Talk:Temperature record of the last 2,000 years/Archive 1
![]() | This is an archive of past discussions. Do not edit the contents of this page. If you wish to start a new discussion or revive an old one, please do so on the current talk page. |
Archive 1 | Archive 2 | Archive 3 |
Page name
(William M. Connolley 19:34, 17 Oct 2003 (UTC)) This page contains useful info, some of it added by me, but there's no way I'm going to add links to anything called "hockey stick graph", so I've renamed it to a more neutral title. Those who like the old one can continue to use the redirect, and that seems fair to me, esp since nothing links to the hsg.
ps: I got the move wrong at first and linked to ...1000years. (with a full stop). Hopefully this is now corrected.
- I agree that Hockey Stick can continue as a REDIRECT. It's a term used chiefly (only?) by 'skeptics' (as you call them) - not 'enviros' (as your colleague Fred Singer calls them). BTW, can you think of any better terms for GW advocates than enviros & skeptics? --Uncle Ed 15:50, 21 Oct 2003 (UTC)
What is up with that graph? there are no units, and there is not even a scale along the bottom. If you add a graph, find one which has units and put a summary below for context. -- Catskul 2004 Feb 13
(William M. Connolley 17:02, 15 Feb 2004 (UTC)) Replaced graph with the ready-to-slot in replacemetn, why couldn't you bother? See http://en.wikipedia.org/wiki/Image%3AGlobwarmNH.png and http://en.wikipedia.org/wiki/Image%3AMBH98_SEW_drawing311.png
Misc
(William M. Connolley 20:50, 20 Oct 2003 (UTC))
- Changed "about 1 oF (0.4 to 0.8 oC)" to "about 0.6 oC". There is no point giving 2 temperature units everywhere, and oC is better (err - there must be some kind of wiki policy on this, I wonder what it might be?). 0.6 not range, not to attempt to suppress uncertainty, but because thats covered elsewhere and is not the point of this page.
- replace "used by the IPCC" with "these"; add "quantitative" to first para. The point being, its what these graphs show, whether used by IPCC or not.
- Add para re qual/quant distinction, which inevitably makes the quant records sound superior. Partly because they are; partly because they are more amenable to analysis.
- Wasn't the Maunder Min a period of low sunspot activity, first, and a cold period, second?
- I don't think the sunspot record goes back into the MWP but leave this for now.
Hottest year
From text:
- The work of Mann et al. and others [1] forms a major part of the IPCC's conclusion that atmospheric temperatures had been on a slow, gradual downward trend until the 20th century when greenhouse gas emissions caused temperatures to increase at an unprecedented rate; and that the present warming is unusual within the last 1000 years. It also shows that 1998 was the hottest year in the record.
Saying that Mann's "work...shows that 1998 was the hottest year" implies that Mann's work is reliable and honest, since it shows a "fact". But Wikipedia ought not to endorse Mann's claims like this. We should not use authoritative language which implies that these are objective "findings" of impartial searcher of truth.
Rather, we should say that the data Mann selected and arranged bolster his contention that temps were flat until recently, and that his graph presents 1998 as the hottest year -- rather than saying that his work proves that 1998 was the hottest year.
Another paper, which I'll cite in a moment, thoroughly discredits his work as little better than a fabrication -- okay, maybe he made an honest mistake, but it's at best shoddy work. He didn't take enough data points, and he simply ERASED the medievel warm period because it didn't fit in with his pre-ordained conclusion.
This division of quantitative (proving Mann and IPCC right) vs. qualitative (uninformed speculations of skeptics) is just UN/enviro POV. --Uncle Ed 21:17, 24 Nov 2003 (UTC)
(William M. Connolley 22:44, 24 Nov 2003 (UTC)) You still don't inderstand qual/quant. Keep trying.
- <grin> Well, doc, it would help if you explained the distinction better. How about writing a short article? --Uncle Ed 22:59, 24 Nov 2003 (UTC)
Recent papers
We should probably add the two recent papers attacking the conclusions by Mann et al.
