RealClimate logo

More on sun-climate relations

Filed under: — rasmus @ 9 March 2010

Four new papers discuss the relatiosnhip between solar activity and climate: one by Judith Lean (2010) in WIREs Climate Change, a GRL paper by Calogovic et al. (2010), Kulmala et al. (2010), and an on-line preprint by Feulner and Rahmstorf (2010). They all look at different aspects of how changes in solar activity may influence our climate.

The paper by Judith Lean (2010) has the character of a review article, summarizing past studies on the relationship between solar forcing and climate. The main message from her article is that the solar forcing probably plays a modest role for the global warming over the last 100 years (10% or less). It’s a nice overview, but I miss treatment of uncertainties.

Her analysis is based on the HadCRUT3 data, and I wonder if she would get similar results if she chose the GISTEMP or NCDC instead. The choice may in particular be relevant for the discussion of the temperatures after 1998.

Personally, I regard the data on solar activity before 1900 as quite uncertain too. The reason is that there are strange things happening to the solar cycle length in the shift from the 19th to the 20th century. Hence, any analysis based on the past centuries is uncertain because of suspect data quality in the early part of the record. Lean mentions that proxy-based records are uncertain, however.

Another source of uncertainty stems from the analysis itself – a regression analysis with chaotic data can easily yield misleading results. Gavin and I showed in a recent paper that multiple regression can produce strange results when applied to the global mean temperature and a number of forcings.

In other words, I think the reader may get the wrong impression from Lean (2010) that the link between solar activity and climate is better established than the data and methods suggest. Especially when she discusses forecasts for the near future (eg. for year 2014) – I fear that such a discussion can be misinterpreted and misused. However, that’s my view, and it does not necessarily mean that her paper is incorrect – quite the opposite, I think her main conclusions are sound (Her estimate of the solar contribution to the global warming over past century – 10% or less – is in good agreement with the figure Gavin and I got in our analysis).

The positive side is that the paper is probably clearer and more accessible without all these caveats. I also think she makes an interesting point when she discusses ‘fundamental puzzles’ associated with claims of strong solar role in terms of the past warming. She puts this into the context of climate sensitivity, arguing that it would imply that Earth’s climate be insensitive to well-measured increases in GHG concentrations and simultaneously excessively sensitive to poorly known solar brightness changes. Furthermore, Lean argues that it would also require that the Sun’s brightness increased more in the past century than at any time in the past millennium – a situation not readily supported by observations.

The paper of Calogovic et al. (2010) is a follow-up of a recent paper by Svensmark et al. (2009), looking into the claim that the cloud water content drops after a Forbush event. Their work involved estimating cosmic ray fluxes for the whole planet, and comparing it to local cloud information derived from satellites. They concluded that the Forbush events had no detectable effect on the clouds.

Moreover, they also argued that the analysis of Svensmark et al. (2009) gave unreliable results since it included a Forbush event on January 20, 2005 which was accompanied by a strong solar proton event. However, they did not explain explicitly why such proton events would disturb the measurements, but referred to another study by Laken et al. (2009) in Geophysical Research Letter. Laken et. al. only discusses the proton events briefly, and refers to a study by Fluckiger et al. (2005), who state that “The cosmic ray ground level enhancement (GLE) on January 20, 2005 is ranked among the largest in years, with neutron monitor count rates increased by factors of more than 50”.

But there is no reference to proton events in Fluckiger et al. (2005), so I’m not convinced that proton events will invalidate the analysis of Svensmark et al. (2009). Perhaps I’m missing something? Anyway, this is only a minor detail, and the rest of the analysis of Calogovic et al. (2010) seems more convincing. Their conclusion is supported by Kulmala et al. (2010): “galactic cosmic rays appear to play a minor role for atmospheric aerosol formation events, and so for the connected aerosol-climate effects as well”. Kulmala’s group in Finland boasts many world-renowned aerosol physicists.

