Guest Post by Willis Eschenbach
Anthony recently discussed a recent paper called “The Role of Atmospheric Nuclear Explosions on the Stagnation of Global Warming in the Mid 20th Century” (PDF, author’s version). It advances the claim that nuclear tests changed the temperature in the period 1945-1980, in a sort of mini-“nuclear winter”. Here’s their main graph:
ORIGINAL CAPTION: Fig. 1 Anomaly in global-mean surface temperature [GST] between 1880 and 2008. Black line: original data and their trend (the broken line). Red triangles: eruptions whose VEI (volcanic explosivity index) is equal or greater than 5 (source). Green vertical bars: annual yield of atmospheric nuclear explosions (UNSCEAR, 2000). Blue line: corrected GST (0.3K was added to GST data of 1945 and later) based on Thompson et al. (2008) and its trend (the broken line). Red line: re-corrected GST anomaly based on effects of atmospheric nuclear explosion (∆t was set at 3 years) and Thompson et al. (2008), and its trend (the broken line). Green line: imaginary linear global warming trend. Gray line: sunspot number (source)
Something caught my eye about this graph, something that generally makes me curious.
What I found odd was the logarithmic scale on the right, for the green bars showing the “Annual yield of atmospheric nuclear explosions (MT/y).” I don’t like logarithmic scales unless there’s a good reason for them. In this case, obviously, a good reason would be if the temperature cooling effect of the bombs was proportional, not to the total yield of the explosions, but to the log of the total yield. However, this would mean that smaller explosions would cause more cooling per megatonne than large explosions, which seemed unlikely.
And in fact, their Figure 6 shows that the amount of fine dust injected into the atmosphere goes up, not logarithmically with total yield, but linearly with total bomb yield. In addition, their Figure 5 shows that the total temperature drop varies linearly with the dust concentration. Which means that the temperature drop varies linearly with the total bomb yield. So the logarithmic display is deceptive.
I got to thinking about the question of how I might falsify their claim. I realized that a) the lifetime of dust in the troposphere isn’t very long, months rather than years or decades; b) there’s only a slow exchange of air between the Northern and Southern Hemispheres; and c) the overwhelming majority of the tests were conducted in the Northern Hemisphere. The Brits blew a few off in Australia, and that was about it. China, Russia, and the US did most of the atmospheric testing, and it was virtually all north of the equator.
This means that if their theory is true, the atom bomb tests should have cooled the Northern Hemisphere more than the Southern Hemisphere.
And that, we can say something about. Figure 2 shows the HadCRUT3 Northern and Southern Hemisphere data, along with a non-logarithmic view of the annual yield of the nuclear and thermonuclear bomb tests:
Figure 2. Temperature anomalies for the Northern (blue) and Southern (red) Hemispheres. Orange circles show annual total yield of all atmospheric (above-ground) nuclear and thermonuclear bombs. Yield data from Figure 1. Vertical gray lines show the start and end of atmospheric tests and bombs, 1945-1980. Fine dust in the lower troposphere has a half-life of days/weeks, and in the upper troposphere, a few months. Stratospheric dust lasts longer, but not much of the dust made it that high.
Immediately, we can see problems. In no particular order these are:
• More than half of the total bomb yield comes from just two years, 1962 and 1963.
• The first big temperature drop takes place in the period 1945-1950, during which time there was little testing.
• During the time when most of the fine dust was injected into the atmosphere, from 1951-1963 (94% of total bomb yield), the temperature was not falling.
So it’s not looking good for the hypothesis.
However, we still haven’t examined what I set out to examine. This was the difference between the Northern and Southern Hemisphere temperatures. Figure 3 shows that difference. We would expect the line to drop if the Northern Hemisphere actually were being cooled by dust injected into the atmosphere.
Figure 3. Difference between the Northern Hemisphere and Southern Hemisphere temperatures (North minus South).
At first glance it looks like they might have something. There is a big drop in the period 1964-72. But there’s a couple of problems with that.
First, if we look at Figure 2, we see that the reason for the drop is the Southern Hemisphere is warming. The Northern Hemisphere is not cooling during that period,. False alarm.
Second, they identify the period of “stagnation of global warming as being 1945-1975. But the relative change between the two hemispheres didn’t happen until 1965.
Overall, I’d say that their explanation of the “stagnation” simply doesn’t hold water. The timing is not right, the size is not right, and the pattern of cooling is not right.