This one by Soon et. al. (There are actually two papers that are substantially the same)...
http://www.kolumbus.fi/boris.winterhalter/EnEpreprintFeb03.pdf
They had it published in "Energy & Environment", which is notable see this excerpt from 'Chronicle of Higher Education'
http://www.davidappell.com/archives/00000293.htm
another discuss at Eureka alert
http://www.eurekalert.org/pub_releases/2003-07/agu-lcs070703.php
and the actual response by Mann et. al.
http://www.agu.org/journals/eo/eo0327/2003EO270003.pdf#anchor
The second paper of interest is
http://www.multi-science.co.uk/mcintyre_02.pdf
again pulblished in 'Energy & Environment'
by McIntyre and McKitrick
and here is the rebuttal by Mann et. al
http://holocene.evsc.virginia.edu/Mann/EandEPaperProblem.pdf
- (William M. Connolley 01:17, 9 Dec 2003 (UTC)) Yes. But some of that is already somewhere... oh yes, on Michael_Mann_(scientist) (Ed Poor started it...). I would actually prefer it to stay there, perhaps augmented: at the moment, the value of M&M is very much in debate, and is more of a "challenge to Mann" (and hence approprite to his page) than a contribution to the T record (my POV). The S+B paper(s) are junk... I'd rather not pollute wikipedia with them. They are of no value to the T record pages... perhaps under global warming controversy? Somewhere.
I realize they are junk, but my concern is, that skeptics will have heard of the papers (due to media saturation), but have doubtfully heard the refutations. Thus if we fail to mention the papers and fail to explain the problems with them, I suspect either readers will doubt the NPOV, or will assume we are ignorant of the papers.
- (William M. Connolley 13:59, 13 Dec 2003 (UTC)) In which case probably the best thing to do is to start a page about just that paper (poss a bit too specialised), or a page about Soon or Baliumas (do they have one already?), or dump it into the hell-hole that is global warming controversy, or... Well, if you want to write up the S+B papers, good luck to you, whereever you put it.
Re-balancing
(William M. Connolley 21:19, 5 Feb 2004 (UTC)) I've de-skepticised the page somewhat, and added a rebuttal from Mann to M&M (forgetting what was on this very talk page about using the Mann page...
Firstly, I've re-instated *all* quant recons. I'll repeat what I said to Ed: you want to get rid of all, you have to find one that says otherwise (and no, M&M doesn't count, can you guess why?).
I cut the "belies activist claims" stuff: what was that there for?
John Dalys page doesn't describe skepticism, it pushes Dalys POV; hence rewording.
Re M&M, I've collapsed it somewhat, in particular making the de-listing the list, because it doesn't (yet) deserve the expanded space. But all the items in the list are still there.
FUD from Soon et al.
(William M. Connolley 19:35, 4 Jul 2004 (UTC)) OK, a POV header perhaps...
- Fear is irrelevant to science. Uncertainty and doubt are hardly surprising when there are many unknowns. FUD is a marketing term, so it depends on what is being sold. -- (SEWilco)
Para removed from page:
- There are difficulties also in presentation of the recent instrumental record, which is often shown near the end of temperature reconstructions.
- Soon, Legates, and Balunias pointed out in 2004 that they were able to replicate the instrumental surface temperature trends as shown in several studies.
- However, they were unable to reproduce the long-term Northern Hemispheric surface temperature trends in Figure 2.21 of IPCC TAR and Figure 2a of (Mann and Jones [2003]). It should be possible to duplicate the results of such a study.
- During one year, three related studies show an increasing temperature change, with the change increasing at the extremely rapid rate of about 1 to 2.5 °C per decade. The authors of those three studies were showing that during the year that their studies had produced increasingly larger rates of change. Each study showed that temperature was increasing ever more quickly. Soon et al. found no justification for the high value in the last paper (Mann and Jones [2003]).
- Soon, Legates, and Balunias pointed out in 2004 that they were able to replicate the instrumental surface temperature trends as shown in several studies.
- Estimation and representation of long-term (>40year) trends of Northern-Hemisphere-gridded surface temperature: A note of caution. Willie W.-H. Soon, David R. Legates, Sallie L. Baliunas Geophysical Research Letters February 14, 2004 http://cfa-www.harvard.edu/~wsoon/myownPapers-d/SLB-GRL04-NHtempTrend.pdf
- Willie Soon's summary of above paper and comments on the scientific method: http://www.techcentralstation.com/062404I.html
The Soon study is a bit weird. They repeatedly talk about claims of 1oC/y or whatever - a claim made by nobody but them.
