Blam! Blam! Willis Eschenbach takes a scattergun to solar-temperature datasets.

WUWT’s resident cowboy-turned-climate-expert WIllis Eschenbach responded to a comment I recently made on his thread ‘Cycles Without the Mania’, where he set out to prove a negative – that the rest of the Solar system can’t have any effect on the Sun, and thereby, Earth’s climate systems. It was a followup riposte to Nicola’ Scafetta’s latest paper, under discussion here at the Talkshop.

This was my comment:

tallbloke says:
July 30, 2013 at 12:04 am

There are roughly speaking, three el ninos per solar cycle. the big one occurs soon after solar minimum when the ocean goes into reverse and kicks heat out instead of absorbing it, unless there was a volcanic eruption in the previous cycle, in which case the PWP is already partially discharged. Consequently the following big la nina usually occurs near solar max. That’s why there’s often a dip in global T near solar max.

If you smooth the temperature data at 1/3 solar cycle or at the average frequency of the ENSO cycle (around 40 months) you get a good correlation between solar activity and global temperature. The amplitude isn’t that big, but this is due to the antiphase nature of ENSO surface temps and solar cycle described above. The Sun is having a large effect, but it’s hidden below the surface, most obviously in the Pacific Warm Pool.

http://woodfortrees.org/plot/hadsst3sh/from:1955/mean:37/detrend:0.6/plot/sidc-ssn/from:1955/scale:0.001/offset:-0.3/mean:12

All fairly obvious easily empirically observed from the records, and not controversial, or so I thought.
But Willis wanted to set me straight:

Willis Eschenbach says:
July 30, 2013 at 7:23 pm

Thanks, tallbloke. If you’re going to say that there is a relationship between two datasets, say HadCRUT temperatures and sunspots, there’s a couple of things you need to do.

First, you need to use all of the data available. You’ve only shown the data from 1955.

Next, you need to actually calculate the statistical significance of the relationship.

Finally, I prefer scatterplots to time series plots because the human brain is a cycle-finding machine. We survive in large measure because of our ability to anticipate what’s going to happen, based on the fact that we’ve seen a particular cycle before. As a result, we tend to see relationships where none may exist.

With that as a preface, here’s the long-term relationship between sunspots and HadCRUT4 temperatures:

It’s at this point Willis pulled out a scattergun from under his Lee van Cleef style coat and;

BLAM!

And the numbers:

Coefficients:
Estimate Std. Error t value Pr(>|t|)
(Intercept) -4.078e-03 6.888e-03 -0.592 0.554
ssnhadwindow 7.442e-05 9.591e-05 0.776 0.438
The right-hand column show the statistical significance of the intercept and the trend. They need to be below 0.05 to be considered significant …

Here is the same thing, for the BEST data (land only):

BLAM!

And the corresponding numbers:

Coefficients:
Estimate Std. Error t value Pr(>|t|)
(Intercept) 0.0189089 0.0139720 1.353 0.1761
ssnhadwindow -0.0003451 0.0001945 -1.774 0.0762
Now, that one is approaching statistical significance for the trend (although not for the intercept). However, look at the size of the trend—for every hundred sunspots, the land is (supposedly) cooling by three hundredths of a degree … and max sunspots is about 200.

So the maximum effect, IF it were statistically significant, is a COOLING as the sun’s activity INCREASES, of about six hundredths of a degree.

Nothing left of the data to see here folks, move along. I decided not to repond to this. At the time I was embroiled in a rather unpleasant dispute with Leif Svalgaard, who had just falsely accused me of fabricating a quote from exoplanet researcher Katja Poppenhaeger. However, Talkshop contributor Bart took on the task and commented on Willis’ statistical expertise:

1. Bart says:

   July 31, 2013 at 12:06 am

   Willis Eschenbach says:
   July 30, 2013 at 7:23 pm

   Willis – of course you are not going to see a relationship in the way you are doing it. If you quickly turn the heat up and down under a pot of water, do you expect you would see a relationship between your instantaneous twisting of the knob and the temperature of the water? Of course not. The scatter plot would look like an amorphous cloud of dots, just like these.

   You are dealing with a system with enormous thermal mass. I recommend, as a first cut, that you try filtering the sunspot data to lower and lower bandwidth, up to the limiting case of a pure integration. Then see if you don’t start to see a correlation at some point.

   There are far more sophisticated techniques available, but, you might find something this way.
2. Bart says:

   July 31, 2013 at 12:08 am

   And, don’t detrend the temperature data. The low frequency regime is where you are going to see the action, due to the low pass characteristic of all that thermal mass.

I started work on integrating the solar data to make it more comparable to the high-mass slowly warming and cooling ocean in 2008, culminating this year in my simple model which contains a major solar integration component and successfully replicates HadSSTV3 to an R^2 of 0.9 using monthly data back to 1874.