The study by Kulmala et al. (2010) was based on near-ground measurements of aerosols, magnetic field, cosmic rays, sunlight intensity (solar radiation), and ionization over a 13-year long period (~1 solar cycle). They also used airborne Neutral cluster and Air Ion Spectrometer, LIDAR and Forward Scattering Spectrometer Probe measurements. They failed to detect any correlation between cosmic ray ionization intensity and atmospheric aerosol formation.

Feulner and Rahmstorf address a speculation stated by Lean: the possibility of solar forcing countering anthropogenic global warming. Their paper examines the effect a solar grand minimum (low solar activity similar to that inferred for the Maunder Minimum) would have on the global mean temperature by 2100. By accounting for a corresponding reduction in forcing for the future in a climate model study, they conclude that the effect is negligible (less than 0.3K compared to 3.7 – 4.5K if the SRES A1b or A2 emission scenarios were assumed).

So what can we learn from these articles? What we see is how science often works – increases in knowledge by increments and independent studies re-affirming previous findings, namely that changes in the sun play a minor role in climate change on decadal to centennial scales. After all, 2009 was the second-warmest year on record, and by far the warmest in the southern hemisphere, despite the record solar minimum. The solar signal for the past 25 years is not just small but negative (i.e. cooling), but this has not noticeably slowed down global warming. But there are also many unknowns remaining, and the largest uncertainties concern clouds, cloud physics, and their impact on climate. In this sense, I find it ironic that some people still rely on the cosmic rays argument as their strongest argument against AGW – it does involve poorly known clouds physics!

192 Responses to “More on sun-climate relations”

  1. 1
    Edward Greisch says:

    “The Forbush decrease is usually observable by particle detectors on Earth within a few days after the CME, and the decrease takes place over the course of a few hours. Over the following several days, the solar cosmic ray intensity returns to normal.”
    CME=coronal mass ejection from the sun.
    A few days is weather.

    ” solar forcing a global warming is 7 ± 1% for the 20th century” 7% of what kind of degrees? Kelvin? That seems large to me. I can’t see your whole article. Please expand on this.

  2. 2
    meteor says:

    Hi Rasmus

    The 11 years cycle involves a 0.1K (from min to max) global temperature variation.
    To explain it, we need an amplification factor of the solar radiative forcing.
    This solar RF is only 1W/m2*0.25*0.7 = 0.175W/m2.
    So with modeling thermal inertia (a sort of oceanic diffusive model) this amplification factor must be about 3.
    This doesn’t change very much the solar influence in the current warming (about 10-15%) but how can we explain this amplification factor?
    You say that there is no slowdown of climate warming during the last years.
    It isn’t true.
    There is really a slowdown, and the decreasing of solar activity, with the solar amplification coefficient, may explain an important part of this slowdown (with ENSO,…)between 2003 and 2009.
    I don’t understand why you are so blind…

  3. 3
    Steve Carson says:

    I’ve only read a few papers on the solar reconstructions, including Solanski & Fligge (1998). Everyone works out quite detailed reconstructions – we all want to know TSI over the past 100, 300, 1000 years. But as Solanski & Fligge note, their reconstructions all rely on the assumption that the measured relationships have remained unchanged over more than a century. And in reading other papers similar caveats seem to apply.

    Pre-1978 can we really place much confidence on the estimated TSI values? Or only say “so long as recently measured relationships haven’t changed in the past” which to some might say “very confident” and to others might say “pick your own number”?

  4. 4
    MR SH says:

    Thank you for good post. These are really good examples how the science community contributes to the understanding of the behavior of nature.

    I have hoped such new findings rather than the confusion of academic society (IoP) which might again harm the reliability of science.

  5. 5
    Bob Tisdale says:

    rasmus: When you wrote, about Judith Lean’s choice of HadCRUT3 data, “The choice may in particular be relevant for the discussion of the temperatures after 1998,” were you referring to the fact that HadCRUT3 is biased high after 1998 due to the change in the source of their HADSST2 data in 1998? That change created a shift that appears in no other SST dataset. Here’s the difference between global HADSST2 and HADISST for example.
    The 1998 shift occurs if one compares HADSST2 to ERSST.v2 or to ERSST.v3b or to OI.v2 SST datasets.

