Thursday, April 12, 2007

Found in the margins

Recently, Eli has been improving his German over at Oekolismus, drawn there at first by the reappearance of the good Diplom Beck and his Jaworowski-like “180 years of CO2 gas analysis by chemical methods” which Princess Denial, Sonja Boehmer-Christiansen published in Energy and Environment in an attempt to lock up this year's S. Fred. Eli dealt with the subject matter in a friendly matter earlier. He will not repeat that mistake. As for Jaworowski, well Jim Easter in Some are Boojums has torn that fakery apart. You can get the flavor from

In Dashiell Hammett’s story The Golden Horseshoe, much of the action takes place in a bar of that name in Tijuana. At one point the narrator, an operative for the Continental Detective Agency, kills a few strategic seconds by studying the decorations:

I was reading a sign high on the wall behind the bar:
I was trying to count how many lies could be found in those nine words, and had reached four, with promise of more …

Sometimes I come across an article, web posting, advertisement or other statement that makes me feel when I read it just as I imagine the Continental Op did in that Tijuana bar.
How can they possibly pack so much misinformation into such a small space?
To honor exceptional achievement in mendacity, I would like to present the Golden Horseshoe Award to that writer who has out-performed his or her peers in density of false statements per column-inch.
To receive the first Golden Horseshoe Award, I can think of no more worthy recipient than Zbigniew Jaworowski.
If you have not had the opportunity, go read the lot. If you don't have time read this. But let us reluctantly return to the good Diplom Beck and the to and fro in Oekolismus.

To summarize, there were several sets of atmospheric CO2 measurements made by wet chemical methods between ~1870 and 1970. Many of them have been critically reviewed, most recently by Eric From and Charles Keeling. Suffice it to say that there were both methodological and meteorological problems with many of the earlier methods, although some are better than others. At best the accuracy of the wet methods was more than 1%, more often like 3%, which is somewhere between 3 and 10 ppm. The infrared absorption method introduced by Keeling had an accuracy of a part in four thousand with proper calibration and is the gold standard.

John A. is the hero in this story. A few weeks ago, he looked at Rabett Run's posting on the matter and commented:
Re: the two links to Keeling. The first one requires a login and password, and the second link goes to a "404" message.

I'm impressed.
The link was to a history of the Mauna Loa Observatory CO2 measurements written by Keeling in 1993. In searching for another version, Eli came across a 1998 Keeling autobiography which appeared in Annual Reviews of Energy and Environment, "The Rewards and Penalties of Monitoring the Earth". It answers several questions.

One of the issues about the wet chemical measurements is how well they were carried out. There is an interesting comment in the Keeling autobiography:
This Scandinavian program, started by Rossby in 1954, had been a major factor in triggering interest in measuring CO2 during the IGY. Nevertheless it was quietly abandoned after the meeting, when the reported range in concentrations, 150–450 ppm, was seen to reflect large errors. 3

3. At two stations in Finland, samples collected by station personnel had been sent to Scripps. These samples yielded nearly the same concentrations as those measured at Mauna Loa Observatory, proving that the errors in the Scandinavian program were mainly analytical rather than due to variable CO2 in the air being sampled.
Another is about how well the MLO and other observatory measurements match those high in the atmosphere
You may have to click on the image to see it clearly

The solid line is from a series of aircraft measurements. It clearly follows the Mauna Loa measurements, showing that the CO2 mixing ratios measured there are equivalent to those measured at altitude.

Yet another issue is the daily variation of CO2 concentrations, something that Keeling had sussed out quite early (long but interesting excerpt)
. . .these data showed an intriguing diurnal pattern. The air contained more CO2 at night than during the day. Also, the heavier carbon-13 isotope of the CO2 at night was depleted with respect to the lighter carbon- 12 isotope, as though the CO2 that caused the nighttime rise had been released by the plants and soil. The degree of depletion of carbon-13 for a given rise in CO2 concentration varied from site to site in a manner suggesting that the plants during daytime at some sites reabsorbed CO2 previously released into the air locally the night before....
The diurnal patterns were similar everywhere I went, from the rain forests of the Olympic peninsula near Canada to the high mountain forests of Arizona near Mexico. (US National Forests at that time had large tracts of land not yet disturbed by logging.) Moreover, the air in the afternoon seemed always to have nearly the same amount of CO2, about 310 parts per million (ppm) of air, after correcting for water vapor. The concentrations were highly variable at night and always higher than in the afternoon. Also, the carbon isotopic ratios in the afternoon were all about the same, though systematically variable with concentration at night.

