Friday, December 11, 2009

Did the hockey stick ever matter?

Climate alarmist Stephen Schneider wrote an article for the Huffington Post where he argues that the hockey stick graph
  • has never been disproved
  • was never a basis for AGW likelihood assessments
Holy cow. These people usually rely on the stupidity of the laymen - after all, they are severely limited themselves, or at least they try to be - but I think that if they rely on too huge stupidity of the laymen, their expectations are likely to drag them into trouble.

It's because this story is really not difficult and many people fail to be sufficiently stupid. See also Roger Pielke Jr's blog.

Hockey stick methodology is flawed

First, the hockey stick graphs are known to be wrong beyond any reasonable doubt today and I am confident that even a majority of the "neutral" readers of this blog have either understood the reason, or they will understand it now.

The basic point of Mann's methodology is that he chooses proxies (trees etc.) - or gives them weight - according to their agreement with the 20th century temperatures measured by the thermometers (which have been mostly increasing).

The logic behind this selection (or bias) is that the trees that show the correlation with the 20th century temperature are more likely to be sensitive to temperature.

However, a significant fraction of the proxies will be correlated with the 20th century temperature - which essentially means that they will show the increasing trend in the 20th century - by pure chance, even though they are not good temperature proxies. A portion of the trees and other proxies satisfies this condition. Statistics makes it inevitable.

They are selected - or given bigger weights - and it is clear what the result is going to be. For the 20th century, their average will be close to the instrumentally measured temperature - because it was a necessary condition for the trees to get a voice. On the other hand, there was no condition for their behavior in the previous centuries. So they behave as simple noise and their average trend for every century except for the 20th century is essentially zero.

Consequently, the hockey stick is obtained even from red noise i.e. randomly, continuously changing temperatures following a "random walk" - and a part of the proxies in the real world are actually red noise. See a Mathematica notebook that reproduces this mechanism:
PDF preview, NB file
The preview contains a nice, sharp hockey stick obtained from pure red noise. Check it out.

(By the way, someone has complained that the tree data are not red noise. Well, the quantities they reflect are clearly continuous - as seen from all reconstructions - so they can't possibly encode a white noise. The very continuous character of the hockey stick and other graphs proves that the underlying noise is red, or at least more red than white. Needless to say, the argument above works for such "pink" colors of the noise, too. If white noise were inserted into Mann's algorithm, it would still produce the increasing "blade" aligned to the instrumental temperatures, for the same reason as for any other noise, but the previous centuries would remain white noise and there could be a sudden temperature drop at the beginning of the instrumental period.)

The mechanism above may look sophisticated. However, Phil Jones summarized it in a simple way: the trees are used to reconstruct the temperatures before the 20th century. They show almost no trend or significant variations because they're way too random and they're sensitive to other parameters. However, the 20th century part of the graph has been de facto replaced by the instrumental temperatures (by the condition that the trees have to be correlated with it) to "hide the decline".

The result is an inconsistent mixture of data from two sources: the distant past shows chaotic noise from the real trees while the recent century shows the increasing temperatures measured by the thermometers (because even if the data come from the trees, it is only the trees that agreed with the thermometers well).

However, the previous centuries actually dragged our planet through pretty much the same swings in temperature as the 20th century - and even the whole ensemble of all proxies shows that nothing has changed about the variations during the last 100 years. However, most of the trees have been eliminated by the flawed Mann methodology and only a subset - one that "proves" a pre-determined conclusion that the 20th century was special - was kept.

Hockey sticks were and are essentially for the AGW orthodoxy

The other point is that the hockey stick has been essential for the logic of man-made global warming fears. For example, the picture at the top shows Sir John Houghton of the IPCC with a hockey stick behind him. It was taken from a 2001 BBC article,
Human effect on climate 'beyond doubt'
It was thought to be beyond doubt because the hockey stick graph showed that the 20th century warming was faster than what the Earth experienced during the previous 1,000 or 2,000 years. The graph also prominently appeared on the cover of the 2001 IPCC report because it was always the "main" piece of empirical evidence that the humans are changing the climate. However, the graph has been shown to be wrong.

If you remove the hockey stick graphs, there is really no evidence left that would suggest that the natural temperature swings before the industrial civilization were any different than what we experience these days. The industrial civilization probably does add some warming trend but this trend is small enough so that it can't be isolated and empirically proven by any available scientific technique today. If the CO2 emissions increased 20-fold, their effect could become measurable - but this question is a rhetorical one because the CO2 emissions will never increase by a factor of 20.

The hockey stick graph was the only hint that the human influence was kind of unprecedented. But the graph is known to be wrong. If you are one of those deniers who deny that the AGW propaganda has been deconstructed, try to tell us about another paper aside from MBH98, MBH99, and their mutations that uses some measured data to "prove" that the recent human influence is unprecedented.

More up-to-date reconstructions such as "alarmist" Moberg et al. (up) and "skeptic" Loehle (down) don't look like hockey sticks and they don't show any statistically significant "unprecedented changes" in the industrial era.

You won't find any because the statement is not true. So let me urge Mr Schneider and dozens of other surviving demagogues on their aisle of the barricade:
Stop the distasteful obfuscation and tear down the wall!

1 comment:

  1. Correct me if I'm wrong, but understand that the tree ring "thermometers" are weighted and calibrated to average global temperatures, not to the historical temperatures of the region where the trees have been growing. This would seem to be a glaringly obvious flaw. A tree may have grown in a region where temperatures declined during the 20th century, but their ring widths would be heavily weighted if they correlated with global temperatures. The only way that they would correlate with global temperatures would be if their ring width had nothing to do with temperature.