Guest Post by Willis Eschenbach [See Update at the end]
Sometimes climate research is just plain funny. I wrote before about Irish rain and investigated whether there is any effect on the rain from the solar variations linked to sunspots. I was looking for evidence that the Svensmark hypothesis is true. Svensmark said that changes in the heliomagnetic field related to the sunspot cycle affect the number of cosmic rays (true), and that this, in turn, affects the number of clouds (unproven). I found no evidence of any such effects in the Irish rainfall data.
In response, folks said a) rain is not the proper measure of the Svensmark effect, I should look at clouds instead, and b) I should look at some place that is drier than Ireland, where it’s basically all clouds all the time.
Fair enough, reasonable objections. So I thought I’d see what other cloud datasets I could find in drier areas. Since I’d already looked (unsuccessfully) for evidence of the Svensmark effect in US cloud data, I went to see what I could find about clouds in Australia. I found some good data, but along the way, I read something pretty hilarious. It had to do with what is called the “pan evaporation” data.
The idea behind pan evaporation is simple. Put some water in a pan. See how fast it evaporates. This measurement involves wind, temperature, humidity, solar input, and rainfall. It’s an important measure for farmers, who use it to determine when and how much to water their crops.
Figure 1. Flat pan-type container used to collect pan evaporation data, along with an anemometer to measure wind speed.
When I saw a link to the Australian pan evaporation data, I thought hmmm, that could be interesting. So I took a look … and found a truly funny statement, viz:
However, the installation of birdguards on the pans during the late 1960s and early 1970s is known to have created an inhomogeneity in the climate record.
Ya think??
Seriously, they put out pans of water to see how fast they evaporated and they didn’t do anything to stop birds from drinking the water? Gotta love climate science. But I digress …
I did find cloud data, from what is called the ACORN-SAT network of climate stations. This is a network of high-quality stations with longer-term records. The ACORN-SAT data is here. Figure 2 shows the locations of the ACORN-SAT stations.
Figure 2. Locations of the ACORN-SAT Australian climate stations
Now, the indefatigable Joanne Nova has done sterling work over at her blog showing that the temperature records from the ACORN-SAT networks have been … well … let me call it “massaged” and leave it at that. It’s a shame.
But I doubt greatly whether any such massaging has been done with the cloud data, because clouds are not political like the temperature data is.
So I went and got the cloud data. Actually, it is very detailed, in that they have recorded cloud coverage both in the morning (9AM) and the afternoon (3PM). I downloaded the data, which turns out to be a total of 888 separate files … urgg. So I girded my loins and started writing computer code. I identified all of the files containing monthly cloud data for both mornings and afternoons, that’s 260 files. I extracted them, did some simple QC on them, and saved them as a pair of CSV files (morning and afternoon clouds) so folks could look at them without the hassle I had extracting them. The data runs from July 1954 to April 2015.
Once I had the cloud data, I did a straight average on them. Yes, I could probably get a more refined answer by doing some kind of geographically-weighted average, but a straight average is generally more than adequate for this type of analysis.
Then I created a periodogram of the average Australian morning and afternoon clouds, and compared that to a periodogram of the sunspots for the same period. Figure 3 shows that result:
Figure 3. Periodograms, Australian clouds and sunspots.
Once again I don’t find any sign of any relationship between the sunspots and the clouds. The clouds have the usual variety of small cycles that you find in any natural dataset, including one at 12 years … but during the period the sunspot cycles were at ten and a half years. Close, but no cigar … and close only counts in horseshoes. And hand grenades. But this is neither one of those.
So, nothing to see regarding the Svensmark effect. However, there is more to be seen in the Aussie cloud data. For starters, my hypothesis that clouds and thunderstorms act to regulate the temperature implies that cloudiness should increase with temperature. And since temperatures are higher at 3PM than at 9AM, my hypothesis would suggest that there should be more clouds in the afternoon than in the morning. I’ve shown elsewhere that this is generally true in the wet tropics, but not in a mostly desert landmass like Australia. Here is the distribution of the morning and the afternoon Australian clouds …
Figure 4. Boxplots, Australian morning and afternoon clouds. Heavy black lines show medians. If the notches of two plots do not overlap this is ‘strong evidence’ that the two medians differ (Chambers et al, 1983, p. 62). “Whiskers” extend out 1.5 times the interquartile range. Circles are outliers.
