but S&F change the subject from trends to the solar cycle, from surface temperature to other variables.

I’m also surprised that they do such a poor job of it. They first attempt to establish a solar-cycle signal in tropospheric temperature in the top of figure 1:


To the eye, there seems to be correspondence from about 1980 to 1992, but very little correspondence for the remainder of the time span. And I’ve done enough analysis to know how easily the eye can fool us; that’s why we apply statistical tests. Unfortunately S&F don’t report any statistical evaluation of this correspondence.

Much if not most of the apparent correspondence is due to the temperature decreases that occured around 1984 and 1992. But we know that those decreases aren’t due to the solar cycle or changes in cosmic rays — they’re due to the eruptions of the el Chicon and Mt. Pinatubo volcanos. Maybe S&F just didn’t think of this? But that can’t be, because they follow their discussion of figure 1 with discussion of figure 2, “an analysis of tropospheric temperatures for the European Space Agency’s ISAC project.” Their conclusion? “After the removal of confusions due to El Nino, volcanoes etc. and also a linear trend, as in the middle panel of Fig. 2, the negative correlation between cosmic-ray flux and tropospheric air temperatures is impressive.”


Unfortunately they don’t say how they remove the “confusion” due to el Nino and volanic eruptions, or what “etc.” is. But they’re surely aware of the impact of volcanic eruptions. This is problematic because it looks like after removing the impact of volcanic eruptions, the cooling in 1992-1993 associated with the Mt. Pinatubo explosion actually gets bigger! In fact I digitized the graphs to confirm this (which is necessary because the graphs are on different scales). How can removing the cooling effect of volcanic activity make the cooling larger? If they told us how they removed non-solar factors, what data they used, and what the elusive “etc.” is, maybe it would be clear. But they don’t.

The ensuing discussion reveals at least part of the reason for their focus on the solar cycle:

When the response of the climate system to the solar cycle is apparent in the troposphere and ocean, but not in the global surface temperature, one can only wonder about the quality of the surface temperature record.

This is the number one favorite method of denialists world-wide: when the data contradict your hypothesis, cast doubt on the data. Yet having implied that the surface temperature record can’t be trusted, they then use it to claim that global warming has recently halted. They first say

In any case, the most recent global temperature trend is close to zero.

Lest you think they’re not referring to the surface temperature record they’ve just disparaged, they follow by saying

Lockwood and Frohlich erase the solar cycle from various data sets by using running means of 9 to 13 years. While these may be appropriate to illustrate trend reversals since about 1985 in various indicators of solar activity, in the case of global mean surface temperatures the use of a long running mean creates the illusion that the temperatures are still rising rapidly early in the 21st Century. Their Fig. 3f (ref. [1]) suggests a remarkable 0.1 K increase between 1998 and 2002, when the curves terminate. In reality, as shown in the unsmoothed presentation of monthly data in their own Fig. 1e, global surface temperatures have been roughly flat since 1998.

Yes, they’re talking about the surface temperature record, the same one used by Lockwood & Frohlich. They quite ignore that this time frame begins with the strong el Nino of 1998. It seems like quite a double standard, to remove the “confusion” of el Nino when you want to support a solar-cycle influence on tropospheric temperature, but exploit the effect of el Nino to discredit surface temperature warming.

This is especially ironic because even choosing the optimal “cherry-picking” start year 1998, their claim is still false. A statistical analysis of the data shows beyond doubt that global surface temperatures have been nowhere near “roughly flat” since 1998, instead they show a very strong and undeniably statistically significant warming. In fact the warming is statistically significant even without removing the impact of el Nino! The “global warming stopped in 1998″ mantra is provably false but will never die; you can only make that claim by ignoring a proper statistical analysis.

There are other problems too. Their table of trends in cosmic rays, tropospheric temperature, and ocean temperature includes error estimates, but it’s clear that these error estimates are far too low because they’re based on assuming that the random part of the data is a white-noise process. But we know that it’s a red-noise process, so the probable errors in trend rates are much higher, which casts considerable doubt on the conclusions of S&F based on them.