- (SEWilco 04:05, 6 Jul 2004 (UTC)) They say the claim is in the figures.
As to their problems in 2b - look at it. THe green line looks like a better fit to the data than the red.
- (SEWilco 04:05, 6 Jul 2004 (UTC)) It looks like a better fit, but it won't be if next year is much warmer or colder. Just as the green line would be too high or low on a graph ending in 1931 or 1940 (without using data in the future of that date), due to the obvious change in trend in those years.
The paper appears to be FUD. The *calculation* of the trends isn't done from reading lines off graphs so its all quite irrelevant.
- (SEWilco 04:05, 6 Jul 2004 (UTC)) Soon's calculations also were based on real data, but could not duplicate the trends shown without finding problems. As you say, there are trends shown in the figures.
- (SEWilco 14:30, 5 Jul 2004 (UTC)) - WMC knows most of this, but I show more detail here for other readers.
- Looks to me like they got the trends from the slope of the lines on the graph, and they are looking at the rate of change shown for very recent years. They started with formal instrumental data and tried to duplicate the published graphs. But using the last year's data as the average for the next 20 years is not a reliable way to get the pretty line to the end of the paper. As the article states, we don't have information about future events so the method which involves 20 years in the past and 20 years in the future produces a graph which should end 20 years before the end of the data. Pushing the line to the last year involves increasing uncertainty.
- Yes, I see in 2b (page 3 of Soon's paper) that the green line looks nice. But if the graph is cut off at 1941, the green line would go upward and away from what actually happened. If the graph is cut off at 1930, the green line would go downward from what actually happened.
- (SEWilco 14:39, 5 Jul 2004 (UTC)) Reinserted in the article the paragraph about Soon's study.
- (SEWilco 04:05, 6 Jul 2004 (UTC)) Rereinserted another version of Soon paragraph.
(William M. Connolley 08:37, 6 Jul 2004 (UTC)) This is silly. Soon et al come to the grand conclusion that:
- concluding that there are problems in recent suggestions of an extremely rapid rate of about 1 to 2.5 °C per decade
BUT THIS SUGGESTION IS ONLY MADE BY SOON ET AL. Not by anybody else. They are rebutting their own nonsense! FUD removed again.
- (SEWilco 16:07, 6 Jul 2004 (UTC)) The suggestion is from the published figures.
- (William M. Connolley 16:30, 6 Jul 2004 (UTC)) Thats why its FUD, not an outright lie. Soon et al are trying to give the impression that people have been saying that the recent T trend is 1-2.5 oC/y. This is nonsense, only Soon et al are saying this so they can create a strawman to knock down.
- (SEWilco 08:02, 7 Jul 2004 (UTC)) It's not the T trend which they refer to. It's the rate of change of the T trend during the year 2002-2003. The illustrated T trend increased by as much as 0.25°C/y during a single year. 0.25 * 10 years = 2.5
- (William M. Connolley 16:30, 6 Jul 2004 (UTC)) Thats why its FUD, not an outright lie. Soon et al are trying to give the impression that people have been saying that the recent T trend is 1-2.5 oC/y. This is nonsense, only Soon et al are saying this so they can create a strawman to knock down.
- (William M. Connolley 16:06, 7 Jul 2004 (UTC)) I don't know where you get this weird interpretation from. I don't find it in the paper. And the tech central article talks about "the extreme warming trend of 1 to 2.5 oc/decade...". Thats a trend, not a rate of change of trend. Mann et al (despite Soons insinuations) don't say this. Nor does theor figure c support it either. Their figure c appears to show 0.5-0.6 oC in about 2 decades. Furthermore, the assertion (lower down) that T trend estimates are quite arbitrary is total cr*p. They have failed to distinuish (perhaps to even understand) the different between lines on a graph and fitting trend estimates. I do hope that you understand this.
- (SEWilco 17:01, 7 Jul 2004 (UTC)) That's why I mentioned 0.25, so you can search the PDF: look at the end of paragraph [3] on page 1 of Soon, paragraph 13 on page 3, and 19 on page 4.
