Bishop Hill Wolff on the Hockey Stick
May 29, 2011 Climate: HSI Climate: MWP I thought I'd set down my thoughts on one aspect of Eric Wolff's email to Neil Craig - namely the millennial temperature reconstructions. Eric has adopted what I called the "NAS defence" when I discussed it in the Hockey Stick Illusion. This is the idea that, whatever the failings of the Hockey Stick itself, a series of other temperature reconstructions have reached broadly the same conclusions - Mann may have used in appropriate data and a biased methodology but he still reached the correct conclusions.
Suffice it to say that I find this highly unsatisfactory.
In the spirit of Eric's "what we agree on" approach, I would have thought that we should be able to agree:
- that bristlecone pines and their ilk should not be used for temperature reconstructions
- that Mannian "short-centred" principal components analysis is biased.
These were the conclusions of the NAS panel (and Wegman too), so it is hard to see that there can be much argument on this score.
Can we agree that this is not a thermometer?If we can agree this, then I would hope that we can also agree that temperature reconstructions that feature either bristlecones or short-centred PCA cannot be cited as support for the idea that current temperatures are unprecedented.
I realise that this is not Eric's area and that he is probably therefore repeating the received wisdom - the IPCC line if you like - but it seems very clear to me that the IPCC line is insupportable on this count. I would have hoped that this area was somewhere where a measure of agreement could be reached.
Reader Comments (58)
I see there are two threads partly initiated by my replies to Neil Craig. Since this thread seems to be the one that has more comments, I will post a very brief one. I don't intend to become a regular blogger.
History: Neil Craig sent me (via normal email) his list of questions. At first I assumed (from the tone) that they were rhetorical. When he made it clear he really did hope for a response, I answered him in an email. I had no intention of posting the response, although when he asked me if he could post it, I agreed. I therefore hope that answers those who say I should not pontificate unless I have read a particular tract. I am also very reluctant to go beyond discussing the science – I am sorry if this appears evasive, but my views on energy policy should not be of any more interest than those of anyone else.
Expertise: Again, I am not an expert in all areas of climate science (and no, not on tree rings): for that reason, I am much more conservative about those areas than I am about, say, ice cores. I suspect that some of those making very definite statements on this site are also not experts on tree rings.
There seem to be two things I said to Mr Craig that really worked people up: my comment about the significance of trends over a short time period, and my comment that the basic shape of the hockey stick has been confirmed by subsequent studies.
On the former, there is no reason why the statement by Phil Jones would have come to my attention, as it says nothing unexpected. It really is simple: on the instrumental data from the last century (whether Met Office, NOAA or NASA) there is multiannual variability of order 0.2 degrees, clearly visible in the early part of the record as well as the recent part. There is also an underlying trend of order 0.1-0.2 degrees per decade. In the light of this, the trend within a particular decade is never going to be significant. Just to convince myself, I tried the following. I took the global mean annual temperatures from 1880-1920 from the NOAA website – I used this as my “noise”. I then made a trend of 0.2 degrees per decade (ie a straight line from zero at year zero to 0.8 degrees at year 40). I added the two columns together and plotted my “noisy trend”. So, I know there really is a trend here because I have imposed it, and indeed it is obvious over the 40 year period: and yet there are two 15 year periods in which I see no apparent trend because the noise has counteracted the trend. Try it yourself! This is why the statement you quote at me causes me no surprise, and says nothing about the reality or otherwise of a warming trend at multidecadal scales.
On the latter issue, I can see that the ad hominem issues behind the hockey stick are really important to people on this blog. Several different reports and investigations have taken place: I understand that you don’t like them. My interest is to know how reliable the product is, and this is why I look at the numerous attempts at reconstruction (including those of Esper, Moberg, Mann, Jones, etc) shown in Fig 6.10 of IPCC AR4, WG1 report. Nothing in my understanding of the science relies on a particular reconstruction.
Eric
Thanks for taking part once again. I think it's fair to say that commenters here are broadly supportive of the conclusions of the NAS and Wegman panels - that Mann's implementation of PCA was incorrect and that bristlecones are unsuitable for temperature reconstructions.
What pains us is the argument that the other reconstructions shown in Fig 10.6 mean that Mann reached the correct conclusions regardless. Almost all of the reconstructions in the figure that touch on the question of the MWP use bristlecones and one uses Mannian PCA.
So my question would be, do you dispute the NAS panel conclusions on the use of bristlecones?
Dr Wolff,
I can see you are very suave with your words but you are wrong on your read of the hockey stick situation.
I am not interested in the gossip behind the hockey stick, nor do I "not like" the various investigations and their reports. On the other hand, I *know* what the hockey stick debate is about, and I know the story behind how the paleoclimate 'assessment' performed by the 'IPCC' (which people such as yourself read in order to make one's mind about the hockey stick), came about. Unfortunately both of these things matter however unpleasant they might be to our reductionist minds.