Note that the same arguments apply for the usual culprit advanced for the “stagnation”, which is aerosols, particularly sulfates. As with the bombs, the main sulfate and other aerosol sources were predominantly in the Northern Hemisphere at that time, and they last no longer in the atmosphere than does fine dust from bombs. So the lack of NH cooling argues against the sulfate/aerosol hypothesis as well.
Best to all,
w.
PS – Before you ask, yes, I know that the Partial Test Ban Treaty (PTBT) went into effect in 1963. But the Chinese and the French didn’t pay any attention to that, did they? After all, we’re talking China and France, and besides we were asking them not to do something we’d done over a hundred times.
Thanks for your thoughts as always, Mosh.
steven mosher says:
April 7, 2011 at 12:10 pm
Well … no. That’s not what the paper says. It says:
So they are talking about three things – first, fine dust in the troposphere, with a very short (one month or less) removal times.
Next, fine dust that makes it to the tropopause, which lasts longer, but which stays in the same hemisphere.
Finally, fine dust that makes it to the stratosphere. This varies in removal time from three months to 3.5 years.
Now, the dust starts out at ground level. According to the paper, on average about three quarters of it makes it to the stratosphere.
OK. A quarter has a residence time of one month. Three quarters of it have a residence time from 3 months to 3.5 years.
Average stratospheric residence time, therefore, seems to be 3.75/2 = 1.8 years. This is in agreement with other estimates, viz:
So overall, we’re looking at a residence time for the whole thing of a year or a bit more … which is what I said.
However, and most curiously, the paper uses a residence time of 5 years … I find no explanation of that.
Next, the testing through differential warming should still be valid even if the residence time is longer … because with new dust being constantly added in the north and only slow interchange between hemispheres, the majority of the dust will be in the Northern Hemisphere.
In addition, while there is more interchange between hemispheres in the stratosphere than the troposphere, the “Brewer-Dobson” circulation generally circulates the stratosphere from the Equatorial region to the extra-tropics, viz:
This, of course, tends to keep the stratosphere in the two hemispheres separated.
Finally, the inter-hemispheric temperature difference is only part of the reason why bombs don’t explain the 1945-70 temperature plateau. Neither the timing nor the size of the global temperature changes match up with the yield of the bombs.
All the best,
w.
Alexander K says:
April 7, 2011 at 2:50 am
It’s lightweight, but it has gotten quoted all over the web. Gotta fight them where they live.
w.
mike restin says:
April 7, 2011 at 3:48 pm
It was peer-reviewed and published in Energy and Environment, the journal that AGW folks love to hate … see here for a bit more.
w.
Not so sure about what your claiming.
The model they used looks like it put dust in the strat.
The emprical test they propose puts dust in the strat.
The geoengineering they reference, in the strat.
Then comes the issue raised by the video of how many explostions were in the SH,
the large number at the equator an the large number within 30 deg of the equator.
One of The models they consulted did the estimation for the strat.
5
The other model looks like it was sued to estimate the trop effect
Residence time?
You wrote “Finally, fine dust that makes it to the stratosphere. This varies in removal time from three months to 3.5 years.” No. HALF removal time. Plus, you cannot just
average 3 months and 3.5 years. Why, because 3 months looks like what you get at the poles. 3.5 years is for the equatorial stratophere. Thats from 0 to 30degrees.
see the report they site which defined the polar strat as 30-90degrees. It also shows the circulation. Then look at the explosions and where they actually happen. figure 4
page 163 of the source they site for half removal time will show you some interesting NH versus SH. not what you expect.
Your temperature graph, like most others, shows World War Two as a heat wave and cooling that starts in 1950. This is complete BS. Cooling started in the winter of 1939/1940 when the Finnish Winter War was fought in the bitter cold of minus 40 Celsius. When Hitler invaded Russia a year later the bitter cold caused many casualties and helped the Russian cause. And GIs had to fight their way from the Battle of the Bulge to the German border in the coldest winter that West Europeans could remember. My theory is that due to the war records were screwed up and later everyone just copied each other from unreliable sources that did not relate to reality.
steven mosher says:
April 7, 2011 at 11:24 pm
Yes, and I said they put dust in the stratosphere, and I quantified it. The paper says about three quarters of the dust made it to the stratosphere.
Yes, half removal time, or half-life. Pardon my error.
Actually, they were about what I expected. You need to learn to mistrust log scales like I do, Mosh, you’ve mis-read Figure 4 completely. Log scales are teh suxxor. I digitize log data and convert it back to real numbers. According to the numbers in their Figure 4, the concentrations in the two hemispheres were as follows (without the log scale):
Note a couple things. First, note that the levels drop fast from the 1963 peak. In one year they’ve dropped by just about half — what does that say about the half-life?