- (SEW) The figures are a statement of trend, as you recognized when you commented one trend line looking like a better fit.
- (William M. Connolley 16:30, 6 Jul 2004 (UTC)) No, the statements of trend are the numerical numbers quoted in IPCC and elsewhere. But Soon et al can't touch those, so they have to create strawmen instead.
- (SEWilco 08:02, 7 Jul 2004 (UTC)) The numbers and methods don't all match the illustrations. Although not all the numbers and methods are available, or are not stated correctly. Are there really enough numbers to show what the illustrations show? Often studies use illustrations as summaries, and extracting data from them is done often enough that there is software to do that.
- (WMC) As to the lines and the fit... yes Soons lines don't seem to fit very well.
- (SEW) There should be no problem replicating the studies, so the lines should fit. You know how science works.
- (WMC) As to the lines and the fit... yes Soons lines don't seem to fit very well.
- (SEW) The studies present a pattern, rather than just the raw data, but in some cases the pattern is not supported. Soon's point is that the calculation of the fit is unexplained or seems to be based on artificial future temperatures
- (WMC) This is a (minor) signal processing matter. People drawing average lines through noisy data like to get to the end. Which you can't do by simple averaging.
- (SEW) These people seem to not agree how to reach the end. Even when one of them is working on all three drawings. As Soon says, they should better explain their methods.
- (SEW) , and more information should be given with such figures. Thus I am pointing out that there are problems with the representation of temperature records. Review and replication of studies is part of the scientific method, including failed replication (ie "cold fusion").
Are any of the above issues unresolved? Which? (SEWilco 07:48, 17 Jul 2004 (UTC))
(William M. Connolley 21:31, 24 Jul 2004 (UTC)) Back from hols... anyway: my central point is that Soon et al is FUD (I only say that to explain my overall view of its: I'm not pretending its a complete criticism in itself). Anyway, the unresolved points are: is the 1-2.5 oC stuff referring to T trends or to increases in T trends; where anyway do S *get* this value from; and why its a valid criticism when they are the only people saying it. There also seems to be some confusion that S are encouraging about the *computation* of T trends as opposed to their *display*: the end-padding S talk about is irrelevant for the computation.
- (William M. Connolley 08:56, 27 Jul 2004 (UTC)) I've added comments below, but this is going round in circles.
- The 1-2.5°C is increase in T trends. That is, the published trends are showing increasingly higher trends for the same period. They also say that they found no justification for one high value, so they obviously got the value from the chart because they can't duplicate that result. The Tech Central Station article shows that they couldn't get an explanation of how to duplicate it, other than some "circular explanation".
- (William M. Connolley 08:56, 27 Jul 2004 (UTC)) The only people talking about 1-2.5 oC stuff are Soon et al. No-one else. They are just making these claims to knock them down: strawmen.
- (SEWilco 15:59, 27 Jul 2004 (UTC))
- Do you see that those 3 graphs do not end the same way?
- If you don't see the graphs are different then let's examine them more closely.
- If you see the graphs are different then you're now one of those who has observed there is a difference in the published trends.
- Do you see that those 3 graphs do not end the same way?
- (SEWilco 15:59, 27 Jul 2004 (UTC))
(William M. Connolley 19:38, 28 Jul 2004 (UTC)) This is where you have failed to understand the difference between drawing pretty lines on graphs, and the published temperature trends. When IPCC says the trends over the last century are 0.4-0.8 oC (or whatever) they are *not* reading the lines off graphs. They are going it by LS fit or somesuch. Which don't suffer from any of this nonsense. Do *you* understand that.
- (SEWilco 05:08, 29 Jul 2004 (UTC)) Yes, if Dr X reports a range like that then it is from the results of his calculations; he might also show results on a graph but he is producing the graph from his numbers. However, if he only publishes the graph and not the numerical results then all he's offering is for us to read the lines off the graph. The graph can be studied, just as similar graphs from analog instruments can be studied.
- Measure the differences in the endpoints of the graphs.
- I see trends of 0.3, 0.41, and 0.55 °C per decade. Do you see similar values?
- Measure the differences in the endpoints of the graphs.
(William M. Connolley 19:38, 28 Jul 2004 (UTC)) Don't do this!