Clearly, there are more clouds in Australia in the afternoon when it is warmer, as my hypothesis implies.
How much difference will the cloudiness make in the amount of sunlight that gets past the clouds to the surface? We can estimate this by using the CERES data. Here is the scatterplot of Australian cloud coverage (%) and Australian cloud solar reflections (W/m2).
Figure 5. Scatterplot, CERES monthly cloud coverage (%) and cloud solar reflections (W/m2) for Australia.
Now, the average difference between morning and afternoon clouds in Australia is 4.6% … which would imply a decrease in afternoon sunshine on the order of 4.6 W/m2. This is a significant amount of cooling.
There’s another way to consider this, again using the CERES data. This is to look at the relationship between the cloud reflections and the temperature.
Figure 6. Scatterplot, CERES monthly cloud coverage (%) and temperature (°C) for Australia.
This shows that in Australia when the temperature goes up by 1°C, the increased clouds reflect an additional 1.4 W/m2. Again, this is in agreement with my hypothesis about the clouds being part of the temperature regulating system.
Finally, note that this is just one of the ways that the cooling increases with the clouds. In addition to the change in the amount of reflected sunlight, there are the effects of rain and thunderstorms. Both of these cool the surface strongly through a variety of effects. See my post entitled Air Conditioning Nairobi, Refrigerating The Planet for a discussion of these effects.
So that’s the result of my wandering around the Australia outback … no sunspot effects on the clouds, and increasing clouds with temperature. Plus a good laugh about the birds drinking out of the Australian evaporation pans …
My best regards to all of my Aussie mates, it’s a great country with interesting folks. If you haven’t visited there you should … and if you are Down Under and you need a tattoo, and who doesn’t need one, go see my mate Tu and his lovely wife Ify, they are both fantastic artists.
And of course, warmest wishes to everyone,
w.
PS: As always, I politely request that you quote the exact words that you are discussing, so we can all be clear who and what you are talking about. Misunderstandings are the bane of the internet. Plus, if you wish to refute something you need to quote exactly what it is that you are refuting. I ask politely, but if you don’t quote what you are talking about, I may indeed cast aspersions on your cranial horsepower or the personal habits of your antecedents …
[UPDATE] A commenter asked to see the CEEMD analysis of the Australian clouds. Here it is …
As you can see, there is no solar signal visible in the cloud data.
I dont think counting clouds in order to find a sunspot signal is the correct way to do it. The process of how clouds created by moist air raising is not going to make the cloud bigger or make more clouds appear due to increase in cosmic rays. The clouds will stil be the of the same size due to the size of the bubble of moist air rising. Possibly it will be slightly brighter.
If I do not remember wrong, I heard Svensmark mentioning the sky or clouds becomming brighter and thus increase earths albedo.
That means that even a cloud free sky can be affected. Even if the sky is blue it is not free of water vapor or water droplets hovering up there. The blue sky would become slightly whiter if GCR alters the formation rate of water droplets.
If one would want to find the effect of GCR modulated by the sunpspot cycle and cloudreflectivity I think the best way to do that is to look at earths albedo from space.
I do not belive GCR can create a cumulus cloud or increase the extent of a massive cloud cover, at most GCR can make them slightly brighter/thicker.
Innit great how some folks cling to their ‘baby’ -Cosmic Cloud Nucleation.
The Magical Thought Bubble at work, watch for it in yourself.
In the final final ultimate limiting case of *nothing* else to nucleate clouds (stuff like dust and the vast litany of organic stuff that gets whooshed up into the sky by weather, NOT Long Welling Radations) then Cosmic Rays might have a measurable effect on cloud creation. Yes?
So……
Taking it that the contents of the Solar Wind are, to a first approximation, similar to what’s in Cosmic Rays….
(Not as energetic but much greater in quantity)
So why is the Aurora Borealis even visible from the ground?
Wouldn’t all that incoming solar stuff make clouds and thus hide the Aurora?
Peta,
No, the solar wind wouldn’t make clouds to obscure an aurora.
Cosmic rays aren’t just a little bit more energetic than the solar wind, the thermal energy of which lies between 1.5 and 10 keV. GCRs are many orders of magnitude more energetic, reaching at the upper end 3 × 10^20 eV.