  6. 6
    Gilles says:

    “Her estimate of the solar contribution to the global warming over past century – 10% or less – is in good agreement with the figure Gavin and I got in our analysis”

    I’m still perplex concerning the first warming period in the XXth century, namely the 1900-1940 period. What has been the solar contribution to this warming ? what was the weight of other possible contributions? As a matter of fact, people around 1940 would have been entitled to say the same kind of things as us : the warming is unprecedented, this is the warmest decade of the whole instrumental period, proxies show that such a warming is exceptional in the last millenium, etc, etc…..

  7. 7

    Aw, heck! I rely on multiple regression for most of my analyses!

  8. 8


    The slowdown is not statistically significant. It takes 30 years to establish a climate trend.

  9. 9

    Nice discussion of the papers, thanks. Good to get back to the science after all the defence against propaganda attacks.

  10. 10
    Icarus says:

    meteor says:

    “You say that there is no slowdown of climate warming during the
    last years. It isn’t true. There is really a slowdown, and the
    decreasing of solar activity, with the solar amplification
    coefficient, may explain an important part of this slowdown
    (with ENSO,…)between 2003 and 2009”

    I wonder whether some of this apparent ‘slowdown’ is due to accelerated melting of the Greenland and Antarctic ice sheets (Velicogna 2009)? Presumably if some of the energy in the climate system is going into melting ice then there will appear to be less warming, if measured as global average temperature, but the total energy content of the climate system will not necessarily show such a slowdown.

  11. 11
    rasmus says:

    The interesting part seems to be that the early century warming seemed to take place mostly in the northern hemisphere high-latitudes, whereas the warming since 1980 is more uniform over all latitudes. Any theory explaning both these periods ought to account for this difference. -rasmus

  12. 12
    rasmus says:

    I mainly had the differences in the global mean temperature in mind:

  13. 13
    Hank Roberts says:

    > We demonstrate that naive application of linear analytical methods
    > such as regression gives nonrobust results.
    (from your paper)
    I hope a competent statistics blogger picks this up and does some tutoring, it would be interesting to watch.

  14. 14
    Kav says:

    Not had chance to read those papers yet but is the reference to solar proton events (SPE) to do with ground level events? Some of the larger SPE (sometimes known as solar energetic particle events – SEPs – as they contain more than protons) are have sufficient high energy fluxes to be detected on the ground in cosmic ray detectors as ground level events.

    Thus if the SPE on 20 Jan 2005 was sufficiently large (I’d have to check) it would skew any estimates of cosmic ray flux for that period. I think I am right in saying that the highest energies tend to peak towards the start of the event and most proton events precede the geospace effects of the CME as they move significantly faster through the solar wind. No time to dig out references now but will try to supply some later if anyone wants them (otherwise google is your friend).

  15. 15
    Tim Jones says:

    Re: 2 meteor says: 9 March 2010 at 3:15 AM
    “There is really a slowdown, and the decreasing of solar activity, with the solar amplification coefficient, may explain an important part of this slowdown (with ENSO,…)between 2003 and 2009.”

    A strong La Niña, stratospheric water vapor and the low ebb in the sunspot cycle did indeed contribute to the cool winter of 2008. But with the highest
    global land surface temperature being in 2005 and 2009 being the second highest as the world emerged out of the low ebb in the solar cycle and the Na Niña, how does this infer we’re seeing a cooling trend?

    The past year was a small fraction of a degree cooler than 2005, the warmest on record, putting 2009 in a virtual tie with a cluster of other years –1998, 2002, 2003, 2006, and 2007 — for the second warmest on record.

    Natural variability doesn’t go away with aghg climate forcing. The temperatures of the last decade never averaged less than or even equal to the temperatures of the previous decade. The lowest temperature of the last decade was about equal to the highest low temperature of the previous decade.