S&F begin by saying that they’ll “rebut their argument comprehensively.” But it’s all smoke and mirrors; they don’t even attempt to rebut L&F’s conclusion that recent trends in solar activity are in the wrong direction to cause global warming. They claim that surface temperature since 1998 is “roughly flat” with no statistics to back it up — which is understandable because the claim can easily be proved wrong. So when it comes to the two main theses of L&F (solar trends in the wrong direction, surface temperature warming), the reply from S&F ignores one and makes a blatantly false claim about the other.

But it may surprise you to learn that I am not convinced that S&F are being deliberately disingenuous. It seems to me that they truly believe their alternate hypothesis, and have fallen prey to the ease and eagerness with which we accept evidence that supports our pet theory uncritically. They’re fooling themselves.

We are not impressed. But it will give Senator Inhofe yet another “scientific paper” (albeit not peer-reviewed) to link to in his anti-global-warming blog.


A reader asks

Would you be so kind as to plot atmospheric CO2 levels against these recent temperature fluctuations. I have a hunch it correlates a lot worse than even the unadjusted cosmic ray plot.

The full comment, and another comment (on another thread) clearly indicate the reader doesn’t believe in AGW.

The question itself reveals ignorance of the nature of climate change. In fact, it illustrates another way to fool yourself (and other people): apply a simplistic view of climate change, the naive belief that temperature has to follow CO2 concentration faithfully, as though CO2 were the only forcing which impacts global temperature.

NASA GISS provides climate forcing data, and it includes a lot more than greenhouse gases. Other climate forcings estimated by GISS include ozone, stratospheric water vapor, land-use changes, snow albedo, stratospheric aerosols, black carbon, reflective aerosols, and the aerosol indirect effect. Climate responds to all these forcings; only the uninformed expect it faithfully to follow any single forcing.

In case you’re wondering how global temperature compares to the total climate forcing as estimated by GISS, here’s a superimposition of both:


The large, but short-lived, downward spikes in forcing are the cooling effect of volcanic eruptions. One factor which is not included in the GISS estimates of climate forcing is el Nino/la Nina. That’s because the GISS data are meant to represent the external climate forcings, to be input to coupled ocean-atmosphere climate models. The el Nino/la Nina phenomenon is not an external forcing to the ocean-atmosphere system, it’s an exchange of heat between subsystems which are internal to the system.

Categories: Global Warming · climate change

22 responses so far ↓

  • BrianR // October 21, 2007 at 4:56 pm

    Thanks for another straightforward post … other climate blogs that include tons of detail (i.e., lots of graphs, tables, and acronyms) do a horrible job of connecting the dots and communicating the significance. And when I ask for clarification or a comment about what it all means, I get chased away by loyalists chiding me for being skeptical of their skepticism. Good times.

  • inel // October 21, 2007 at 5:11 pm

    one can only wonder

    Recently, a teenager declared L&F had been well-and-truly debunked (by S&F). I had read L&F but not thought it worth spending time on S&F, and in any case, I was not the intended audience for the remark, as a fly on the proverbial wall (!) at the time.

    Now, thanks to you, at least I have some idea as to how not-so-well though ever-so-truly S&F managed to convince themselves, and by extension, others.

    P.S. I would appreciate your input on a completely different topic that bothers me, after hearing students’ reactions to AIT, TGGWS, Monckton’s latest con, etc.

  • Frank Rizzo // October 21, 2007 at 6:19 pm

    As you say, S&F-C are failing to see the wood for the trees. Lockwood & Frölich’s article highlighted that warming in the last 20 years couldn’t be explained by the sun’s trends. It’s obvious to everyone that the Sun’s impact on climate didn’t magically disappear in the 80s - which is what S&F-C have shown here.

