Good line Gavin: Some contrarian commentators have recently fallen into the habit of mass mailing… there certainly is an astroturf campaign going on, and your guys’ bandwidth certainly has taken a hit because of it.
But I’d like to thank you for the solar forcing synopsis. It is rare that one paper can be pointed to as proof of anything – most papers are bricks in the wall, albeit some bricks are more important than others.
I taught stats for soc sciences a couple of times, and a big issue was CORRELATION DOES NOT MEAN CAUSATION (which got the tobacco industry off the hook for may decades until proof at the cellular-molecular level came in). In class I used the example of the correlation between storks & birth rate (turns out both are increased by rural locality).
So what is needed in addition to a simple correlation is a GOOD, LOGICAL THEORY (plus some common sense) and a load of other possible causal variables being controlled for. It seems we have tons of these other supports for AGW, but very little for these 2ndary solar factors.
The other main point (raised by this article) is just because the sun impacts climate, this does not disprove the impact of GHGs.
Finally, if (suppose) the 2ndary solar factor contrarians are right, and these 2ndary solar factors (in addition to GHGs) cause warming, then that’s all the more reason to take charge over what we can control, and reduce our GHG emissions, or we may be asking for even worse problems than what the climate scientists are predicting! And then what if a bunch of volcanos decided to erupt? We really need to get to it and do what we can, and reduce GHG emissions pronto!
I was surprised at this statement: “Sometimes even papers in highly respected journals fall into the same trap.”
The ‘high-impact’ journals have much higher rejection rates than other journals. This is worrying because the basis of this rejection is often that the new study doesn’t significantly change our view of something. Therefore, you might think that strange results which go against common understandings would be more likely to get accepted. Because 5% (depending on the specific field’s convention for Type I statisitcal errors) of rejections of null hypotheses will generally be unwarranted across all studies, the journals with the highest rejection rates are likely to be publishing spurious rejections of null hypotheses more frequently than journals that aren’t above publishing results that confirm the work of others. That is, in statistical terms, Science and Nature might publish Type I errors much more frequently than the stated alpha would suggest. Given that studies demonstrating a failure to reject a null hypothesis are much more difficult to publish and given the extremely high rejection rates in the ‘high impact’ journals, it is possible that large proportions (much higher than 5%) of conclusions published in them are wrong.
To me this shows the strength of the AGW science. Different methods applied to a variety of data are suggesting the same thing. Individual studies debunking AGW don’t seem to be sustainable on their own for any length of time, and (as far as I know) those studies contradict each other or fail to reinforce one another. I wonder if any commentators on science (David Woljick) would agree with this assessment.
I just noticed that my post at 2:28 pm was unclear. When I wrote “rejection rates,” I meant rejection of papers for publication. When I wrote “rejection of null hypotheses” I was referring to how frequently a statistically significant result was obtained. Sorry for any confusion.
Can you explain the correlation that should exist between NAO and solar cycle?
There are some forecasts concerning a cold winter 2005-2006 in western Europe .
These forecasts for example Met Office (UK) come from a link between SST and winter NAO, then between winter NAO and temperature negative anomaly.
Some people see a link between solar cycle and NAO.
They say that since we have a minimum solar activity between 2005-2007 we get negative NAO and temperature anomaly.
I tried to see any correlation between the solar cycle and the NAO indice but the two curves seemed completely independant.
Can you give some informations about this?
[Response: The mechanistic theories for why there should be link involve either stratospheric influences on the NAO (similar to the process that leads to a stronger winter NAO after big volcanoes for instance), or changes in tropical SST, or both. A tropical SST link would explain why the signal is strongest with a 10 to 20 year lag of the long-term changes (Waple et al, 2001), but the noise in the NAO record could mean that you only see significant changes after long term averaging. There is still a lot of work being done on this, but I wouldn’t anticipate much of a significant change in NAO over 2005-2007 since there is a lot of noise and a lot of other influences – not least greenhouse gases, which with either mechanism, provide an opposing tendency. -gavin]
I think there is a pretty complex dynamic governing that relationship. Still, on balance I agree they will be MUCH greater than 5% and the best journals will probably have more of them than second tier journals.