[1] The original hockey stick made a strong rhetorical case for a number of things: unprecedented-ness in the amount of rise, high CO2 sensitivity, alarming rate of temperature rise etc. This is now shown to be an artifact of dodgy methodology. The case made by the original Mann stick correspondingly weakens
[2] The IPCC, for this reason, gave up on showing a supposed multitude of graphs. But these graphs do not all, tell the same story about the temperature course as the original one did. And one can only make the hockey stick shape to appear on this graph (Fig 6.10), by dishonestly graphically overlaying two different kinds of data (a paleoproxy with an instrumental proxy), and by chopping off the decline in proxy temperatures. In other words, the sphagetti graph (Fig 6.10) makes a different rhetorical implication: Esper and Moberg tell a different story compared to Mann.
[3] Thirdly the hockeystick-ness of other reconstructions, wherever they do pop up, clearly are artifacts of the inclusion of certain, very specific datasets as I am sure you'd be aware by now. These were present in the original Mann hockeystick as well. While it may be addicting to go back to artifactual data in order to support one's original scientific position by writing more papers, the resulting picture that emerges due to a reliance on such data can hardly be termed as one of a 'single finding confirmed by multiple studies' (!). Its just the same thing over and over again.
So, can we ask: what is your understanding of paleo-temperatures that 'relies on no particular reconstruction' then? If possible, I think a more explicit statement would be useful here. How is it possible to have an understanding that is line with the original hockey stick (if that is your take), when all the new hockey sticks have the same issues, and the others do not a hockeystick shape?
In other words I believe that it is not possible to have 'one' understanding, by not relying on any particular graph from the current mix.
Lastly it doesn't matter, in one sense, whether a lot of Bishop Hill commenters make categorical statements about tree rings, but (unfortunately) the same doesn't apply to you. As you might have observed, people don't allow you (a climate scientist) the freedom to make a few generalized observations, without freezedrying and shrinkwrapping your opinions to put them up for stuffed display. This practice has to be condemned wherever it appears.
Correction:
Read above para [2] as: "[2] The IPCC, for this reason, gave up and showed a supposed multitude of graphs.
Professor Wolff
If there is an underlying miltidecadal trend of 0.1-0.2 upwards since presumably at least 1800 that is one thing & something which does not suggest we will see historically unprecedented temperatures at least in this century and indeed fits previuos historic trends such as the rise to the Medieval Warming & would suggest a subsequent fall likely.
However Mann's Hockey Stick claims that there has been an unprecedented acceleration since 1979 and to a lesser extent since the 1920s and the entire warming alarm appears to me to be dependent on it, at least the IPCC thought so at the time.
Both cannot be correct.
Whence that underlying temperature trend? We are going to just take it as read, without an explanation?
If the example given is only intended to show how a cycle on a trend may look flat for a while, well, that's fine. If it is meant to say that THIS trend is a given before we begin mention where the cycle comes from, I am not sure we are actually going to the root of the problem. What we need to be able to do is to explain all the variability. Regionally and globally. I don't think we even have the observations to do that. So what actual qualified scientist (which I am not) is going to tell me we even have a basis for saying what is going on now, never mind the future? You can't do it without taking massive leaps of logic and faith.
Please read what I said and did carefully:
I was not in any way assuming a trend, and certainly not assuming one as far back as 1800. I did my example calculation with a hypothetical trend over a mythical year 0- year 40, and real "noise" (from 1880-1920), precisely to show that even with an underlying trend as high as 0.2 deg/decade, the multiannual variability in the real record gives 10-15 year periods with no visible trend: you have to look over a longer period. I really do suggest doing it yourself: it is highly instructive.
Andrew: especially after the other comments on here, I am not going to be led down the road of commenting on something I have no expertise in (bristlecones). I am happy to agree that our knowledge of climate before the intrumental period is still uncertain, and that future efforts (addition of new data, critiques of methodology, new reconstructions, all of them peer-reviewed so we can be sure they pass a first test of plausibility) are likely to improve things.
Well, professor, explanation understood on the trend/cycle illustration. The fifteen-year 'flat' claim never meant much to me anyway. My scepticism is based on what we think we know, and what we can reasonably make of it. It seems to me that the so-called consensus view starts from the properties of CO2 and proceeds far too quickly to the conclusion of ongoing warming without necessarily passing through all the required intermediate stages. Thereafter the consensus supporters seem (to me, subjectively) to be making things fit around their hypothesis. Things which fit are accepted, things which do not are to be dismissed. We all do it in our daily lives, but it isn't really the scientific method. The fact is, we don't know enough to reach the warming conclusion, or if we if accept some warming as a likelihood, just how much, and how does it compare with cyclic and other variations. I say other, because I don't know whether the millenial trends are cycles or derive from events, randomly.
Of which the operative term is 'we don't know'. Until we know, or get some handle on it at least, our best action is to find out what is going on. It does not help if most of our research is directed so as to confirm the consensus view. It should be to find an explanation for our observations, That is not what sceptical me sees happening. What I see happening is that things which fit are accepted, and things which do not are explained away by ever-more-stretched stories. What I detect is like the epicycles used to explain planetary motion. Problem cooling in the 70s? Oh, it was probably aerosols. So that's OK then. Frost fairs on the Thames? Never 'appened, Squire. Roman warm period? What did they know, they didn't have thermometers. Out of an ice age in a decade or two?...well, I don't know how they explain that.