GCRs collide with air molecules, producing secondary particles, some of which decay to muons capable of reaching Earth’s surface. Muons can even penetrate for some distance into shallow mines.
https://www.climato-realistes.fr/svensmark-cloud-cern/
https://huemaurice5.blogspot.fr
Another problem with pan evaporation data, apart from water being disturbed by bathing magpies or scoffed by thirsty camels, is possible changes in local wind patterns due to changes in the immediate “topography” surrounding weather stations. Encroaching buildings can affect the temperature at a recording site.
Similarly encroaching buildings or changes in vegetation can effect wind patterns and, therefore, evaporation rates.
https://journals.ametsoc.org/doi/full/10.1175/JCLI4181.1
Sunspots are not Cosmicrays. If you try do compare something and get results. Don’t use apples when onion is needed!
i’m super impressed by the incredible devotion to science that boffins have scrutinized nature at such fine resolution to produce an algorithm that adjusts the data for the effects of an average bird on an average day with an average thirst in the morning. models are freakin awesome.
and this average thirst of this average australian bird is known to be representative of approximately 1500 sq km gridcell so interpolation is no porblemo
Yeah well, neglecting bird covers ain’t the half of it…. Australia is the stupidest country in existence. They actually have a Prime Minister who thinks that pumping water up a mountain to fill a dam will create enough hydro electric power to be viable….. They are calling it the Snowy River Scheme 2.0. I kid you not.
Yep. I suggest we nuke them from orbit….. It’s the only way to be sure. If that kind of stupid ever gets out, it’s game over man…. game over.
Willis, do you agree that the CEEMD analysis you posted show something for substantiating an increase in cloud cover when the solar sunspot count reaches its peak, with the exception of the breakdown beginning year ~2000? Could the inverse relationship change to one in lockstep due to some other forcing or effect that is also affecting winter Arctic temperatures >80°N as per DMI’s graph beginning in 2005 which shows the start of repeated warm spikes above the winter long-term median? Or could the spate of warmer Arctic temperatures cause changes in global cloud cover patterns which shows the shift?
I’m really grasping at straws here, of course, but it seems that much of the brouhaha of CAGW first is substantiated post-2000, especially with sustained long-fetch winds removing large portions of thick, old sea ice through the Fram Strait. Or perhaps changes in rainfall frequency and/or intensity alters the pattern of cloud cover to show a shift relative to SSN’s?
Wolf April 4, 2018 at 12:58 am
I’ll take a look.
For a man who hasn’t done one single analysis looking for observational evidence of the Svensmark effect, you sure have an inflated opinion of the value of your thoughts on the subject … I’ve looked lots of places, many of them suggested by people just as knowledgeable as you seem to think you are, without finding anything.
Your implication that I’m deliberately avoiding looking where it might be found is asinine, unpleasant, and totally untrue.
w.
Well, Wolf, I haven’t found any long-term data for the ITCZ. I do have the TRMM rainfall data which looks like this:
I’ve highlighted the area I’m looking at in red. Here is the CEEMD analysis of that data:
Any suggestions you might have of better data for the ITCZ would be useful …
w.
“Any suggestions you might have of better data for the ITCZ would be useful”
Willis, I said the net effect of the atmosphere over the oceans was to cool them. You bought into the crazed idea that net atmospheric effects including radiative exchange was to warm the oceans.
You are stuck with that. There can be no forgiveness.
Your work on the cloud thermostat effect is excellent. But your shame in promoting the AGW conjecture must burn forever.
The Wolf said “surface Tav without radiativly cooled atmosphere will be above 300 kelvin”.
If you want to challenge Willis, you are going to be better than Monckton. (And he already folded).
Are you going to step up to the plate, sunshine?
Wolf April 6, 2018 at 3:29 am
Quote where I said that. I don’t recall saying anything like that.
Quote where I have said any such thing.
Sorry, I’m not dealing with your imaginary planets. You’ll have to build them on your own time.
Sure … as soon as you QUOTE WHATEVER IT IS THAT YOU ARE BABBLING ABOUT … sunshine.
w.
“Sorry, I’m not dealing with your imaginary planets. You’ll have to build them on your own time”
I only deal with real planets and solar thermal gain in real materials.
I believe imaginary planets with infinitely thin, infinitely conductive surfaces, no thermal inertia, constantly illuminated by a ¼ power sun and able to absorb and emit equally at all frequencies is your playing field, not mine.
