RealClimate logo

The CERN/CLOUD results are surprisingly interesting…

Filed under: — gavin @ 24 August 2011

The long-awaited first paper from the CERN/CLOUD project has just been published in Nature. The paper, by Kirkby et al, describes changes in aerosol nucleation as a function of increasing sulphates, ammonia and ionisation in the CERN-based ‘CLOUD’ chamber. Perhaps surprisingly, the key innovation in this experimental set up is not the presence of the controllable ionisation source (from the Proton Synchrotron accelerator), but rather the state-of-the-art instrumentation of the chamber that has allowed them to see in unprecedented detail what is going on in the aerosol nucleation process (this is according to a couple of aerosol people I’ve spoken about this with).

This paper is actually remarkably free of the over-the-top spin that has accompanied previous papers, and that bodes very well for making actual scientific progress on this topic.

The paper first confirms some results that are well known: aerosol nucleation increases enormously when you add H2SO4 into the air (the biggest contributor to human aerosol impacts via the oxidation of our emissions of SO2), it increases further when you add ammonia (NH3), and it increases even more when you increase ionisation levels from neutral, to ambient ground levels, and to upper atmospheric levels (as long as you are below what is called the ‘ion-pair’ limit). However, the most intriguing result is that despite going to a lot of trouble to make sure the chamber was ultra-free of contaminants, the researchers found that within most of the aerosols that formed, there were traces of organic nitrogen compounds that must have been present in almost undetectably low concentrations. The other intriguing finding is that aerosol nucleation rates in the chamber don’t match (by a an order of magnitude or more) actual formation rates seen in real world near-surface atmospheric layers at realistic temperatures (only in unrealistically cold conditions do rates come close). The authors speculate (quite convincingly) that this is precisely because they didn’t have enough volatile organic compounds (which are ubiquitous in the real world) to help get the nucleation started. This result will surely inspire some of their next experiments. All-in-all this is a treasure trove of results (and potential future results) for people tasked with trying to model or understand aerosol processes in the atmosphere.

Annotated Fig 5 from Kirkby et al
Figure 1: Annotated version of fig 5 in Kirkby et al. Small dots are in situ observations, lines are other lab data. Colours for the CLOUD results are coded with respect to temperature. Going from open to filled symbols denote increasing NH3. All results are for ambient CR ionisation (changes in CR only make a difference below the ion-pair limit).

However, aerosol nucleation experiments are not usually front page news, and the likely high public profile of this paper is only loosely related to the science that is actually being done. Rather, the excitement is based on the expectation that this work will provide some insight into the proposed cosmic ray/cloud/climate link that Svensmark (for instance) has claimed is the dominant driver of climate change (though note he is not an author on this paper, despite an earlier affiliation with the project). Indeed, the first justification for the CLOUD experiment was that: “The basic purpose of the CLOUD detector … is to confirm, or otherwise, a direct link between cosmic rays and cloud formation by measuring droplet formation in a controlled test-beam environment”. It is eminently predictable that the published results will be wildly misconstrued by the contrarian blogosphere as actually proving this link. However, that would be quite wrong.

We were clear in the 2006 post that establishing a significant GCR/cloud/climate link would require the following steps (given that we have known that ionisation plays a role in nucleation for decades). One would need to demonstrate:

  1. … that increased nucleation gives rise to increased numbers of (much larger) cloud condensation nuclei (CCN)
  2. … and that even in the presence of other CCN, ionisation changes can make a noticeable difference to total CCN
  3. … and even if there were more CCN, you would need to show that this actually changed cloud properties significantly,
  4. … and that given that change in cloud properties, you would need to show that it had a significant effect on radiative forcing.

Of course, to show that cosmic rays were actually responsible for some part of the recent warming, you would need to show that there was actually a decreasing trend in cosmic rays over recent decades – which is tricky, because there hasn’t been (see the figure).


Figure 2: Normalised changes in cosmic rays since 1953. There has not been a significant downward trend. The exceptional solar minimum in 2008-2010 stands out a little.

The CLOUD results are not in any position to address any of these points, and anybody jumping to the conclusions that they have all been settled will be going way out on a limb. Indeed, there is a lot of evidence that (particularly) point 2 will not be satisfied (see for instance, Pierce and Adams (2009), and a new paper by Snow-Kropla et al).

So what changes did they show as a function of the CR activity? In going from neutral (shielded) conditions to ambient CR levels typical of the lower atmosphere, the ionisation changed by a factor of 2 to 10 (depending on the temperature – colder conditions are more sensitive). However this is a much bigger change (by an order of magnitude or more) than the percentage change in CR activity over a solar cycle (i.e. ~10-20%). A rough calculation (by way of Jeff Pierce) that takes into account the square root dependence of ion concentrations on GCRs and the neutral nucleation in the CLOUD results, suggests that for average conditions the solar modulation of GCR would impact nucleation by about 1% – rising to perhaps 12% for the biggest changes in GCR seen in figure 2 at very cold temperatures. Thus the nucleation change as a result of real world GCR modulation is going to be much smaller than seen in these experiments, and much less important than the amount of pollutants.

In summary, this is a great example of doing science and making progress, even if it isn’t what they first thought they’d find.

220 Responses to “The CERN/CLOUD results are surprisingly interesting…”

  1. 51
    Jeff Id says:

    “Of course, to show that cosmic rays were actually responsible for some part of the recent warming, you would need to show that there was actually a decreasing trend in cosmic rays over recent decades – which is tricky, because there hasn’t been (see the figure). ”

    Isn’t it true that a ‘trend’ wouldn’t be required but simply an offset over recent decades? The offset creates the imbalance which results in the energy build up…

    It is also important to understand that an offset is most difficult to identify.