- (SEWilco 05:08, 29 Jul 2004 (UTC)) What do you want us to do? If the numerical results are available, we can use those. Perhaps Soon et al. used numerical result data sets, although the references only mention the published studies. Aren't published studies what you say should be used as authoritative sources?
- (William M. Connolley 09:21, 29 Jul 2004 (UTC)) What do I want you to do? I want you to calculate trends by LS fits to data, not by reading wiggly lines off graphs by eye.
- (SEWilco 15:36, 29 Jul 2004 (UTC)) You indicated a lack of understanding where the 0.1 and 0.25 came from, so I presented a back of the envelope method. If you don't want to eyeball the graphs, dump their numerical representation of their results in your spreadsheet and compare the numbers. Are you getting values which are significantly different?
- (William M. Connolley 15:55, 29 Jul 2004 (UTC)) You have misunderstood me. You said, you were getting trend values by eyeballing wiggly lines. I said, this is a poor method of deriving trends, amongst other reasons because it suffers from the end effects that so excise Soon et al. If you want to derive trends, do what all sane people do and derive them from fits to the original data, which doesn't suffer from these problems.
- (SEWilco 05:02, 1 Aug 2004 (UTC)) So have you found yet that the three graphs show final trend values which differ? I don't care whether you want to use the red pixel on the image or numerical data. I'm just asking if there is a difference, the amount of difference is the next item.
- The differences are about 0.1 and 0.25 °C per decade. Do the math yourself because reproducibility of results is important.
- The three studies appeared in one year.
- I don't see mention of when the studies were authored, so the differences in publication date seems to be used instead of the actual times.
- The change in trends of 0.1 and 0.25 within one year are projected as changes of 1 and 2.5 per decade. The decade timescale matches that used in the studies.
- (William M. Connolley 15:55, 29 Jul 2004 (UTC)) You have misunderstood me. You said, you were getting trend values by eyeballing wiggly lines. I said, this is a poor method of deriving trends, amongst other reasons because it suffers from the end effects that so excise Soon et al. If you want to derive trends, do what all sane people do and derive them from fits to the original data, which doesn't suffer from these problems.
- (SEWilco 15:36, 29 Jul 2004 (UTC)) You indicated a lack of understanding where the 0.1 and 0.25 came from, so I presented a back of the envelope method. If you don't want to eyeball the graphs, dump their numerical representation of their results in your spreadsheet and compare the numbers. Are you getting values which are significantly different?
- (William M. Connolley 09:21, 29 Jul 2004 (UTC)) What do I want you to do? I want you to calculate trends by LS fits to data, not by reading wiggly lines off graphs by eye.
(William M. Connolley 19:38, 28 Jul 2004 (UTC)) Indeed. This is where the total twaddle comes in. Thanks for explaining it so clearly - I hadn't realised that Soon et al were really dumb enough to do this.
- 0.1 change in a year * 10 years in a decade = 1 / decade
- 0.25 change in a year * 10 years in a decade = 2.5 / decade
- If your calculations also show 1-2.5 then now you are also making the claim. It's no longer only Soon et al.
- Quantity of participants does not determine a straw man, it is the ingredients. In which of the above steps is straw growing?
- The number of people saying something does not affect whether it is true or not -- are the charts for the same period the same or not?
- The computational issues are an attempt to figure out what causes such differences, but the differences are there whether explained or not. This study shows there are difficulties in dealing even with the very newest temp records. -- (SEWilco 06:05, 25 Jul 2004 (UTC))
- (William M. Connolley 08:56, 27 Jul 2004 (UTC)) Let me take this very very slowly as I'm clearly not getting through. (1) If you want to draw pretty lines on a graph of smoothed data, then you need to do something to "pad it to the end" if you want the smoothed lines to go to the end. There is no unique way to do this. (2) If you want to calculate the trend in a series then point one is TOTALLY IRRELEVANT because the trend can be and is calculated without any need for padding. I don't understand what you mean by "there are difficulties in dealing even with the very newest temp records".
- (SEWilco 15:59, 27 Jul 2004 (UTC)) (0) The published trends are different.
(William M. Connolley 19:38, 28 Jul 2004 (UTC)) No. The wiglly lines are different. The published trends aren't.