I differ with #2 & #7. “Good theory” is still an important requirement of science, so that should moderate the publishing of wacko false positives. However, I did notice when doing research in the early and mid-90s that The New Scientist tended to publish more sensational articles than Science or Nature. Since AGW did not reach scientific certainty in studies until 1995, that meant The New Scientist was publishing more articles (more assuming AGW validity) than the somewhat more scientifically cautious journals. And I think that’s good to have a least one journal that does that, even if they might possibly publish more contrarian research that doesn’t pan out later.
The other point is that AGW is obviously not your normal science topic, so publishing research that reconfirms AGW is in itself sensational & would keep the readership interested & subscribing. I know that the mainstream media is afraid to touch even bonafide, mundane studies that reconfirm AGW, because of their sponsors or fear of eliciting a slew of negative responses from contrarians.
I’m thankful that most CC scientists are continuing to do their research and publish honest reports in the face of threats & wearisome bogus contradictions. When they entered science, they probably didn’t realize they’d be called on to be troopers as well.
Comment by Lynn Vincentnathan — 16 Jul 2005 @ 11:03 AM
Hi, Gavin. In a different discussion I asked you to explain the fall in temperature between 1940 to 1970. Your response was:
[Response: The IPCC itself doesn’t have any models, however the results from the this GISS model have been submitted to the archive run by IPCC so that they can be considered for the next assessment report (AR4).
With respect to your question, the slight cooling trend seen in the results is a function of a combination of increasing aerosols, significant volcanic activity especially in the early 1960s and a plateau in both the greenhouse gas forcing and solar. Together that produces the ‘blip’ that you can see. Once internal variability is also taken into account the trend goes from near zero to negative over the different realisations. -gavin]
I still don’t understand the plateau in greenhouse gas forcing, but you also seem to be suggesting changes in solar activity as a significant driver of climate. Do you have figures (with uncertainties) about how much changes in solar forcing affect climate? I would really like to know how relatively important this is.
[Response: I’ve made it clear that I think that the solar variations are indeed a factor in driving climate change, though my opinion is that it is a relatively small factor over the last century. It’s difficult to assign a specific number because of the uncertainties in solar and other forcings particularly early in the century and in even working out the correct calculation to do – i.e. since there are multiple positive and negative forcings it will depend on how you group them. If you lump them as ‘natural’ forcings vs. ‘anthropogenic’ forcings, you end up with something like 20% natural from 1900 to 1940 (note that I haven’t done this calc exactly), and even less since then. There is the potential for some wiggle room there though. You can do the calculations yourself based on the forcing fields available at http://www.giss.nasa.gov/data/simodel/F.indiv.data.txt. -gavin]
It is my understanding, that a consensus has arrived on the influence of the sun on the climate. This originates from an increased knowledge of the variation in the sun. Please refer to the following link, from where I have copied the most important results:
” How Strongly Does the Sun Influence the Global Climate?
Studies at the Max Planck Institute for Solar System Research reveal: solar activity affects the climate but plays only a minor role in the current global warming
Two scientists from the MPI for Solar System Research have calculated for the last 150 years the Sunâ??s main parameters affecting climate, using current measurements and the newest models: the total radiation, the ultraviolet output, and the Sunâ??s magnetic field (which modulates the cosmic ray intensity). They come to the conclusion that the variations on the Sun run parallel to climate changes for most of that time, indicating that the Sun has indeed influenced the climate in the past. Just how large this influence is, is subject to further investigation. However, it is also clear that since about 1980, while the total solar radiation, its ultraviolet component, and the cosmic ray intensity all exhibit the 11-year solar periodicity, there has otherwise been no significant increase in their values. In contrast, the Earth has warmed up considerably within this time period. This means that the Sun is not the cause of the present global warming.