Notice also that the corresponding peak in the Southern Hemisphere occurs in 1964.
Finally, note that the NH peak is NINE TIMES AS LARGE as the SH peak.
So despite all of your theories, the observations show that the time constant is short, far from the 5 year half-life claimed in the paper. We also see that concentrations are much higher in the north than the south. Finally, we see that despite certainly making it to the stratosphere, only a fraction of the fallout made it to the SH. This is also confirmed by the cumulative totals shown in Figure 5 of the same paper. NH totals are about seven or eight times SH totals.
All the best,
w.
Willis,
Unfortunately, you cant use the strontium 90 distribution to talk about the distribution of the fine dust. The fine dust was only created by a selection of
all explosions. So it still takes more than the back of an envelope to determine
the distribution of fine dust into the hemispheres. Look again at the video of the
explosions, or again at the table of which explosions they used. Even in the NH
you are seeing dust injected at 30-40N at the most. Given the half removal time
of 3.5 years for dust in this zone, its not clear that this will not be mixed into
the SH. That’s why I’d suggest looking at latitude bands rather than hemispherical
figures.
Willis, plus there is a reason to use a log scale. Especially when you are talking about the obscuring effects of particles. So, you want to leave it in log scale because the effect
(blocking) is going to have a log response, depending of course upon the size of the particles in question. I’d waste some time and go look at scattering to get some more clarity
steven mosher says:
April 9, 2011 at 2:34 pm
Thanks, Mosh. Whether you are looking at “most” atmospheric explosions (what the original paper did) or “all” explosions (what the USCEAR Report did) is not a large difference. And in fact, your point works against you — many of the equatorial explosions were over the Pacific, so there would have been strontium, but little dust. So we’d expect less dust in the SH, proportional to the NH, than strontium
My point is, however, that the proportions will be not too far different. If about a quarter of the strontium made it to the south, about a quarter of the dust is very likely to have done the same.
So if so much dust is making it into the SH as you claim … why isn’t the strontium doing the same? And why did they put the graph of the strontium in there, if not to show where the fallout was falling out … which turns out to be, as I said, in the NH. Only 25% of the Strontium made it to the SH, so we can assume that even less of the dust made it to the SH.
You also say:
steven mosher says:
April 9, 2011 at 2:37 pm
While that log-relationship is certainly true for medium and high concentrations, we’re talking about very, very low concentrations which are principally in the lower stratosphere. At that level, the log and linear will only be trivially different, because both will be about linear. The log relationship is because when you get lots of particles, they start to interfere with each other. But when the numbers are tiny, it’s not a large effect. And as I mentioned above, the people who wrote the original article agree with me, they show the temperature deviation to be linear with dust, not logarithmic.
Does some of the dust make it into the SH? Sure … but as the strontium data shows … not much. There just isn’t a whole lot of interchange between the atmospheres of the two hemispheres. Even a long-lived molecule like CO2 takes a while to move from the NH to the SH.
w.
Mosh, I’d also like to know why the paper uses 5 years as the half-life of the dust in the atmosphere … do you happen to have any clues on that one? Because all of the numbers that both you and I have estimated have been well below that.
w.
Also, Mosh, according to the UNSCEAR report, the majority of the explosion ends up in the lower stratosphere. You say regarding this zone:
However, the UNSCEAR paper says:
‘
So it seems your numbers are still too high.
And in section 31, p 165, it gives two estimates of the average half-life for bomb debris in the atmosphere, which are 1.3 and 1.1 years … remember that my estimate was one year.
It also shows in Table 8 the hemispheric and zonal distribution, about which they say:
w.
Chronology of all above ground nuclear tests. http://www.johnstonsarchive.net/nuclear/atest00.html
USA popped off 216. USSR did 214, but their combined yield was nearly double the American bombs, with the Tsar Bomba contributing quite a lot.
A bit of irony, if you’ve seen “Dr. Strangelove”. The USSR actually did contemplate building a giant nuclear doomsday bomb, which would have been built into a ship cruising the north sea. Detection devices spotted around Russian territory would automatically set it off should they be attacked with nuclear bombs. Pretty much exactly the ending of the movie, but Stanley Kubrick could not then have known about the real doomsday bomb plan – which fortunately the Soviets weren’t crazy enough to actually go ahead and build.