  2. 52
    tea partier says:

    #34 “…The difference between the Kristjánsson and Svensmark papers is likely the difference in the number of FDs used in their primary results…”

    If memory serves correct neither paper filters the data based on weather events coincident with FD events. The presence of waxing/waning major low pressure systems would obscure the GCR effect I would think. The fact that major FD events have a signal of any kind showing through cloud data indicates the GCR effect is likely significant.

    Filtering weaker FD events for only those coincident with high pressure weather systems over the east Pacific etc. might improve the chances of seeing a GCR signal from weaker FD events.

  3. 53
    Edward Greisch says:

    Please add FD “Forbush decrease, in astronomy, a decrease in observed cosmic ray intensity” to the acronym-index/

    45 Jeff Pierce: Thanks. By the feel of my lungs, I would say that there are plenty of somethings in the air around here.

    47 SecularAnimist: Content-free “bombshells” are the best kind when leading content-free minds. That is the beauty of that kind of “leadership.” No thinking is required. The followers never think to ask for content. Yet they keep on reading the same stuff.

  4. 54
    Ian Forrester says:

    I’ve always known that Scottish scientists punch way over their weight but for 63 foreign scientists to essentially come up with similar results to Charlie Wilson almost exactly 100 years after he invented the first cloud chamber (he observed real clouds and didn’t need to invent a crazy mnemonic) says much for the quality of science conducted in Scotland 100 years ago. I bet that Charlie operated on a shoe string budget unlike what scientists get today.

    Of course the big problem with the cosmic ray theory of global warming is that there is no long time trend for cosmic rays but that wont stop the deniers from spinning this report like crazy.

  5. 55
    Edward Greisch says:

    See the latest spin at:
    7 Artimus K
    It needs to be stated more clearly: “CO2 is Still the cause of GW.”

  6. 56
    Qwinn says:

    “Oddly, not one of those comments that I have seen bothers to explain exactly what the supposed “bombshell” is.”

    Here, let me explain it for you.

    We already know that the AGW cult hasn’t even produced one decent predictive model that A) when applied to historical data, actually predicts known future results, and B) makes even a half decent effort at properly accounting for *clouds*.

    [Response: So supporting a theory that works and produces validated predictions makes you part of a cult? Curious. – gavin]

    And now we’ve just added that all the models you adore have been completely wrong – by at least an order of magnitude – in accessing the impact of cosmic rays as well.

    [Response: You’ve just made this up. The modelling of the full aerosol growth and decay cycle is relatively new – with people like Jeff Pierce (see above), Peter Adams or Susanne Bauer leading the charge. It’s fascinating (and uncertain) stuff, but it is a small part of the overall picture. The overall effects of CR changes estimated via these models are indeed small, but that is pretty much in line with the very limited results shown in this paper. – gavin]

    This post introduces several strawmen. The big one is that the anti-AGW crowd is claiming the cosmic rays are the “dominant driver” of climate change. We don’t have to prove that. We just have to prove you don’t have a clue.

    [Response: Have you even read Svensmark’s book? Thought not. – gavin]

    [edit – political tedium removed]

    But when you guys assume, as always, that the burden of proof is on the -other- guy to continually prove a negative, and if we don’t, we have to give you trillions of dollars… why, I bet you feel like you win every single debate, don’t you?

    [Response: Oh please. No one is giving scientists ‘trillions’ of dollars. And if this is your answer for why this is a ‘bombshell’ paper, you completely fail to make your case. You appear to be arguing that anything that you perceive as anti-AGW (incorrectly, and in stark contradiction to what the authors themselves claim) must be a bombshell because, well,…. ‘trillions of dollars’. That isn’t an argument, it’s a knee-jerk. Try and be a little more reflexive, and even sceptical. – gavin]

  7. 57
    David Horton says:

    Ian Forrester #52 yes, and the other big problem is that cloud formation hasn’t prevented major climate swings in the past.

  8. 58
    Clearview says:

    I you look at the Oulu neutron counts during solar maximas from 1970 to 1990, there’s significant drop in the neutron count during hight solar activity. This could partly account for the warming from 1970’s to 1990’s.

  9. 59
    Septic Matthew says:

    11 gavin inline: Undeteriminable at this stage because we don’t actually know the impact on real clouds, or what the net radiative effect might be.

    Gavin’s presentation was good, and this sentence needs to be remembered.

  10. 60
    vukcevic says:

    Any hypothesis should at least be able to backtrack reasonably accurately what is known in the recent years, decades and even centuries, as I show here with the help of the N. Atlantic currents:

  11. 61
    Paul A says:

    An interesting comment found at the UK Daily Mail (not a place I normally go for climate science):

    “Commenting on the Nature paper, Prof Arnold Wolfendale, Professor of Physics at Durham University has pointed out several other holes in the cosmic ray idea, notably the fact that we would expect to see cigar-shaped clouds in the high atmosphere following cosmic ray tracks – we don’t.”

    All the focus has been on the actual CLOUD experiment, but surely there are other signatures like clound shape.

  12. 62
    Mac says:

    Congratulation for another VERY good post (and well-timed)!

    Just one quick question – isn’t all this cosmic-ray link (in the absence of any trend in recent times) mostly an argument for a higher actual sensitivity (but probably ‘fully kicking in’ more on the medium-term, after 2-3 solar cycles)?