The scientific literature didn’t suggest that daytime concentrations should be so similar from place to place. A recently published book on geochemistry (53) indicated that arctic air could contain as little as 150 ppm, tropical air as much as 350 ppm. Moreover, photosynthesis by plants in the area of my sampling should have drawn CO2 down during the day, making the concentration lower than in air over bare ground. I broadened my study by sampling on a high mountain during strong winds over barren ground. . . . .

Even at these places, sampled in the free atmosphere, the concentrations and carbon isotopic ratios were nearly the same as in the afternoon near vegetation (30, 32). Why didn’t photosynthesis, which takes CO2 out of the air during the day, cause low and variable concentrations when respiration by plants and soil, which puts CO2 into the air at night, causes high and variable concentrations? I found an explanation in a book that attracted my attention because of its apt title: The Climate Near the Ground (21). All of my forest measurements had been made during fair weather. On such days heating by the Sun typically induces enough turbulence in air near plants to cause thorough mixing of this air with the free atmosphere by early afternoon. Where I had sampled, the free air evidently had been of nearly constant composition with respect to CO2. In contrast, during the nighttime the air near the ground cooled, forming a stable layer that allowed CO2 from respiration to build up within the forest canopy.

The highly variable literature values for CO2 in the free atmosphere were evidently not correct.1 Rather, a concentration of 310 ppm of CO2 appeared to prevail over large regions of the northern hemisphere. I had detected this near constancy under the implausible circumstances of studying air in old-growth forests where variability was to be expected.
The 1986 From and Keeling paper deals with the many problems of the data sets taken before 1905 and there are other indications of problems with the subsequent wet chemical measurements. For example, a paper by Lockhart and Court describes 1940 and 41 measurements of air composition in Antarctica. They find abnormally high CO2 mixing ratios, 600 ppm, but they also find abnormally LOW oxygen ratios, and there is no indication of measurements on calibration mixes. A big no-no.

The 1955 Fonselius, Koroleff and Buch paper would have won the S. Fred with Golden Horseshoes. One of the things you have to do to measure CO2 content is sample the air. These guys could not have thought of a worse way to do it.

You may have to click on the image to see it clearly

The right way to sample is to use large evacuated bulbs which you open rapidly to sample. You always use carefully calibrated mixes to calibrate your measurements. You really want to understand and eliminate wall effects by making the volume large compared to the surface, optimizing pumping, etc. Keeling, for example, used 5 L bulbs when he started his California measurements.

The Buch method is an invitation to error. They have a small 250 cc cylinder filled with air one end of which is connected to a rubber bulb. They opened both stopcocks and held the bulb outstretched in Mr. Roland Ploennige's arm and pumped the thing a 150-200 times. The flow of air through the cylinder is not constant. This (a) is pretty close to Mr. Roland Ploennige, (b) does not guarantee a complete exchange of air in the bulb, (c) encourages backflow from the opposite end which is pretty close to Mr. Roland Ploennige's armpit if not his mouth (d) is not much volume to work (e) it is not clear if the measurements were calibrated against standard samples. That is for starters.

This is not the gold standard.


guthrie said...

Ahhh, wet chemistry techniques. Just like the good old days.
So many sources of error, I am amazed they managed to get results at all. But then Eli has already pointed out the most important problem- sampling bias.

A quick straw poll- how many people here were told about sampling bias and other sources of error at university, assuming they did a degree that might involve sampling in some way?

(My bog standard chemistry degree didn't as far as I can recall)

Magnus said...


Anonymous said...