These findings bring the question as to what is the connection between variations in solar activity and the terrestrial climate into the focal point of current research. The influence of the Sun on the Earth is seen increasingly as one cause of the observed global warming since 1900, along with the emission of the greenhouse gas, carbon dioxide, from the combustion of coal, gas, and oil. “Just how large this role is, must still be investigated, since, according to our latest knowledge on the variations of the solar magnetic field, the significant increase in the Earthâ??s temperature since 1980 is indeed to be ascribed to the greenhouse effect caused by carbon dioxide,” says Prof. Sami K. Solanki, solar physicist and director at the Max Planck Institute for Solar System Research.”
As I read this paper, the debate on the impact of the sun on the climate is closing. Is this correct ?
I will be pleased if some will be able comment on this.
Comment by Klaus Flemloese, Denmark — 16 Jul 2005 @ 1:55 PM
Astute readers will notice that there is a clear problem here. The widespread predisposition to believe that there must be a significant link and a lack of precise knowledge of past changes are two ingredients that can prove, err…., scientifically troublesome. Unfortunately they lead to a tendency to keep looking for the correlation until one finds one. When that occurs (as it will if you look hard enough even in random data) it gets published as one more proof of the significant impact that solar change has on climate. Never do the authors describe how many records and how many different smoothing methods they went through before they found this one case where the significance is greater than 95%. Of course, if they went through more than 20, the chances of randomly stumbling onto this level of significance is quite high.
The proof that this often happens is shown by the number of these published correlations that fall apart once another few years of data are added, cosmic rays (which are modulated by solar activity) and cloudiness for instance.
Sometimes even papers in highly respected journals fall into the same trap.
A superb post which very persuasively throws into doubt claims that solar variation can explain recent temperature changes. I applaud your courage, since this is the very same logic that has thrown so much doubt on Mann’s results and very neatly summarizes why Mann’s work has drawn so much fire. You have gained a great deal of credibility (in my eyes at least).
[Response: I’m glad you’re happy. I don’t see the connection with Mann et al though – I am talking here about spurious correlations being trumpeted as proof of an effect. The work of reconstructing past climate changes, though it uses correlations between multiple proxies and climate, is subject to validation and in and of itself is not proof of any effect (be it solar or greenhouse gases or whatever). – gavin]
I also saw the connection that Terry suggested. I do not necessarily see it as some sort of slam-dunk refutation of Mann or whatever (so hold any defensive reflexes just for now). I will try and explain it as I see it.
Mann showed that there was a correlation between proxies and ‘global’ temperature over the past 100 or so years. Further investigation has suggested that this is primarily a correlation between some Bristlecone pine ring widths (or densities) and ‘global’ temperature – they are the dominant element of Mann’s PC1 and in their absence there is very little correlation (MM and subsequently Wahl and Amman have shown this effect). The validation you refer to is an out-of-sample forecast test that does not overcome the problem that correlation does not prove causation. Furthermore, the Mann studies and those that followed are essentially bivariate regressions which map between temperature and the distilled essence of the proxies (the point of the PC analysis). Importantly, there is no multivariate investigation in Mann – e.g. a regression of the PC of the proxies on temperature and rainfall and CO2 concentrations – to name a few. Thus the concern about spurious correlation still exists – there could well be omitted variables bias. Furthermore, temperature is a strongly trending series over the instrumental period. This leads to concerns that there may be problems caused by non-stationarity of the series (which detrending or similar processes does not cure). This reinforces concerns that it may be a spurious correlation (using that term technically to refer to I(1) series).
The problem then becomes that if the correlation shown over the past 100 years or so is spurious then any reconstruction is flawed. Doesn’t necessarily mean its so, but the exact same concern you discuss with respect to solar input does definitely affect Mann et al et al.