  13. 63
    James Allan says:

    I’m familiar with a fair handful of the authors of this paper (a few on a personal level) and have a lot of respect for them as scientists, so the lack of spin is nothing more than I would expect. There is a lot of very good science out there, it’s just a shame that it is the spin-heavy stuff that more routinely gets the public attention. In terms of presentation, I think what the authors deserve a certain amount of credit for is the lack of gems available for the denialist quote-miners. It’s a fair assumption that the usual suspects are currently trying to mangle this article for all it’s worth.

    And for the record, one of the most noteworthy things for me about the experiment was how clean they managed to get the chamber. This, if nothing else, is what made this experiment truly unique compared to those conducted at other facilities. Although the instrumentation is pretty cool, I admit.

  14. 64
    Charlie Z says:

    I agree that this is not the slam dunk proof that the denialists claim it is. But it does lend support to Svensmark’s theories. At the very least, it doesn’t disprove them. In the text you provide additional steps that would be needed to show the correlation, so, you clearly are not dismissing the possibility that cosmic rays influence on cloud formation could be a much more significant contribution to global temperature change that previously thought.

    I think what this paper most significantly shows is that the science is definitely not settled.

    [Response: This idea far predates Svensmark – going back to Ney in the 1950s or Dickinson in the 1970s. What people have objected to in Svensmark’s work is not the idea that there are potential connections between GCR fluxes and climate, but rather the ridiculous overselling of their results, the inappropriate manipulation of data, and the lack of predictability of any of their proposed correlations when new data arrives. There are many issues in climate that are worth more study and this is certainly one of them, regardless of the previous overwrought hyperbole. – gavin]

  15. 65
    SecularAnimist says:

    Qwinn, whose litany of belligerent right-wing bumper sticker slogans inexplicably received a thoughtful, respectful response from Gavin rather than being consigned to the darkest depths of the Bore Hole where such Ditto-Head drivel belongs, wrote:

    “We just have to prove you don’t have a clue.”

    Which you have utterly failed to do by regurgitating nonsense and falsehoods.

  16. 66
    Martin Vermeer says:

    Jeff Id #49:

    Isn’t it true that a ‘trend’ wouldn’t be required but simply an offset over recent decades? The offset creates the imbalance which results in the energy build up…

    But the imbalance would be there immediately from 1953 on, see Figure 2 in the post. You would need to explain why temps didn’t immediately start rising then by at least the current rate…

    It is also important to understand that an offset is most difficult to identify.

    Jeff, “God of the Gaps” much? ;-)

  17. 67
    Scott says:

    while there isn’t a downward or upward trend in the amount of CR, is there a trend in the amount of aerosols over the past 100 years or so?

    [Response: Yes. Mostly from coal, oil, agriculture, biomass burning. – gavin]

    Gavin, Is this trend up or down? Haven’t the amount of aersols gone down in the past 60 years due to clean air act and similar measures?

    If so, *might* we be seeing a net positive contribution to ave. Global temperatures because less clouds are forming at this boundary layer due to this aerosol nucleation phenomena?

    Also, has anyone actually done a study on nighttime global trends vs. daytime global trends?

    thanks in advance!

    [Response: Over the last 100 years, definitely up globally. But regional emissions have varied enormously. For instance, US black carbon probably peaked in the 1930s (from Greenland ice cores), sulphate aerosols in the US, Japan and EU peaked around 1990 (down since), in China/India they have not yet reached the peak, biomass burning in the tropics is peaking now. – gavin]

  18. 68
    Scott says:

    Wow, the right wing picked up the spin on this paper in a hurry!

    The hypocrisy is really unbelievable. There are thousands of papers on this subject that are completely overlooked and dismissed as some sort of scientific cabal to fool the world into a marxist new world order, yet they latch onto any scientific paper that even hints at supporting their misguided and uninformed ideas.

    This cherry picking of science to support politically motivated propaganda is depressing.

  19. 69
    Adrian Smits says:

    The reality is there is now a real alternative to agw theory and its called gcr. If the Danes estimates turn out to have credibility which appears to be the case the models for agw have to be cut in half! At the very least.

    [Response: This is hilarious. You read something that you percieve to be against mainstream thinking (wrongly) and then without any work on attribution at all, you arbitrarily reduce the attribution to anthropogenic effects by 50% (at least!). The irony here is impressive. – gavin]

  20. 70

    62, Charlie Z,

    I think what this paper most significantly shows is that the science is definitely not settled.

    I’ve seen this particular moved goalpost arise a lot of late. It’s almost as if it’s the new denial stance. It’s not that AGW isn’t or couldn’t be happening, it’s all of that stuff that we gee whiz just don’t know yet, so maybe we’re being hasty.

  21. 71
    Snorbert Zangox says:

    The missing contaminant is nitrogen dioxide, which catalyzes the oxidation of sulfur dioxide to sulfuric acid in the presence of moisture. The reaction is rapid and yields are near 100%. The nitrogen dioxide is not consumed. The particle growth rates are limited only by the availability of moisture.

  22. 72
    SecularAnimist says:

    Scott wrote: “… they latch onto any scientific paper that even hints at supporting their misguided and uninformed ideas.”

    Actually the deniers will latch onto any scientific paper that doesn’t remotely support their misguided and uninformed ideas, and pretend that it does.