    Oddly, S&F-C go beyond their point and decide to take some random swipes at the surface temperature record. Also, I’m puzzled why they decide to remove the “confusion” of El Niño in their graph (Fig 2), but then proceed to neglect its effect when they more or less claim that warming ended in 1998. Both lend a rather cynical tone to the article that doesn’t add much to the science.

  • Steve Bloom // October 21, 2007 at 10:06 pm

    BrianR, bear in mind that most climate scientists are specialists in one or another areas such that they don’t have time series analysis skills at their fingertips. So when they are asked to engage in what is for most of them a fairly time-intensive task of proving a point which is (to them) obvious and will be completely lost on most of the audience, they tend to be resistant. We are fortunate to have Tamino to fill the gap.

    Also, as anyone who has attended college has experienced, specialist knowledge and communication skills often do not have perfect overlap. Equally fortunately Tamino has the latter as well.

  • John Mashey // October 21, 2007 at 10:07 pm

    And for more, perhaps related statisticalist fantasy, the littliest denier, Kristen Byrnes is still at it in ponderthemaunder, as of 10/20.

  • BrianR // October 21, 2007 at 11:52 pm

    I was mostly referring to those climate blogs out there by non-scientists evaluating and discussing climate data … they do not communicate what they are doing effectively (which may be part of their goal anyway)

  • D // October 22, 2007 at 7:53 am

    Sunspot numbers for pretty much the first time dont seem to be following the temperature records, however new tree ring data is doing the same - maybe as we cross check we might find that everything correlates except the smooth temperature data - hmmm I wonder - could the numbers be being fiddled or is the sun, the trees, the environment etc all out of whack?

  • luminous beauty // October 23, 2007 at 12:15 am


    You make a good point. The environment is out of whack.

    We did it. We Homo sapiens have stressed our environment in ways that only huge cataclysmic events like giant asteroids and super volcanoes have exceeded.

    It’s not just greenhouse gas. There is a host of other pollutants as well. Land use changes account for a significant proportion of temperature increase ( and incidental environmental stress on eco-systems as well), the least significant of which is the urban heat island effect.

    There are other stressors besides temperature and moisture that affect tree growth, like soil PH. Hmmmmmm, I wonder how that correlates?

    Temperature is only a mean proxy for but one aspect of the complex inter-play of the changing attributes of environmental systems.

  • JamesG // October 24, 2007 at 3:06 pm

    Would you be so kind as to plot atmospheric CO2 levels against these recent temperature fluctuations. I have a hunch it correlates a lot worse than even the unadjusted cosmic ray plot.

    Also, are you agreed with the assertion that the Sun was probably the main influence up to around 1985 even though we don’t know the exact mechanism, which is roughly the Solanki position?

  • cce // October 24, 2007 at 5:06 pm

    I believe anthropogenic forcing exceeded solar forcing in the late 1800s. 1985 happens to be the year that all of this stuff peaked, and since then has been going in the opposite direction, while the temperatures have continued to go up.

    And since the climate is influenced by many things, not just CO2, the proper way to judge our understanding of climate is to look at model runs of the past which do a very good job of recreating past temperatures.

    If you want something simpler, you can look at Tamino’s recent “Many Factors” post. It looks at the contributions from three factors: increasing GHG, el Nino, and volcanoes from 1975. Any contributions from the solar cycle are lost in the noise.

  • Alan Woods // October 24, 2007 at 11:42 pm

    Tamino - a question about the GISS climate forcing you have faithfully plotted. As this is modelled forcing, and is calculated from theoretical estimates of forcings from a number of variables, and that the IPCC ascribes very low confidence in our ability to attribute a quantitative effect to many of these forcings, what should the cnfidence intervals be on that graph?

  • JamesG // October 25, 2007 at 9:13 am

    Alas your beliefs don’t coincide with the available data, or indeed with the views of many climate scientists (eg Hadley centre, NASA, NCAR) or with any solar scientist. The sun’s effect, in some way or another, is actually the only plausible explanation for climate shifts in the past from the last ice age up to 1985 and probably for the multiple ice-age cycles too. There are also models for this - see the NASA earth observatory website - which explain the past rather better than alternative models. Hence trying to establish a sun connection for the recent warming may not turn out to be correct but it clearly bears close examination, if only to rule it out.