[Response: If Mann et al was simply based on a correlation, you may have had a point. However, there are clear validation steps in the MBH methodology – i.e. the proxies are matched over the period 1900 to 1980, and then validated against the the earlier period 1850 to 1900 to see if there is any predictablity – this is the step missing from most solar correlations. Only the proxy networks that valdiate in this way were used in the reconstruction. Note also that the issue with the bristlecone pines which was first brought up by Mann et al 1999 only affects the reconstruction between 1400 and 1490 because of the sparsity of data for that period. -gavin]
I will offer up my apologies in advance because this is probably the wrong thread to ask this, however…
In the paper Global Surface Temperatures over the Past Two Millennia, by Michael E. Mann and Philip D. Jones (Geophysical Research Letters Vol. 30, No. 15, 1820, August 2003), how are the uncertainties for the Northern and Southern Hemispheres (Fig. 2a, b) calculated?
The uncertainties for the NH and SH appear to be the same and they appear to remain constant over the entire 2000 years of the reconstructions.
This seems to be counter intuitive because:
1. The SH has far fewer records which suggests a greater uncertainty as suggested in the abstract; and
2. I would have thought the uncertainty would increase with time (in the past) from the present.
Any help understanding this would be greatly appreciated.
[Response: This paper tried to use a smaller number of long term records to make a uniform reconstruction. MBH on the other hand used different networks for different periods, hence the error bars changed as the networks get smaller further back in time. In Jones and Mann, the errors are constant because the network is constant. -gavin]
You may have noticed that I am a fan of enhanced solar influence on climate than currently included in climate models. Let me explain this again…
About past temperatures and sensitivity for solar forcing
First, there are more indications that the MWP and the LIA show larger temperature variations than what is found by different multi-proxy reconstructions (Mann, Crowley, even Moberg). Also beyond the North Atlantic. See the stalagmite data from China with variations of +4/-3 ºC compared to current annual temperature. Also coral data of the Great Barrier Reef where the seawater temperature varied +0.5/-1 ºC, compared to today in the past centuries. And not to be forgotten borehole data, which show a full 1 ºC change globally in the past five centuries.
It seems to me that most multi-proxies (except Moberg) rely (too) much on tree ring data, for one can question the reliability as temperature indicators.
If there was more climate variation in the past, then models should be adjusted for larger sensitivity towards natural variability, especially solar.
About models and solar forcing
GCM’s give some rather good simulation of past temperatures. But there are problems with volcanic influences. As the Shindell paper shows: the influence of volcanic is within the modelled unforced variability, except for Europe, where it is outside, but with the wrong sign (!), compared to proxies, thus probably overestimated in the model in question.
An investigation by Stott ea. to check the attribution of different forcings in the Hadcm3 model did find an underestimate of solar (factor 2), within the constraints of the model (like fixed influence of aerosols!):
“It is found that current climate models underestimate the observed climate response to solar forcing over the twentieth century as a whole, indicating that the climate system has a greater sensitivity to solar forcing than do models.”
The largest problem in current climate models is the feedback from cloud cover. That is responsible for most of the broad (1.5-4.5 ºC) range in sensitivity for a CO2 doubling between different models. But there is a direct connection between cloud cover and solar irradiation, which still holds until now (last data 1999). During a sun cycle, the global cloud cover changes with +/- 2%, good for a change of several W/m2 (depending on type of clouds and region), far higher than the effect of insolation change as result of the sun’s energy variation. The (negative) correlation between (low) cloud cover and solar irradiation is high and significant, see KristjÃ¡nsson ea.
As far as I know, not one of the current climate models includes the cloud cover response to the solar cycle, neither to longer time changes in solar activity.
About solar variability and climate indications
Solar variability is found in tree rings of Northern Europe (see page 3 of the 5 MB record), while there is no visible influence of increased GHGs. The summer temperature indication by tree rings in Finnish Lapland is decreasing in the period 1950-2001, see page 14 of Pages News. But this also may be an indication that the use of tree rings as (past) temperature indicators is not free of problems.