  23. 73
    Jeff Id says:


    I don’t agree with your interpretation. Warming can have multiple sources, including CR and CO2. My point was that the post claimed that we would need to identify a trend in the data. — This is incorrect by my understanding.

    We need to identify an offset from historic values to see a change in cloud forcing (it would be considered a forcing in this case right?). I think the post should be tweaked to represent this truth.

    [Response: Jeff, the point is that there is no evidence of a change, offset, trends, whatever you wish to have. Of course, it is always possible (in principle) that the current CR flux is significantly different than in the past. But even if that were true (and there is no evidence, of course), then you’d still have to contend with the orders-of-magnitude-too-small effects! –eric]

  24. 74
    Jeff Id says:

    “[Response: Jeff, the point is that there is no evidence of a change, offset, trends, whatever you wish to have. Of course, it is always possible (in principle) that the current CR flux is significantly different than in the past. But even if that were true (and there is no evidence, of course), then you’d still have to contend with the orders-of-magnitude-too-small effects! –eric]”


    Thank you for responding. Please understand that I haven’t claimed that this is a ‘big hole’ in any sort of model or AGW theory. There is no ‘gotcha’ in my thoughts here. I saw a comment by Gavin that we should look for trend in the data and this is flatly not correct. It is an easy fix at my blog, it should be equally so here. It should say offset in CR as temperature is an integration of forcing minus outbound flux. It is a word change and a shoulder shrug. That’s what I would do.

    You say that there is no evidence that CR is ‘significantly’ different (I’m starting to hate that word), but then you claim that there are orders of magnitude too small effects. I don’t know. Flat out, don’t have a clue. Someday I might. Without reservation, if you would tell us why you are certain that the effects are too small to make a difference, all I can do is listen. Last time you commented at tAV, I was hoping you would clarify those comments as well.

    Again, thanks for letting these difficult comments through and I am interested in any of your group’s replies. If we are going to waste our time blogging, why not learn.

    [Response: Jeff: An epoch to epoch change, when fit linearly, will give a trend. There’s nothing wrong with calling it a trend (even if that does invoke images of a monotonic increase). I get your point, but it a semantic one. You’re essentially saying “maybe if we had data from 50 years ago, we would find that there *is* a difference, and hence a trend, and maybe (if we take into account the response time of the system, plus unforced variability), we’d be able to show that there is actually a plausible correlative relationship between climate and CR. But you’d still be left with no evidence that the effects are big. I didn’t say I was certain they are small, just that there is no evidence otherwise. And of course there are lots of other things that we know do have big effects on cloud nucleation (like, as Gavin said, plain old pollution (SO2 emissions)) and on radiative forcing (like, for example, CO2).

    It seems to me you are trying to say that the ‘jury is still out’. That’s fine. But to take that analogy further, there isn’t even really much of a defendant for the jury to decide on! Meanwhile, the CO2 case has already been prosecuted and sentenced.–eric]

    P.S. Your comments aren’t difficult!]

  25. 75
    MikeCoombes says:

    Since clouds can either have a warming or cooling effect, to what extent can an increase in cosmic rays be expected to have a warming effect?

    [Response: Undetermined – you should ask the people who are so sure that it can only be cooling why they think that. – gavin]

  26. 76
    Snorbert Zangox says:

    My point was that when nitrogen dioxide, which is always present in the outdoor atmosphere is added to the mixture in the experimental chamber, the rate of conversion of sulfur dioxide to sulfuric acid will accelerate dramatically. The order of magnitude deficiency that Gavin referred to earlier will disappear.

    I do not know why your rendition of the historical frequency of cosmic rays differs so radically from the data of others.

    [Response: The CR data are direct plots of the pressure corrected data. If other people have something different, they are not looking at the primary source. (Oh, NOx is not the issue since they injected H2SO4 directly, not SO2). – gavin ]

  27. 77
    David Horton says:

    #73 “it is always possible (in principle) that the current CR flux is significantly different than in the past.” On the other hand it is CERTAIN that the rate of increase of CO2 in atmosphere in last 30 odd years, and the causative increase in industrialisation, is unprecedented. Lo and behold, so is global warming! What are the odds that whatever CR might or might not be doing they would be mimicking the effects of that CO2 rise at precisely this time in our history?

  28. 78
    Jeff Id says:

    Thanks again Eric. This is the part which I have a problem with:

    “Of course, to show that cosmic rays were actually responsible for some part of the recent warming, you would need to show that there was actually a decreasing trend in cosmic rays over recent decades – which is tricky, because there hasn’t been ”

    To be accurate, it should say:

    “Of course, to show that cosmic rays were actually responsible for some part of the recent warming, you would need to show that there was actually a difference in cosmic rays over recent decades – which is tricky, because the data is too short”

    Now I have introduced the ‘too short’ point here for the first time, but Figure 2 data is far too short to claim any difference or lack thereof inside the noise. But again, you don’t need to find a trend as explicitly stated above, you need to find an offset which would integrate into a temperature blade over time. It is not semantic as I understand it. Isn’t it accurate that physics dictates that the integral of a power (Watts) imbalance leads to temperature rise, so the statement as written misleads the casual reader into expecting the hockeystick blade from a short bit of not-yet-integrated, high variance forcing data?

    The real answer is that from Figure 2 as shown, we don’t know anything new and shouldn’t expect to know anything new about a change in cosmic ray’s which would affect climate in recent decades.