- (1) The same method of smoothing should be used, so the results from all three similar studies should be the same. Instead 3 studies show increasing trends, so differences were examined, and one could not be duplicated. (2) The padding was examined in an attempt to duplicate the published results, as the published studies showed the trend was changing although the data was not.
(William M. Connolley 19:38, 28 Jul 2004 (UTC)) But this is where the total nonsense occurs. Different smoothing leads to different wiggly lines. They might go up, they might go down. Only Soon et al are dumb enough to create a fake time series from these different wiggly lines. Wikipedia should have nothing to do with this.
- (SEWilco 15:36, 29 Jul 2004 (UTC)) Soon et al had to use most of their paper to show that "different smoothing leads to different wiggly lines".
- (William M. Connolley 10:17, 1 Aug 2004 (UTC)) No. Different smoothing leading to different lines is obvious.
- They had to do this because others produced a fake result of their wriggly lines. If you're going to call Soon's study fake, then that term should apply to the result which Soon found "unjustified". Should Wikipedia have nothing to do with Mann and Jones (2003)?
- (William M. Connolley 10:17, 1 Aug 2004 (UTC)) Sigh. You're being silly here. Soon et al have produced a fake timeseries, if your description above is accurate. You say, they have produced a timeseries from differences in other peoples wiggly lines. But those differences in wiggly lines are not well fixed in time (as you say above) and aren't really a time series anyway: they could just has easily appeared in another order had publication dates been different.
FUD from Kyoto supporters
Wilco and William, please stop sparring a moment and help me write about why the IPCC orginally used a temperature graph that showed the Medieval Climate Optimum but then changed to using a temperature graph which conceals that warm period. --Uncle Ed 19:37, 1 Oct 2004 (UTC)
- (William M. Connolley 20:27, 1 Oct 2004 (UTC)) Welcome back to climate, Ed. But you've been away a while: take a time to familiarise yourself with whats already on the pages before you rush in: in particular the section "Skepticism and rebuttals thereof" subsection "Historical temperature estimates".
Bias
If anyone censors the report of M&M's findings about bugs in Mann's hockey stick program, then I will put the "neutrality disputed" notice back up. "I'm a lover, not a fighter." --user:Ed Poor (deep or sour) 17:32, Nov 11, 2004 (UTC)
- (William M. Connolley 19:07, 11 Nov 2004 (UTC)) M&M's findings are unpublished. See-also http://www.wmconnolley.org.uk/sci/mbh/ if you want some detail. I've just had a look at your additions, which confirm my worst fears. You cannot possibly get away with dumping a pile of undigested skepticism, not even from a published paper, into the intro of the article. Try putting it somewhere sensible, or better still here, and we can talk about it.
Well, you've certainly given me something to chew on, doc. But I don't swallow it. Just what exactly is the flavor of your discontent? Wikipedia is not a refereed journal. I can put in any skeptical remark I like, as long as the source has SOME credentials related to the issue. --user:Ed Poor (deep or sour) 21:18, Nov 11, 2004 (UTC)
- (William M. Connolley 21:30, 11 Nov 2004 (UTC)) I'll repeat myself (and []) then: You cannot possibly get away with dumping a pile of undigested skepticism, [which is] not even from a published paper, into the intro of the article. Try putting it somewhere sensible, or better still here, and we can talk about it.. You have a bad history of doing this: adding material that (although dubious) might survive lower down in a sensible section but which cannot possibly justify being in the intro.
Are you throwing down the gauntlet now? Let's do this properly, and have your friends arrange the matter with my friends. Pistols at dawn? Icicles at noon? How about a spaghetti-eating contest? Cool off, man. Go outside and breathe some cool British air. --user:Ed Poor (deep or sour) 22:21, Nov 11, 2004 (UTC)
- (William M. Connolley 23:04, 11 Nov 2004 (UTC)) This isn't war. There is space for the most recent McK stuff (properly qualified). But it doesn't belong in the intro.