Willis, the foundation claim of the AGW conjecture is that the surface temperature of this planet would only average 255K (-18C) in the absence of radiative atmosphere.
Do you agree with this foundation claim?
Wolf April 7, 2018 at 10:42 pm
First, AGW doesn’t rise or fall based on the specifics of the claimed temperature without GHGs, so it is hardly a “foundation claim”.
Second, this is a very difficult question, to which no one has a satisfactory answer. Dr. Robert Brown wrote a most informative post on this question. Hang on, let me see if I can find it … OK, here you go. Bear in mind that Dr. Brown teaches physics at Duke …
Best regards,
w.
So that’s the result of my wandering around the Australia outback … no sunspot effects on the clouds, and increasing clouds with temperature.
Why don’t you plot/regress the am cloud cover and pm cloud cover versus the sunspots? It seems to me that is the effect that you are looking for.
Thanks again for your essay. Bravo! on the heroic downloads. What a chore.
Noise … both datasets are quite noisy, so regression is the wrong tool. Here are the regressions. First, morning clouds as a function of spots:
Residuals: Min 1Q Median 3Q Max -17.419 -4.332 0.079 4.229 21.929 Coefficients: Estimate Std. Error t value Pr(>|t|) (Intercept) 46.848989 0.373434 125.45 <2e-16 *** as.vector(b) -0.002632 0.002991 -0.88 0.379 --- Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1 Residual standard error: 6.163 on 728 degrees of freedom Multiple R-squared: 0.001063, Adjusted R-squared: -0.0003091 F-statistic: 0.7747 on 1 and 728 DF, p-value: 0.3791And here’s the afternoon clouds …
Residuals: Min 1Q Median 3Q Max -19.5144 -3.9522 0.1129 4.2605 18.4014 Coefficients: Estimate Std. Error t value Pr(>|t|) (Intercept) 51.337550 0.380346 134.976 <2e-16 *** as.vector(b) -0.002557 0.003046 -0.839 0.402 --- Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1 Residual standard error: 6.277 on 728 degrees of freedom Multiple R-squared: 0.0009666, Adjusted R-squared: -0.0004057 F-statistic: 0.7044 on 1 and 728 DF, p-value: 0.4016I do like the R^2 … 0.0009666 …
w.
Thank you. I thought that you’d tell me to do it.
matthewrmarler April 4, 2018 at 6:50 pm
You’re welcome. I do my best to fulfill reasonable requests from reasonable people, particularly when I don’t know the answer. You have always been courteous, you have focused on the science, and you’ve never sunk to personal attacks.
My goal is to mirror the attitude of the people who direct their comments at me. If they come out talking about the science, I’m more than happy to discuss the science. If they come out attacking me, I’m more than happy to hit back twice as hard. When I hit back twice as hard, people think I’m angry, but I’m not. It’s all fun to me, I’m laughing at the contortions that they twist themselves into trying to bite my ankles. There’s nothing in words to get angry about.
Best regards, and thanks for the nature and style of your participation in the ongoing climate dialogue, much appreciated.
w.
Willis Eschenbach: I do my best to fulfill reasonable requests from reasonable people, particularly when I don’t know the answer. You have always been courteous, you have focused on the science, and you’ve never sunk to personal attacks.
In that case, why don’t you try the regression again with both temperature and sunspot number as predictors, since you have a solid case that the temperature is important.
If Svensmark’s hypothesis is true, I expect the R^2 to be small, under 1%.
Here you go, Matt.
Call: lm(formula = cld ~ sunspts + temperature) Residuals: Min 1Q Median 3Q Max -15.956 -4.088 -0.015 3.738 21.799 Coefficients: Estimate Std. Error t value Pr(>|t|) (Intercept) 40.159469 1.081691 37.127 < 2e-16 *** sunspts -0.002376 0.002911 -0.816 0.415 temperature 0.303789 0.046061 6.595 8.31e-11 *** --- Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1 Residual standard error: 5.99 on 707 degrees of freedom (1 observation deleted due to missingness) Multiple R-squared: 0.0591, Adjusted R-squared: 0.05643 F-statistic: 22.2 on 2 and 707 DF, p-value: 4.451e-10Sunspots not meaningful, temperature is …
w.
Here you go, Matt.
Thanks again.
Willis, may have sent you this elsewhere….
https://m.youtube.com/watch?v=_7RBJia7Q-w
(14 mins in)