    From the certainty of the articles conclusion, I wondered if anyone here knew of a good quality long term data source for cosmic rays that should replace figure 2. If it exists, it would explain the conclusion quoted here. I don’t know of any because my background is limited and a better plot may be common knowledge for you guys. If it isn’t available,then it seems the quote here should be corrected to demonstrate the proper uncertainty to the public.

    PS, Bringing the CO2 argument seems moot at this point. It is quite possible that this effect doesn’t have to be a ‘warming’ one. It could explain why observed temperature trends are running somewhat below many of the models as well and you might have a valid, ‘worse than we thought’ on your hands.

    Again, if I’m wrong about any of the above, I’m listening.

    [Response: Jeff. Gavin’s point about ‘no trend’ however is partly in response to those that have claimed that the warming of just the last 30 years is GCR. But sure, ‘too short’ is fine, and perhaps more relevant. But the take home here is a correction to your P.S.: “…observed temperature trends are running somewhat below some models, well above others, and since we have absolutely no clue which direct the GCR effect ought to go, even if it is important (for which there is no evidence), it is not clear whether it would help explain anything whatsoever.” That would be a little more honest than Watts ridiculous “a non-visible light irradiance effect on Earth’s cloud seeds has been confirmed”, dontcha think?–eric]

  29. 79
    Macro Tel says:

    I wonder if a GRACE is possible for mapping water tables in the boreal forest, tundra, and maybe soon rock, North of 45^0-50^0 latitude? Also a clay satellite might be useful. Peat has a different density than does water, than does ice, than does the living Acrotelm layer. Extra points for differentiating non-fuscum vegetation. Clay might be useful to spread over microsite depressions where water may be scarce or slope high. IDK about mapping clay deposits but the terrain is isolated and sometimes marshy; easier with satellite. The clay satellite need only be temporary; IDK yet if clay’s capillary physics acts as a water reservoir for acrotelm like peat does.
    Water table to within a few cms is worth $$. Few tens of cms is okay. Sphagnum is best at -15cm water table except jr sphagnum at maybe -5cm. This would be best a permanent satellite.

  30. 80
    Brian Dodge says:

    “…which is tricky, because the data is too short” Jeff Id — 27 Aug 2011 @ 8:16 AM

    No. As Svensmark’s analysis showed, clouds respond within days to the transient changes from Forbush events. Further, the diurnal variation in temperature and its change between cloudy and clear skies show a rapid response to this forcing. Download 30 years of monthly data, CR from Oulu, and UAH from woodfortrees, and plot them using a spreadsheet. There is an obvious cyclic variation in CR data corresponding to the solar cycle, but little obvious correlation to temperatures. Sort the monthly data pairs by CR intensity, and plot that data. This puts the temperatures during low CR counts, ~’82, ’90, ’00-’02 on the left side of the graph, and temperatures during high CR counts, ~’87, ~’97, ~’09, to the right. This will tend to remove the trend in UAH temperature due to global warming, since early and late years get mixed together. There remains a weak POSITIVE correlation with Oulu CR and UAH Temperature, 0.274. I may be able to upload graphs to an image site later.

  31. 81
    seven says:

    The CERN press release is rather clear. There are 2 different results;
    1 High clouds do form in greater numbers thanks to cosmic rays.
    2 Low clouds are formed far more more by vapours ammonia than gcr.

    Jasper Kirkby. “We’ve found that cosmic rays significantly enhance the formation of aerosol particles in the mid troposphere and above. These aerosols can eventually grow into the seeds for clouds. However, we’ve found that the vapours previously thought to account for all aerosol formation in the lower atmosphere can only account for a small fraction of the observations – even with the enhancement of cosmic rays.”
    The CLOUD results show that a few kilometres up in the atmosphere sulphuric acid and water vapour can rapidly form clusters, and that cosmic rays enhance the formation rate by up to ten-fold or more. However, in the lowest layer of the atmosphere, within about a kilometre of Earth’s surface, the CLOUD results show that additional vapours such as ammonia are required.
    Conclusion 1. cr does influence high clouds ‘enhance the formation rate by up to ten-fold or more’ and 2. don’t make much diff. for low clouds and where trace vapours do far more than thought.

    [Response: Not quite. The connection to high clouds is tenuous since they didn’t do experiments that were appropriate to that altitude (press, concentrations etc.). Rather they took the experiments for the boundary layer and just made it colder in order to increase nucleation. Of course, if they had really concluded that GCR were a bigger issue for higher clouds, that would mean that they were likely to be a global warming effect. – gavin]

  32. 82
    Jeff Id says:

    Brian, thanks for the reply but Fig 2 data is quite obviously too short to determine any differential from prior time.

    Regarding air temps, they are controlled by ocean surface temp. Expecting the oceans temp to oscillate on decade timescales, misunderstands the total energy capacity of the oceans. I don’t think anyone would expect a cosmic ray forcing to drive the ocean temps up and down every solar cycle, but it is possible to add/subtract a little every year with a net positive/negative forcing that would add up over multiple decades.

  33. 83
    Kevin O'Neill says:

    >> … that would mean that they were likely to be a global warming
    >> effect. – gavin

    I’m confused by this statement. Could you please rewrite or elucidate.


  34. 84
    Hank Roberts says:

    > long term data source for cosmic rays
    Has more been done with these? (If mentioned before, my bad, I’ve barely looked for this subject at all)

    J. GEOPHYSICAL RESEARCH, V.106, NO.A10, PAGES 21,585-21,598, OCTOBER 1, 2001
    Solar cosmic ray events for the period 1561-1994
    1. Identification in polar ice, 1561-1950
    … 2. The Gleissberg periodicity

    “A total of 125 large fluence solar proton events identified from the nitrate
    deposition in ice core from Greenland for the period 1561-1950 are examined in an exploratory study of the geophysical information that will be available from such data in the future. These data have been augmented with ionospheric and satellite data for the period 1950-1994….”