    No matter how you twist and squirm with the facts greenhouse gases are keeping temperatures elevated at an elevating rate. There is no cooling trend.

  16. 16

    In your opinion, does any of the information in these articles explain the melting of the polar ice caps on Mars, or is that merely due to dust storms on the Red Planet?

  17. 17
    meteor says:


    I don’t speak really about a trend.
    We must not focalize on this.
    I repeat that the consensus appears to be 0.1K (between solar maxi and mini)
    How can you see this?
    How can you see ENSO influence?
    If not in studying global temperature.
    So, between 2003 and 2009 there is ABOUT +0.15K/dec *7 = +0.1K from GES
    at least -0.1K for solar influence.(at least because a very deep minimum in 2008-2009 so it’s more likely between -0.1K and -0.15K)
    For this period we can consider that solar influence, at least, counterbalances the anthropogenic influence.
    OK there is also other climatic variability. (ENSO for example)

    Now, when you look at the “trend” (sorry) (NASA-GISS for 2003-2009) you get 2009 (the end of the linear regression) 0.02K colder than 2003.(the beginning of the linear regression)
    This is not too bad.
    But my question to Rasmus was “Why 0.1K for only 0.175W/m2 of solar forcing without a amplification solar coefficient?”
    Have you, or no, an idea on this, Rasmus?

  18. 18
    meteor says:

    Sorry, GES = Gaz à Effet de Serre = GHG in English

  19. 19
    Ray Ladbury says:

    Scott W. Somerville,
    There is a global climate model for Mars. Dust storms are one of the most critical factors and are the main explanation of the differential melting of the Maritan polar caps.

  20. 20
  21. 21
    Completely Fed Up says:

    Scott, the problem is your question is so vague it can be answered, but the only way to answer it with less than an essay is to answer it with something that is accurate but worthless.

    If the sun had done something different, the melting would have been different.

    Worthless as an answer, though.

    A better question would be:

    “I’ve read about melting ice caps of Mars and wonder whether the changes in the sun from these articles are sufficient to explain why they are melting, or are they overwhelmed by, for example, the dust storms on Mars?”

    That starts with the predicate, the axiom you’re working from (the ice caps are melting on Mars).

    It then proposes a postulate (does this explain the melting, or is it something else).

    Your question doesn’t actually have any postulate.

    Answering it doesn’t tell you anything more than you knew before it, unless the change in mars is contrary to the changes the sun is introducing.

  22. 22
    stevenc says:

    It seems to me the uncertainty in solar radiation is fairly great. The study by Willson indicates the underlining radiation may be going up by as much as 0.05% per decade. Over 100 years it may account for a 0.5% increase. At (1366 watts/meter^2)(0.005) this could account for 6.83 watts/meter^2 or nearly 2C of warming. I did this fairly roughly as I try to concentrate on concepts not details so perhaps my math is severly in error or I have made some other fundamental mistake?

  23. 23
    Hank Roberts says:

    CFU, _look_ at the guy’s website before taking that bait further. Please.

  24. 24
    John E. Pearson says:

    22: stevenc . Using your numbers the stefan boltzmann law implies:

    (T+dT)^4/T^4 = (1+dT/T)^4 ~= 1+4dT/T = 1373/1366=1.005

    T is the equilibrium temperature without the 7W/m^2 additional forcing and T+dT is the equilibrium temperature with the additional 7W/m^2 of forcing.

    In any event we find: dT/T ~= .00125

    Pick the unperturbed T however your want within reason say 250K, 300K whatever. You don’t get a dT of 2C. You get a dT of around .2-.3K.
    You’re off by a factor of 10. “Details” matter.

  25. 25
    Oxford Kevin says:


    What study by wilson? In which peer reviewed journal?


  26. 26
    Pete Dunkelberg says:

    Is the “cooling trend” having a rough year?