About 14C/10Be reconstructions and climate
From the Mangini ea. paper:
“The highly significant correlation with delta 14C underlines the important role of solar forcing as a driver of Northern Hemisphere climate during the past 2 millennia”, while you wrote that the explained variance with the 14C record is only 5%? No matter that, solar activity in the past 60 years has never been as high in the past 8,000 years. It’s level since the 1940’s is constantly higher than in any period before in the past millennia and the sun’s magnetic field more than doubled in the past century, including a 40% increase in the 1960-1999 period. While there is a direct effect on climate, there may be a residual effect from this constant higher energy level of the sun (+ cloud feedback), even if several of the indications show a current levelling of solar activity. The oceans need more time to heat up until a new equilibrium is reached.
The science about the different sensitivities (solar, volcanic, GHGs, aerosols and their feedbacks) is far from settled. What is clear is that solar influences are underestimated in all current climate models.
A persistent argument used by proponents of solar climate forcing is supported by the
physics of CO2 infrared opacity (band saturation) at wavelengths of the electromagnetic
spectrum where CO2 atmospheric heating occurs.
David Archer has written in “Global Warming: Understanding the Forecast, 2005″
geosci.uchicago.edu/~archer/PS134/ Chapter 3 that wavelengths, wings on either
side of the 600 to 800 cycles/cm absorption band are modeled to be sufficient to
continue trapping from 2ºC to 5ºC of heat.
A counterpoint to this argument is exemplified by Gerald Marsh in his “Global
Warming Primer” (www.nationalcenter.org/NPA420.pdf) where he claims…”additional
carbon dioxide does have an influence at the edges of the 14.99 micron band. Because
of this marginal effect, the change in forcing due to a change in carbon dioxide
concentration is proportional to the natural logarithm of the fractional change in
concentration of this gas.”
“Specifically, the IPCC gives (change in forcing) dF = 6.3 ln (C/C0) W/m2 where dF
is the change in forcing, and C0 and C are the initial and final carbon dioxide
concentrations. […] The Earth’s temperature is therefore relatively insensitive to
changes in carbon dioxide concentrations, a doubling leading to a dF of only
“If the edges of the absorption bands were completely abrupt, as if CO2 absorbed 600
cycles/cm light completely and 599 cycles/cm light not at all, then once an absorption
band from a gas was saturated, that would it. Further increases in the concentration of the
gas would have no impact on the radiation energy budget for the earth. CO2, the most
saturated of the greenhouse gases, would stop changing climate after it exceeded some
concentration. It turns out that this is not how it works. Even though the core of the CO2
band is saturated, the edges of the band are not saturated. When we increase the CO2
concentration, the bite that CO2 takes out of the spectrum doesn’t get deeper, but it gets a
“The bottom line is that the energy intensity Iout in units of W/m2 goes up
proportionally to the log of the CO2 concentration, rather than proportionally to the CO2
concentration itself (we say linear in CO2 concentration). The logarithmic dependence
means that you get the same Iout change in W/m2 from any doubling of the CO2
concentration. The radiative effect of going from 10 to 20 µatm pCO2 is the same as
going from 100 to 200 µatm, or 1000 to 2000 µatm. ”
They both seem to be saying the same thing. For Marsh “the Earth’s temperature is
therefore relatively insensitive to changes in carbon dioxide concentrations.” For
Archer the models predict a rise in temperature of 2ºC to 5ºC.
Which is it?
[Response: They are both saying the same thing. The only possible confusion would be in what Marsh is comparing the CO2 sensitivity with. From the context, it’s clear that he is comparing it to a gas with less (or no) saturation, and therefore the Earth is ‘relatively’ insensitive to CO2. i.e. for the other gas the absorption would go linearly or something with concentration. It is despite this ‘relative’ insensitivity that CO2 still gives a forcing of around 4W/m2 if it doubles. -gavin]
In your graph, demonstrating the delusions of Mangini et al., you have used a “slightly cleaner” solar proxy, without mentioning that this might be somewhat (even radically) different to the one used in the paper.