  35. 85
    Tea Partier says:

    #33 “…The Laschamp excursion led to a dramatic increase in GCR around 40,000 years ago (seen clearly in the 10Be ice core data), but did not noticeably impact climate at all…”


    I’m not sure the Laschamp excursion is a relevent argument for why GCR can’t modulate climate. At the time the earth was in the depths of an ice age with lots of dust flying around, a depressed biosphere due to low CO2, and a suppressed hydrological cycle. We see from this latest CERN paper that ammonia may be a critical component for the leverage of GCR on cloud formation. What was the ammonia content of the troposphere 40K years ago?

  36. 86
    Jeff Id says:


    Thanks for the time again. It is important.

    Gavin’s reasons for conclusions about no trend are the single problem I discussed with this post. You guys are the pro’s so we look to you for answers to questions when our less than full time studies can’t answer them. In fact, RC was the blog I asked questions of well prior to tAV on other issues. I did not know, until your reply, that climate science didn’t have an answer to the direction of cosmic ray’s. I was educated enough to wonder about it because of course I do know of the solar proxy data and recognized the difference.

    I may write a post on this exchange if I find time, hopefully your group will consider my suggestions for changes above.

  37. 87
    Hank Roberts says:

    TP, nobody’s claimed “GCR can’t modulate climate.”

    The research is working to identify signals in the various proxies for paleo work, the CERN work is looking into what other variable factors may also be involved, and much else. Whatever’s there is a small effect from a lot of factors. Flip the planet’s magnetic field
    “… as this paper shows – we’re not even sure what all of the gases are,” said lead author Jasper Kirkby from CERN in Geneva, Switzerland. “We’ve found that [in addition to sulphuric acid and ammonia] there has to be another vapour (or vapours) involved, which has a controlling influence.”

    Quoted at
    Cosmic rays may influence cloud formation, 25 August 2011
    by Ceri Perkins

    The problem is whatever’s there isn’t big strong obvious and simple.

    It’s easy to speculate, but speculation proves nothing.
    Is the missing multiplier/mystery vapour CO2? nitrogen from agriculture? DMS? Sulfates? hydrocarbons? One of the many thousands of other artificial chemicals maybe? Is a geomagnetic excursion or field flip, e.g. Dergachev et al., needed to make the effect big enough to be detectable in the proxies?

    Whatever it is it’s small compared to the big control knob we know about.

  38. 88

    Gavin, you wrote, “Of course, to show that cosmic rays were actually responsible for some part of the recent warming, you would need to show that there was actually a decreasing trend in cosmic rays over recent decades – which is tricky, because there hasn’t been (see the figure).” There is a trend, and it’s up. The trend of the pressure-corrected Oulu neutron count since April 1964 is y = 0.3935x + 6026.6 (resolution of one month to reduce the size of the spreadsheet and apply some smoothing. (The data are provided by Sodankyla Geophysical Observatory and are available in many resolutions at I generally do not apply a linear regression to so few cycles but thought this might be of interest. (I do apply linear regressions to my atmospheric data, but these observations were begun in 1990 and constitute nearly 22 annual cycles.)

    [Response: Uh.. What are the units of x and y, and what is the confidence interval on that trend. I’ll bet is it much greater than the slope of the best fit line. if so, then, no trend…–eric]

  39. 89
    Edward Greisch says:

    The On-Line Database of Cosmic Ray Intensities at
    looks like it ought to have a lot to say about the issue. Is that what RC people have already looked at? If that data is flat, issue answered.

  40. 90
    Hank Roberts says:

    How about other sites? This paper mentions many (no specific numbers given for most of the other sites, but it seems to say the trend is generally down)
    Long-term decline of South Pole neutron rates
    doi:10.1029/2006JA011894, 2007

    It begins
    “… As illustrated in Figure 1, most neutron monitors
    return to approximately the same count rate from one solar
    minimum (cosmic ray maximum) to the next. However, the
    South Pole monitor displays a remarkably different behavior
    consistent with a steady long-term downtrend. Even the
    monitor at Mt. Washington (not shown), which is perhaps
    most comparable to South Pole in terms of cutoff and
    altitude, displays no such marked downtrend [Lockwood
    et al., 2001].”

    “… In summary, we have not been able to identify any
    instrumental or environmental effect that could cause the
    long-term decrease in the South Pole neutron rate. Unless
    some such cause emerges in the future, it would appear the
    origin of the decrease must be a change in the Sun or solar
    wind, with an attendant change in the strength of solar
    modulation of cosmic rays [Ahluwalia and Lopate, 2001;
    Caballero-Lopez et al., 2004; McCracken et al., 2004a,
    2004b], or possibly a change in the local interstellar density
    of Galactic cosmic rays [Stozhkov et al., 2000].”

  41. 91
    Hank Roberts says:

    Interesting discussion of how to assess trends here:

    Ann. Geophys., 29, 251–262, 2011
    Secular trends in storm-level geomagnetic activity

    Lots of discussion of statistics in this one, for those who like that sort of thing.
    Brief excerpt follows

    “… while many people might claim to recognize a trend when they see one, a more precise, but still usefully general, definition is difficult to pronounce (e.g. Preece, 1987). We are reminded of the limerick by Cairncross (1969): “A trend is a trend is a trend …”. To which we would unpoetically add the hope that a graph of the data would have a visually-compelling slope.