Hey, Eli. I noticed you commented on Lubos Motl's recent global warming post. I was just wondering if you followed the entire debate, including my response. If you did, I would love any feedback, good or bad, from you. I admire your work and your blog, so any advice you have would be greatly appreciated.

jre said...

Eli -
Thanks for the notice. One thing often overlooked (not that anyone should care) is that the Jaworowski post is actually seven posts, because I put all the discussion of topics such as J's 1992 and 1994 papers, Mulvaney, Oeschger, etc., in six subsidiary posts, linked to them and hid the lot under the "References" category. I mention it now because, as it happens, Fonselius et al's 1955 paper was the object of one of Jaworowski's most egregious distortions.
I discussed it here.
Fonselius seems to have been keenly aware of the limitations of measurement technology. In reference to his plot of historical measurements, for example, he says "This scatter … probably depends on the rather crude and incomplete techniques of analysis used during the first fifty years, possibly with systematic positive errors."
He also refers approvingly to the criteria G. S. Callendar used to winnow reliable from unreliable data.
Jaworowski used a little artistic license to redraw Fonselius' plot, then claimed that it showed a "biased selection."
Which takes, I think, balls of brass if not osmium.

jre said...

The review of historical CO2 measurement was

Fonselius, S.; Koroleff, F. and Wärme, Carbon Dioxide Variations in the Atmosphere, Tellus VIII (1956), 176-183

So it's Fonselius et al. 1956, not 1955, as I confusingly said above. Sorry.

Anonymous said...

"John A. is the hero in this story"

Again? Man, he's always the hero.

What cockeyed claim was he making this time?

Wait, let me guess: "Measurements of atmospheric CO2 concentration are meaningless, not unlike measurements of surface air temperature on a windless, cloudy day".

Anonymous said...

your article is full of errors. Please take the time to read Beck´s paper. Obviously you cannot.
1. There were >380 papers on CO2 from 1800-1952.
2. Keeling and From mentioned (NOT REVEIWED!)18, this is not MANY but 5%,and they praised 3 as precice.
3. Keelings NDIR accuracy was 4 ppm up to 1968 (see references)
4. The Scandinavian program by Fonselius and Bischof produced horrible results with no accurate seasonal variation because of methodical errors
5. The measurements by Lockhart in 1940 were made with the famous Haldane gas analyser calibrated by v. Slyke and thousandfold in use in prior times. Lockhart and Court were experienced scientists of Harvard university.
So why writing such a bullshit?

Anonymous said...

Dear Anon 10:34,

"1. There were >380 papers on CO2 from 1800-1952."
And 90000 measurements, and tens of Nobel prizes. So what ? Doesn't prevent the CO2 flying circus to be a flying circus. BTW according to Larouche and some others there were only 175. Or did Beck dig 205 other papers in the meantime ?
"2. Keeling and From mentioned (NOT REVEIWED!)18, this is not MANY but 5%,and they praised 3 as precice."
So what ? They might have some scientific reasons to do that.
"3. Keelings NDIR accuracy was 4 ppm up to 1968 (see references)"
Yes, thanks for references. The CDIAC data rather mentions 0.2 ppm.
"5. The measurements by Lockhart in 1940 were made with the famous Haldane gas analyser calibrated by v. Slyke and thousandfold in use in prior times. Lockhart and Court were experienced scientists of Harvard university."
OK so those measurements make authority, amen.
BTW, having seen the Lockhart paper, the CO2 concentration was 900 ppm on Dec 31, 1940; climbed to 1700 ppm on Jan 2, 1941, then down to 1100 ppm ( Jan 4), 400 ppm (Jan 5), up to 1600 ppm on Jan 6, etc..., with an error of 300 ppm for each measurement. So what is the valid value ? The mean value ? The lowest one ? The highest ?

"So why writing such a bullshit?"
Indeed, flying circuses are full of horseshits!



EliRabett said...

From and Keeling reviewed all the 1880 - 1900 papers. They thought three were trustworthy to some degree.

Indeed Beck's arguments from authority are, well, weird.