This paragraph is meant to confirm that the debate is going on, and actually our side - that insists that these graphs should not be presented as facts - constitutes a majority in this debate. William, it would be nice if you took this observation into account, and if you either allowed the critical remarks about the MBH graphs to appear in the introduction, or accepted the POV label. Concerning newer material along these lines, please don't forget von Storch's recent article in Science. The MBH graphs are very far from an established piece of science. --Lumidek 22:15, 14 Nov 2004 (UTC)
- (William M. Connolley 22:21, 14 Nov 2004 (UTC)) Putting the M&M stuff into the intro is absurd - its not even published material. And even if it was, putting the entirety of the quote into the intro would still be absurdly unbalanced. Von S is another matter - feel free to introduce von S into the text.
- OK, I would be happier if you introduced von S yourself - unless the truth about the record has been showed to you by God, my particle physicist's understanding of the word "science" dictates that the observations of von S are important for this story. Moreover, if the fluctuations by von S are confirmed, I will really think that M&M should take more credit for it, but that's a different issue. At this moment, I find the article plausible - at the edge - so I won't reintroduce POV. --Lumidek 22:55, 14 Nov 2004 (UTC)
- (William M. Connolley 23:09, 14 Nov 2004 (UTC)) Von S and M&M are orthogonal. This is an encyclopedia, not a newspaper - news requires time to settle down. Von S hasn't. Its in Science, true, but very recently. How would you have reacted if I'd put Vinnikov and Grody down as definitive rebuttal of S+C's MSU stuff?
- Von S and M&M are not orthogonal, and both are about the underestimated fluctuations in the past by Mann et al.
(William M. Connolley 10:02, 15 Nov 2004 (UTC)) But there is no connection between their approaches. Neither supports the other.
- As you can see, there has not been enough time for Mann et al. to settle down either, and I think that they will not settle down anymore. ;-) I am surprised by your focus on the appearance of various texts in journals. This is certainly not a strategy that I would expect from someone who understands the field although it may be a good strategy for a laymen to choose the right information. However, if a physicist told me that he decides whether a paper is correct or not according to the journal where it appeared (i.e. she or he trusts the referees), I would conclude that the physicist probably does not understand the material herself. Do you actually understand something about climate science, or are you just a science fan that trusts those experts who are able to publish in well-known journals? Although you are very active and visible at all these pages, I have not understood this point so far. To be sure, I want to answer all your questions, so here's the last one. If you decided to put Vinnikov and Grody in this article, I would erase it because it has nothing to do with temperature in the last 1000 years - rather last 25 years as measured by satellites. I am open-minded whether the surface measurements agree with the satellites, and I agree that Vinnikov and Grody are too new for us to be sure about the outcome, especially if we're string theorists like me. ;-) Yes, as you probably guess correctly, I lean to the opinion that the surface measurements are biased by human activity, and the satellites can show much less change, but it's certainly not a dogma for me. --Lumidek 23:27, 14 Nov 2004 (UTC)
(William M. Connolley 10:02, 15 Nov 2004 (UTC)) Ah, excellent: you've managed to see the point: V+G is too recent. But its older than von S.
- Thank you for your compliments. Things can be recent, but they will not be recent in the future. I think that you know very well that in a couple of years, when the dust settles, Mann et al. most likely won't be viewed as a good piece of work. You just try to slow this process down.--Lumidek 12:40, 15 Nov 2004 (UTC)
MM05
Redirected discussion from Talk:Global_warming
I think that there is going to be a storm of controversity when MM05 is published in February.
The Financial Post:
Breaking the Hockey Stick - Part 1
http://www.canada.com/components/printstory/printstory4.aspx?id=108c0400-4e71-4c55-a279-dd43aed1f224
The lone Gaspe cedar - Part II
http://www.canada.com/components/printstory/printstory4.aspx?id=052554eb-ebdb-483a-8f28-c4ce19458973
--D Norris 15:43:55, 29 Jan 2005 (UTC)
- (William M. Connolley 16:30, 29 Jan 2005 (UTC)) Thats nice. Lets wait till its actually published though shall we? And lets remember that the MBH results are replicated by independent studies too, shall we? Since we're all recommneding our favourite reading, I recomment to you http://www.realclimate.org/index.php?p=114, http://www.realclimate.org/index.php?p=111 and indeed many of the other posts at RealClimate.