    Of course, needed specificity for what is meant by “trend” can be obtained through (2) measuring and testing. These typically involve either deterministic or stochastic analysis, with limitations imposed by data quantity and quality, and the possible presence of superimposed signals. Here, the notion of significance is important, as is the timescale over which the trend is supposed to apply. And there are practical considerations ….

    … While fits to data across all 13 solar cycles seem to show a linear trend of increasing geomagnetic disturbance, fits to shorter durations, such as for 6, 4, 3, or 2 cycles, do not consistently show persistence. Indeed, Figs. 1 and 2 show that the time-dependence of past geomagnetic activity has been complicated.

    Clearly, our observation of a long-term linear trend of increased geomagnetic disturbance is due, in part, to the time span we have considered, the span of the available geomagnetic K time series, 13 solar cycles.

    This is a simple, but important, observation that has been made by others (e.g. Richardson et al., 2002, Fig. 1; Mursula et al., 2004, Fig. 3).

    If we had chosen to analyze the time span of (say) the past 6 solar cycles, we would, instead, be discussing a decreasing trend in geomagnetic disturbance!

    (extra ‘paragraph’ breaks added for online readability — hr)

  42. 92
    rykart says:

    This is the quote I see ALL over the web on denialist sites:

    “When Dr Kirkby first described the theory in 1998, he suggested cosmic rays “will probably be able to account for somewhere between a half and the whole of the increase in the Earth’s temperature that we have seen in the last century.”

    This seems exceptionally sweeping and wild, and I have a funny feeling Kirby never said it, as the denialist vermin have an unpleasant habit of simply inventing stuff.

    Anyone know if this widely circulated quote is indeed accurate?

    [Response: It comes from a 1998 physics world article – and I agree with you, it seems exceptionally sweeping and wild and is not the kind of thing you’d expect scientists to say prior to having any actual evidence. – gavin]

  43. 93
    prokaryotes says:

    rykart, this comes up when i put “kikby” into the search box over at

    ConCERN Trolling on Cosmic Rays, Clouds, and Climate Change

    ANd here i read, quote:

    RealClimate has a good rundown of what Kirkby et al.’s results do and do not mean. The short version is that Kirkby et al. do find increased aerosol nucleation under increased ionization (i.e. “more cosmic rays”), particularly in the mid-troposphere, but the effect is smaller at warmer, lower levels where the cosmic ray-climate myth proponents claim it has its greatest climatic effect. Lead author Jasper Kirkby has tried to set the record straight, stating (all following emphases mine):

    [The paper] actually says nothing about a possible cosmic-ray effect on clouds and climate, but it’s a very important first step.

    Kirkby himself is debunking the claims you mentioned above. That is beyond desperate…

  44. 94
    seven says:

    This 2008 article mentions it in a in dept story about the start of Cloud and decade delay after the quote.

    ps Here are 2 vids where CERN / Kirkby explains the findings that nature now published.
    CERN News (short video):

    Video Release:
    (After 9 minutes a interesting explanation and animation about cloud nuclei formation)

  45. 95
    Pete Dunkelberg says:

    Unsigned CERN pdf

    What has CLOUD discovered and why is it important for our understanding of climate? There are several
    important discoveries from CLOUD. Firstly, we have shown that the most likely nucleating vapours, sulphuric
    acid and ammonia, cannot account for nucleation that is observed in the lower atmosphere. The nucleation
    observed in the chamber occurs at only one‐tenth to one‐thousandth of the rate observed in the lower
    atmosphere. Based on the first results from CLOUD, it is clear that the treatment of aerosol formation in
    climate models will need to be substantially revised, since all models assume that nucleation is caused by
    these vapours and water alone.


    [Response: “Climate models need to be substantially revised” is a cliche that seems to be appended to any experimental result that is connected to climate, particularly when no climate models or their results have been discussed in the paper at all. The context that is missing is that aerosol nucleation processes don’t actually figure in 99% of climate model results, and so it is unlikely that any of those need to be revised in the slightest. People need to remember that climate modelling is a hierarchy going from the relatively simple to the extremely complex, but not every element in the very complex models impacts the top-level results in the simpler formulations. Take aerosol nucleation – this figures only in models that attempt to capture the whole aerosol growth and decay process (such as the model in Pierce and Adams (2008), or the MATRIX module. It does not figure in the AR4 models, and even though the CMIP5 models have substantially more aerosol/chemistry than CMIP3, the nucleation process isn’t included in them either. Instead, these models base their aerosol physics on empirical large scale effects – not explicit microphysics. Obviously one can learn a lot from doing more ‘a priori’ calculations (which is why MATRIX/Pierce and Adams etc. are happening), but it only indirectly affects the bulk aerosol modules that are used in most the current GCMs. And even then, I’m not sure enough about the details of the aerosol microphysics used in the a priori calculations to know whether they need to be ‘substantially revised’ as a function of this paper. Perhaps Jeff Pierce could comment? – gavin]

  46. 96
    Tilo Reber says:

    Gavin: “1.… that increased nucleation gives rise to increased numbers of (much larger) cloud condensation nuclei (CCN)”

    Doesn’t common sense tell you this? You don’t go from single atoms or small molecules to CCNs in a single step. These things must grow gradually from very small to very large. And the more of the small components that are available to grow into CCNs, the more CCNs there will be. Of course the Forbush decreases that Svensmark et al found confirm that this is what happens.