    Saying there are multiple factors is a cop-out.
    You could also have said “the climate is influenced by many things, not just cosmic rays”. So why does the solar/cosmic ray theory have to show a one-to-one correspondence with temperature but CO2 doesn’t? It’s just plain double standards.

    Yes the direct sun effect is lost in the noise but the sunspot correlations are so striking that it has long been obvious that there must be an indirect effect which is not yet established. This is why the cosmic ray theory came about in the first place. It may not be 100% correct but it still fits quite well - better than CO2. It is this unknown but undeniable indirect effect that has not been modeled previously in most models but has now been modeled by Hadley. And guess what? They now say it’s going to cool down for a bit due t natural variation.

  • tamino // October 25, 2007 at 2:22 pm

    Saying there are multiple factors is a cop-out.

    The truth is a cop-out?

    So why does the solar/cosmic ray theory have to show a one-to-one correspondence with temperature but CO2 doesn’t? It’s just plain double standards.

    That’s a straw man argument; I never said anything like it.

    Yes the direct sun effect is lost in the noise but the sunspot correlations are so striking that it has long been obvious that there must be an indirect effect which is not yet established. This is why the cosmic ray theory came about in the first place.

    I don’t see that the sunspot correlations are “so striking” — at least not from this paper by Svensmark & Christensen. I have seen credible evidence of solar cycle influence on upper-atmosphere layers, and on prevailing wind circulation patterns, in other works, but it’s entirely possible they’re due to solar-cycle variations of TSI or UV flux; I have yet to see persuasive evidence of its influence on surface temperature.

    There’s no imperative to postulate another mechanism. The cosmic ray theory was postulated *because* it provides a mechanism to link the sun to modern temperature increase.

  • cce // October 25, 2007 at 3:06 pm

    My “beliefs” correspond to those of mainstream climate science. The sun obviously powers the climate system and most everything else on earth. It cannot, however, explain the changes of the last 30 years, nor has the changes in solar output (solar forcing) come any where near that of anthropogenic contributions for a long time. That is what NASA or Hadley will tell you.

    The climate is determined by many things. That’s not a copout, that’s a fact. Our understanding of GHG, aerosols, and solar explains the temperature changes of the last 100 years. Cosmic rays do not. It is not a double standard to treat failed theories as such. Maybe there is an effect over long timescales, but as an explanation for the warming since 1975, it fails.

  • JamesG // October 25, 2007 at 3:23 pm

    Well the sunspot/temperature connection is summarised well in wikipedia under “solar variation”. But there is no doubt that if the Hadley/GISS graphs are correct then there is no sun link to recent warming. S&C of course are questioning the accuracy of those graphs in this paper. Consider though that Soon has correlated the sun very well to Polyakov’s arctic temperature plot and the sun also correlates pretty well to the GISS US48 plot - which is likely the only part of the GISS data that is sufficiently well covered to be properly correctable. Also the Ulster observatory, with virtually the only unmoved, non-urban, longterm instrument data, shows a very good correlation to sunspots . Combined with S&C’s ocean/troposphere matches, thats 5 to 2 against and GISS/Hadley probably use the same sources which makes it really 5 to 1. Hadley and GISS of course are also in the climate modeling field which makes some of us a little suspicious. A decent 3rd party review of the datasets would work wonders for our confidence in these plots but that’s not likely to happen is it?