  27. 27
    Completely Fed Up says:

    Or, indeed, Lamont (#20).

  28. 28
    stevenc says:

    John Pearson, thank you, I thought there was probably something wrong with what I did to come up with such a high value. I understand details matter and I usually avoid engaging in them for the reason just illustrated.

  29. 29
    stevenc says:

    Kevin, I read that in a press release from NASA dated 20 March 2003. It stated the study would be continuing at least 4-5 years into the future so I am not sure the study has even been completed yet.

  30. 30
    andreas says:

    @ stevenc (#22)

    I’m just an amateur, but I think, you have to divide 1366W/m^2 by a factor 4 or so before. Think of the “dark side” of earth and other angular corrections.

    Could anybody respond to #17? I think, it’s a good question.

  31. 31
    Sekerob says:

    Pete Dunkerberg, 26:

    Looked in past at this… yes 2010 is cooking on AMSU, BUT, select 4400 mtr alt (Ch 5), 1999, 2005, 2007,2010 + 20 year hi/low/record than hover over any data point or hold mouse button for a pop up box of temps for the date. Something wrong with the 1999 values or the 1999 curve.

  32. 32
    Sekerob says:

    Oh wait, duh, the temp scale is negative :O

  33. 33
    Completely Fed Up says:

    what was supposed to come before 27 got et.

    Hank, what about Ray? Don’t you care about him too?

  34. 34
    GFW says:

    Reply to stevenc:

    Willson himself says that compared to greenhouse gasses “Solar forcing would provide only about one-fourth as much warming, if the solar trend persists over the same period,” Willson said. “Solar forcing could be significant, but not dominant.”

    Further, I looked at Willson’s presented graphs of TSI, and he rejects the PMOD corrections to TSI despite the preponderance of evidence that PMOD is likely more correct than ACRIM. See

    Willson’s graphs also (so far that I can find) seem to stop short of the recent relatively deep minimum. So his analysis is as “pro solar” as one can get, probably overstates the solar case, and he still says it’s a quarter the effect of greenhouse gases.

  35. 35
    Philip Lloyd says:

    The Feulner and Rahmsdorf link takes me to a members page, not to the paper.

  36. 36
    Hank Roberts says:

    Just looking it up, that’s probably:
    Wilson: Secular total solar irradiance trend during solar cycles 21-23
    cited by 86; skimming, I don’t see any big excitement about this. Experts?

  37. 37
    Hank Roberts says:

    Paywalled at AGU:

    Feulner, G., and S. Rahmstorf (2010),
    On the effect of a new grand minimum of solar activity on the future climate on Earth, Geophys. Res. Lett., doi:10.1029/2010GL042710, in press.
    [PDF] (accepted 5 February 2010)

    Some info here:

  38. 38
    Jacob Mack says: Though I do not always agree with Hansen’s predictions, we cannot argue with the warmer 2009, or more importabtly a continuation of the warming trend.

  39. 39
    R. Gates says:

    Thanks for the excellent summary of these new papers. I’m curious, given the apparent relative contribution of solar influences vs. GHG, how much role did fact of that the last solar minimum ended last year and the the sun is heading upwards to the next solar max play in the Met office predicting that 2010 will be the warmest year on record? What I’m getting at, is suppose we were in a waning solar cycle, heading toward a solar minimum, would the Met Office still have been so sure of the record warmth for 2010?

  40. 40
    Septic Matthew says:

    I referred to Judith Lean’s paper approvingly a couple days ago. It’s nice as far as it goes.

    It seems that for solar and GHG influences, the earth’s response lags behind the forcing: the full effect of the GHG increase of the 150 years has not yet occurred, and the measures of solar activity that correlate most with earth temperature change are integrated measures (areas under the solar activity curve) of years past. For these reasons, I think that the vector autoregressive analyses (regression on lagged variables) will prove, eventually, to be better for making predictions than the multiple linear regressions on contemporaneous measures. The best to date seems to be the non-linear vector autoregressive analysis by Beenstock and Reingewertz (previously discussed here at Real Climate), but it hasn’t passed peer review yet. If they do publish it in Nature they will put their data and code on line, and we shall be able to examine it all in detail.

