Published on:

Correlation ain’t causation.

But for some time I have been noticing that the correlations between certain aspects of solar activity and certain aspects of climate are getting really rather impressive — far more so than anything relating to carbon dioxide.

Carbon dioxide certainly can affect climate, but so for sure can other things, and in explaining the ups and downs of past climate, before industrialisation, variations in the sun are looking better and better as an explanation. That does not mean the sun causes current climate change, but it certainly suggests that it is at least possible that forcings more powerful than carbon dioxide could be at work.

I am not yet a solar partisan in this debate, nor do I plan to become one. But I find the hypothesis that solar variation has been stronger than carbon dioxide in recent decades sufficiently intriguing that I do not see why it should be dismissed yet.

Here are some of the correlations that have impressed me. Some may be wrong, or misleading. Some come from more trustworthy causes sources than others. Some might have been smoothed or otherwise manipulated. I don’t really know. But it’s interesting to lay them out.

First the Parana river flow and sunspot number:

Second the Nile river flow and sunspot number:


Third, the correlation between cosmic ray flux (as measured by beryllium and carbon isotope levels) and climate in Greenland. Cosmic ray flux varies inversely with solar activity:



Fourth, cloud cover and sunspot number:


Fifth, carbon 14 and climate in Austrian speleothems (can’t get a good picture off the pdf).


Sixth, solar magnetic variation and rate of change of temperature (interannual variation):



Now, I repeat, correlation does not mean causation, and I stress that no individual data set can be considered reliable till thoroughly checked and replicated. So I am not saying all climate change is caused by the sun, nor that recent climate change must have been caused by the sun or anything like that.


1, Some correlations exist

2. Plausible mechanisms exist (stronger solar activity leads to better shielding of the earth from cosmic rays leads to less cloud leads to more warmth)

3. The order of magnitude is not wrong. Carbon partisans will argue that total solar irradiance varies by tenths of a percent over the solar cycle, but UV output and magnetic field strength vary by far greater amounts.

4. The only reason for ruling it out a priori that I have come across is that solar activity has not increased markedly in recent decades while climate has warmed. You hear this used as a final dismissal of solar theories in the most emphatic way (here’s an example:

The scientists’ main approach was simple: to look at solar output and cosmic ray intensity over the last 30-40 years, and compare those trends with the graph for global average surface temperature.

And the results were clear. Warming in the last 20 to 40 years can’t have been caused by solar activity,” said Dr Piers Forster from Leeds University, a leading contributor to this year’s Intergovernmental Panel on Climate Change (IPCC).)

— and it nearly always goes back to one paper, by Lockwood and Frohlich:

Here we show that over the past 20 years, all the trends in the Sun that could have had an influence on the Earth’s climate have been in the opposite direction to that required to explain the observed rise in global mean temperatures.

But this argument assumed a lag between forcing and equilibration (ie, how long before the earth stops warming after a warming stimulus stops changing) of less than ten years. Here is what Lockwood toldAlec Rawls, whose interesting post on this subject caught my attention last week:

In the attached paper, we cite a paper by Schwartz (2007) that discusses and quantifies the heat capacity of the oceans relevant to GMAST changes and so what the relevant response time constant is. It is a paper that has attracted some criticism but I think it is a good statement of the issues even if the numbers may not always be right. In a subsequent reply to comments he arrives at a time constant of 10 years. Almost all estimates have been in the 1-10 year range.

In the attached paper we looked at the effect of response time constants between 1-10 years and showed that they cannot be used to fit the solar data to the observed GMAST rise. Put simply. The peak solar activity in 1985 would have caused peak GMAST before 1995 if the solar change was the cause of the GMAST rise before 1985.


So solar activity peaked in 1985 and the earth should have stopped warming by 1995.

Two immediate problems with this argument are immediately obvious. First, the earth did stop warming around 2000, so there is not a lot of difference. As Rawls puts it:

If the solar activity peak shifts five years then Mike’s temperature response formula says peak GMAST should have occurred by 2000, which is pretty close to when it did occur.

Second, Rawls points out that short equilibration times don’t suit the carbon partisans, because they leave them with too little unfelt warming in the pipeline. Therefore:

Gavin Schmidt recently had occasion to  comment on the time to equilibrium:

“Oceans have such a large heat capacity that it takes decades to hundreds of years for them to equilibrate to a new forcing.”

This is not an unconsidered remark. Schmidt was one of co-authors of The Team’s response to Schwartz. Thus Mike Lockwood’s suggestion that “[a]lmost all estimates have been in the 1-10 year range,” is at the very least passé. The clearly increased realism of the two reservoir model makes it perfectly plausible that the actual speed of equilibration-especially in response to a long period forcing-could be quite slow.


So the rejection of the solar hypothesis rests on an assumption that causes real problems for the carbon dioxide hypothesis. Either the earth is quick to adjust to a warming factor, in which case the mild warming of the past few decades shows that earth’s sensitivity to carbon dioxide is much lower than the models assume, or there is still lots of warming still to come, in which case it might have been caused by the sun’s ramp-up of its activity prior to 1985.

Either way, it’s wholly premature to say that the solar hypothesis has been investigated and disproved.

There is a startlingly good brief talk by Vincent Courtillot here. It is moderate in tone, impeccably rational and translucently clear.

In it he argues that the actual empirical support for the IPCC’s climate models is poor, while the empirical support for a powerful role for variations in the solar magnetic field as a driver of climate change on earth is good.

He argues that the solar effect could vary by as much as 8 watts per squer metre, or roughly double the forcing of carbon dioxide.

Is he wrong?


By Matt Ridley | Tagged:  general