  • JamesG // October 25, 2007 at 3:46 pm

    As far as I can tell the aerosol data in the climate models has no real foundation or proof. It seems that it’s shape has been derived purely by taking the temperature record and subtracting the presumed GHG forcing effect. All other climate factors were clearly forced to cancel each other out. No wonder the combination matched post-1975; it was forced to match. Even that trickery though couldn’t explain the 1945-75 blip so they eventually had to admit that natural effects ie solar must have governed up to 1975. So much for accuracy - it’s just biased fudging. Hence it’s no more a proven theory than any other; at least until they can predict the future with some accuracy - which would be a first. Hadley made it easy for themselves though by predicting cooling in the near future. Nice.

  • cce // October 25, 2007 at 6:21 pm

    I would like to see where “they” admitted that solar “governed” up to 1975. Solar played a part in the first half of the 20th century. There has been little to no change in 50 years.

    And there is more than a little foundation that the Northern Hemisphere underwent rapid industrialization mid-century, resulting in a massive increase in sulfate aerosols and thus regional cooling. That increase leveled off in the 70s and ’80s and has since fallen.

  • John Finn // October 26, 2007 at 2:17 pm

    “And there is more than a little foundation that the Northern Hemisphere underwent rapid industrialization mid-century, resulting in a massive increase in sulfate aerosols and thus regional cooling. That increase leveled off in the 70s and ’80s and has since fallen.”

    Would you care to prove this. Show us industrialised regions which showed the cooling.
    Large scale Aerososl cooling is a nonsense. It’s simply a fudge factor to explain mid-20th century cooling

  • dhogaza // October 26, 2007 at 4:08 pm

    Large scale Aerososl cooling is a nonsense. It’s simply a fudge factor to explain mid-20th century cooling

    Ooooh, now THAT’S a convincing rebuttal of peer-reviewed science! Take that, you white-coated horn-rimmed glasses-wearing commie nerds, you!

    Mr. Finn, do you really think your post is an effective counter to the work linked to in the post you’re responding to?

  • JamesG // October 27, 2007 at 9:42 am

    You need to catch up with your reading. There are several recent papers on solar effects in climate models. No they don’t come out and say “solar governed up to 1975″ - they say weasel words like “internal variability” or “natural effects”. And the end of these natural effects is roughly 1970 at NCAR at least. Hadley split the solar effect with the aerosol effect to weasel it out a bit longer. Mark my words though that soon they’ll agree with Solanki that it is at least until 1980.

    Thanks for the link. Ok now I see the foundation we just need the proof. Aerosol experts tell us tough that the radiative effect of aerosols has huge error bars which makes it little more than guesswork. Far more work needs to be done to reduce these uncertainties. In the light of that, how can you consider models to be accurate? Bear in mind with solar we are comparing measurements with other measurements ie traditional science, but models with highly variable parameters and large error bands cannot be held to the same standards. Real, experimental proof is needed.

  • luminous beauty // October 27, 2007 at 3:23 pm

    “Real, experimental proof is needed.”
    The experiment we are running, we are running with the entire biosphere of the planet Earth. It is our home, we haven’t another. The results of sufficient proof such as you demand, must needs be waited for until the planet likely transformed in ways, we, it’s humble tenants are very, very likely to not find to our satisfaction.

    Thus, the use of models. (Actually all scientific theories are models.) We simulate the planet and run scenarios based on known physical features of known physical inter-actions of known physical components of the real Earth, guided by known maps of their real world spatio-temporal manifestations. I.e., exactly, ‘comparing measurements with other measurements.

    They do not perfectly reflect the real planet, but neither are they so wildly erroneous that they are useless.

    And they get better, the more we learn. Vague assertions about uncertainties and ‘highly variable parameters’, whatever that means, do little to contribute to making them better.

  • luminous beauty // October 27, 2007 at 3:36 pm

    John Finn,

    You could go to Wkipedia and begin your education on the well documented if not perfectly globally quantifiable phenomenon of Global Dimming, but judging from your posts I’ve read here and elsewhere, you are pre-disposed to dismiss, out of hand, any explanation that does not concur with your strongly held a priori convictions.

    There is a certain degree of quixotic heroism in the stance you are taking. But not a lot.

Leave a Comment