    For both the solar theory and the AGW theory without solar, the (partial, reduced, whatever) warming of the last 15 years is within the range of uncertainty of the imprecise predictions, so neither is rejectable with high confidence (or a low statistical significance.) As has been frequently asserted, decades more data are necessary to determine what apparent trends of recent decades are going to persist.

  41. 41
    Septic Matthew says:

    I meant to write “AGW theory with or without solar”. Sorry.

  42. 42
    David B. Benson says:

    Gilles (6) and others — The so-called 1910-1940 anomolous warming shows up clearly in the AMO:
    and using about 1/3 of the decadal average of the AMO together with CO2 gives very good agreement with GISTEMP:
    My (poor) understanding is that AMO is an index for MOC rate, mostly, but all frocings not contributing to the removed linear trend are there as well.

  43. 43
    Pete Dunkelberg says:

    Breaking! “—–X—–” shows a cooling trend since its last maximum! Yet mysteriously, each decade is overall warmer than the last.

  44. 44
    Bob Tisdale says:

    11 rasmus: You wrote, “The interesting part seems to be that the early century warming seemed to take place mostly in the northern hemisphere high-latitudes, whereas the warming since 1980 is more uniform over all latitudes.”

    The Zonal Mean linear trends for the latitudes of 65S to 20N are remarkably similar for the periods from 1910 to 1940 and from 1980 to 2009. It’s above 20N and 65N that they diverge.

    Data through the GISS map making webpage:

    But, yes, it is interesting.

  45. 45
    Milan says:

    From what I understand, the paleoclimatic record supports the view that solar changes could not be causing the climate change we are observing.

    The geological era spanning the last 65 million years is called the Cenozoic. Over that time, the sun’s output has increased by 0.4%. This corresponds to an increase of about 1 watt since the dinosaurs died out. Over this time period, the planet has actually cooled considerably: with mean global temperature more than 8°C higher at the end of the time of the dinosaurs. This, despite the increased solar output.

    Over this timespan, the atmospheric concentration of CO2 has ranged from between 1,000 and 2,000 ppm during those hot years of the early Cenozoic and as little as 170ppm during recent ice ages. This range corresponds to a climate forcing of about 12 watts: at least ten times more than the forcings from the sun and from changes in the configuration of continents. As James Hansen says: “It follows that changing carbon dioxide is the immediate cause of the large climate swings over the last 65 million years.”

  46. 46
    Hank Roberts says:

    re #17, how to fool yourself about trends
    “… here are four different trend-lines you can get from the same data (UAH temperature) just by choosing your periods carefully” and page down to the slider, move it back and forth to see what trend you get with different timespans.

    Shorter timespans are less reliable at finding a _real_ trend in the noise.
    Statistics 101 teaches this lesson; thereafter the world looks different.

  47. 47
    Frank Giger says:

    Hurray! We’re back to science!

  48. 48
    JamesA says:

    > Could anybody respond to #17? I think, it’s a good question.

    I’m not convinced it is. Picking out specific signatures in global temperature trends is a pretty hairy business (not that that’s stopped all the various have-a-go statisticians from trying over the years). Part of the problem is that you can’t simply expect the rate of change of the surface temperature products to follow the sum of the radiative forcing components at any one moment in time, even if you do allow a nominal lag for the oceans etc. If it was that simple, they wouldn’t need the hellishly complicated GCMs that they have.

  49. 49
    David B. Benson says:

    Septic Matthew (40) — Well, instead use the known physics and some estimates of future global warming (so-called greenhouse) gases and other forcings.

  50. 50
    John E. Pearson says:

    48A: JamesA said maybe it isn’t such a good question.

    I agree. Anyone who thinks it is a needs to read (and understand) Gavin’s “Climate Change Commitments” thread.