Are you saying that the atmospheric C14 series is not a reasonable solar proxy?
Intcal98 etc seems to have be used by quite a few as a solar proxy. (eg Wang et al. Science 2005)
( Mueschler et al, 2004 -Muscheler?? Have you a link?)
Many thanks for the links.
Ive done a similar quickie using the INTCAL98 series for the period 5AD to 1935AD (I interpolated the SPA data to get to the Intcal dates with a cubic spline).
Raw points: r2 = 0.23
3point Moving average: r2 =0.34
3point MA but with a quadratic: r2= 0.41
(All Excel numbers)
I obtained the Muscheler paper from ScienceDirect but it didnt have the C14-production data-set with it, do you have a link to it?
As you say the more interesting question is how current climate models match such datasets; Wang et al’s Asian monsoon stalagmite would seem a good test; do the models demonstrate any solar-asian monsoon linkage?
> The New Scientist tended to publish more sensational
> articles than Science or Nature.
I emailed them about that once and got back a very nice letter pointing out to me that editorially, New Scientist is and means to be an _entertainment_ publication. Their aim is to sell magazines and keep people turning the pages.
“However, it is also clear that since about 1980, while the total solar radiation, its ultraviolet component, and the cosmic ray intensity all exhibit the 11-year solar periodicity, there has otherwise been no significant increase in their values. In contrast, the Earth has warmed up considerably within this time period. This means that the Sun is not the cause of the present global warming.”
Someone is forgetting some very basic thermodynamics. The heat source may have reached a constant temperature, but the Earth isn’t necessarily at equilibrium with the new warmer environment yet.
Further to #17 #18
After removing trends in each series with a cubic curve,
The relationship between the stalagmite temperature and atmospheric C14 still holds:
-I must admit that I didnt expect this.
The longer-time lag between insolation and global temperatures is not in land surface, but in the oceans…
Even in the past decades, there is an increase of ~2 W/m2 in insolation in the tropics (30N-30S), due to a change in cloud cover, mainly over the oceans. At the same time there is a lot of extra heat (~5 W/m2) radiated back to space. Insolation directly affects ocean surface temperatures (and longer term ocean heat content). IR radiation may come from the surface or anywhere in the atmosphere, thus redistributing less heat towards the poles.
There is a reverse correlation between the solar cycle and cloud cover. If that holds (directly or indirectly) for longer term changes (in this case continuous higher levels) of solar activity remains to be proven.
See Chen ea.
“What is generally required [for proving solar forcing of climate change] is a consistent signal over a number of cycles (either the 11 year sunspot cycle or more long term variations), similar effects if the timeseries are split, and sufficient true degrees of freedom that the connection is significant and that it explains a non-negligible fraction of the variance.”
In my opinion, climate behaves in a far from linear way, with loads of factors to take into account, so in most cases it would be very difficult to find climate records react consistently (over several solar cycles/decades/centuries) in the same way to say a solar change (see the Hoyt & Schatten 1998 book). Even more, the solar cycles aren’t that linear/consistent either if you ask me.
Perhaps we should not only try to find (or rebut) cycles, but rather look at ‘events’. =Investigate past climate events, on a temporal and spatial scale, and check what the sun and other factors were doing at the time. For this, precise chronological control is of course a major issue.
Reconstructing or understanding climate change ain’t as easy as 1+1=2. For every theory/study there are loads of pros and cons. Indeed, perhaps we’ll never understand…
[Response: You are of course correct. There are many non-linearities in climate (and indeed solar forcing), although since some effects are clear (strat. ozone for instance), it is probably useful to occasionally look for first order linear effects. Understanding is likely only to come from mechanistic modelling of effects, rather than linear correlations. – gavin]
GCR’s, Clouds, and Climate, Oh My!
If you were at a BBQ with a group of atmospheric scientists, and wanted to raise some hackles, there are several subjects that are notoriously easy to get people riled up… …Talk of Hockey Sticks, Inhofe, Crichton, Coulter, and you’re …