    I’m not sure how Svensmark’s name being on the paper is relevant, especially since Svensmark’s own SKY experiments anticipated these results.

  47. 97


    It seems to me that you’re arguing that a stepwise change in GCR before 1950 could have caused (part of the observed) warming after 1975. That seems entirely implausible though:
    a) I would not expect the temperature to suddenly increase (after a period of little change) long after the forcing has stopped changing (if that forcing is dominant).
    b) Timescales for the GCR-aerosols-climate link are pretty short, so even more reason to expect a quick response in “climate” when GCR conditions are changing.

    Btw, I agree with James Allan that the authorlist of this article has many very good researchers on it who have not staked out a contrarian position on climate. Note that Svensmark is absent, which could perhaps explain the missing spin.

  48. 98
    Tilo Reber says:

    Bart Verheggen: “It seems to me that you’re arguing that a stepwise change in GCR before 1950 could have caused (part of the observed) warming after 1975. That seems entirely implausible though:”

    The strength of solar cycles was climbing strongly up through the late 1950s. But there is no reason to think that climate had reached equilibrium at the same time that we reached the peak solar cycle. After solar cycle 19, 20 to 23 were still strong cycles, so temperature could still have been climbing to equilibrium with regards to solar cycles. On a shorter time scale, the temperature doesn’t have to follow the cycles exactly. The solar GCR effect doesn’t act in isolation. Any GCR signal would still be modulated by ENSO, volcanoes, and a CO2 climate sensitivity that could be around 1C.

    [Response: So now we are making progress. You accept that a) all factors need to be weighed in any attribution study, b) that there is thermal inertia and c) that there is internal variability unconnected to any forcing. So far, so good. But how should such an attribution proceed? Perhaps we could look at the impacts of all the individual processes and their likely time-evolution? Then perhaps we could try and see what would happen if we did them all together? And perhaps we would be careful to allow for some multiplicative factor to account for some uncertainty in the magnitude of the effects? And have multiple estimates of the role of internal variability in order to account for that? Still with me?

    Oh yes. – gavin]

  49. 99
    Jeff Pierce says:

    #95, Pete Dunkelberg: The “climate models need to be substantially revised” comment. I don’t think this comment has anything to do with their cosmic ray results (and I’ve found it interesting that the media has used it as a statement about the cosmic rays), and I’ll explain why in a moment…

    But first regarding the IPCC climate models, I think Gavin explains it well. Right now the IPCC climate models don’t take into account any aspect of aerosol number or size (just the total mass). Thus, they do not take nucleation into account. If changes in nucleation with time end up being important for climate change, we will need to incorporate these physics into the IPCC models. The comment really is directed at people like me (and about 10-20 other groups around the world) who use detailed global (or regional) aerosol microphysics models.

    However, the comment is really stating that the CLOUD experiments could not reproduce atmospheric nucleation rates from sulfuric acid + ammonia + cosmic rays + water vapor, “The nucleation observed in the chamber occurs at only one‐tenth to one‐thousandth of the rate observed in the lower atmosphere.”. Therefore, some other component (e.g. organic vapors) needs to be included in order for the CLOUD experiments match atmospheric nucleation in their chamber. These are going to be the results from the next set of CLOUD experiments (which occurred earlier this summer), and they are foreshadowing these results. They do not actually bring up cosmic rays in that paragraph at all, but they make it clear in the paper that even with cosmic rays cranked up in the chamber that they still cannot reproduce atmospheric nucleation rates.

    Since detailed aerosol models (like the one I use) do not include organics (or these other missing species) in the nucleation calculations, THESE missing components are what needs to be revised in our models, not the dependence of cosmic rays. Recent modeling by my group and those out of Fangqun Yu’s group at SUNY Albany does use a nucleation scheme that has a dependence on cosmic rays that is very similar to the CLOUD results. Thus, our models don’t need to be revised based on the cosmic ray results, but they do for the additional missing components (e.g. organics).

    I hope this helps.

  50. 100
    Tilo Reber says:

    Gavin: “Perhaps we could look at the impacts of all the individual processes and their likely time-evolution?”

    Certainly. For example, while the rise of solar cycle strength level off in the late 50s, we switched to an El Nino dominated mode of ENSO around 1977 that could well have continued to drive temperature up even if GCRs were no longer increasing.

    So, with the possibility of the GCR effect not having reached equilibrium and ENSO kicking in to help and CO2 kicking in to help, we could easily have seen what we see in the record. Certainly a CO2 only attribution would be an error.

    Gavin: “And have multiple estimates of the role of internal variability in order to account for that?”

    I’m not sure that multiple estimates increase accuracy. Multiple estimates may be better than the worst of the estimates, but I doubt they are better than the best of the estimates.

    Gavin: “Oh yes”

    Did you find GCRs included in the attribution considerations?

    [Response: The impact on temperatures of an increase in solar activity to 1950 and then flat, would have been an increase to a little past 1950 and a roughly exponentially slowing warming since. The timescale for the exponential is related to the sensitivity of the climate (i.e. the longer the posited delay the bigger the sensitivity – see Hansen et al, 1984 and Lindzen if you want a different perspective), and so you quickly find that large solar forcings combined with everything else just don’t fit. GCR are proportional to TSI and so to first order a mutliplicative factor on TSI forcing would capture the main effects on global mean temperature – that is already taken into account in D&A studies. – gavin]