- For those that want a sneak peak at what is being published in Geophysical Research Letters next month: http://www.climate2003.com/pdfs/2004GL012750.pdf GRL, by the way, published Mann in 99 as well. --D. Norris 17:25, Jan 29, 2005 (UTC)
Hmmmmm ----
In January, 2005, an adapted version of McIntyre and McKitrick's paper was accepted for publication by Geophysical Research Letters (GRL). Judging by the reactions of the referees of GRL, which McIntyre made available to us, the tide may be turning in the climatology field. One referee stated: "S. McIntyre and R. McKitrick have written a remarkable paper on a subject of great importance. What makes the paper significant is that they show that one of the most important and widely known results of climate analysis, the 'hockey stick' diagram of Mann et al.,was based on a mistake in the application of a mathematical technique known as principal component analysis (PCA)."
The same referee also writes: "McIntyre and McKitrick found a non-standard normalization procedure in the Mann et al. analysis. Their paper describes this procedure; it was an apparently innocent one of normalization, but it had a major effect on their results. The Mann et al. normalization tends to significantly increase the variance of data sets that have the hockey-stick shape. In the Mann et al. data set, this turned out to be bristlecone pines in the western United States. Thus the hockey stick plot, rather than representing a true global average of climate for the past thousand years, at best represented the behavior of climate in the western U.S. during that period. This is an astonishing result. I have looked carefully at the McIntyre and McKitrick analysis, and I am convinced that their work is correct."
The referee ends with: "I urge you not to shy away from this paper because of its potential controversy. The whole field of global warming is currently suffering from the fact that it has become politicized. Science really depends for its success on an open dialogue, with critics on both sides being heard. McIntyre and McKitrick present a cogent analysis of the global warming data. They do not conclude that global warming is not a problem; they don't even conclude that the medieval warm period really was there. All they do is correct the analysis of prior workers, in a way that must ultimately help us in our understanding of past climate, and predictions of future climate. That makes this a very important paper. I strongly urge you to publish it." - The lone Gaspe cedar [2]
--D. Norris 19:01, Jan 29, 2005 (UTC)
- I discuss the article MM05 on my blog [3] which also contains the links to the paper itself, and articles about it. Maybe we should try to be nice to these people. They will definitely be in an existential crisis. Starting from February, they may have no future. They will be identified as the leaders of the most costly scientific fraud in the history of the humankind. I think that M&M will become pretty famous. Many people experienced in statistics will quickly follow, and they will most likely identify the problems with all the other similar climate papers. These papers are not really independent. They share the same authors and the same methods, so my guess is that they will collapse easily - the main "argument" behind them was a "consensus" - a consensus among the people neither of which bothered to verify the key calculations. A consensus among very weak scientists. In fact, Mann et al. was the most transparent paper where the algorithm they used to obtain the results could be partially followed. Other papers are much worse. If someone predicted that Mann will commit suicide before March 2005, it would be hard for me to argue against it. It is certainly much much more likely than a climatic disaster in the next 50 years. ;-) This guy is kind of doomed, and William's situation is not much better. William, please don't give up your life - and enjoy the last weeks of your life in which the people are not spitting at you on the street. ;-) There is always a chance for you to start a new life - a life without scientific fraud. All the best, Luboš --Lumidek 20:44, 29 Jan 2005 (UTC)
- (William M. Connolley 21:12, 29 Jan 2005 (UTC)) I suggest you read the RealClimate posts.
- Dear William, be sure that I've read them - although an outsider, I am simply interested in this topic. Having read them does not imply that they make as much sense as the new paper by M&M which is pretty brilliant and powerful. Your article on rc.org are pretty lousy, they have really nothing to do with the technical findings of M&M, and you must really believe that all other people are complete idiots if you think that they won't be able to identify whether MM or rc.org make more sense. Enjoy the last weeks of your consensus of lies! Saddam also tried to enjoy the last weeks. ;-) --Lumidek 21:35, 29 Jan 2005 (UTC)
More from The Financial Post: http://www.canada.com/components/printstory/printstory4.aspx?id=db3b26e9-edb9-417c-ab92-fa78432433bf
For instance, in 2004, scientists in the journal Geophysical Research Letters concluded that Mann and Jones' methods were "just bad science" and that they had undertaken a "selective and inappropriate presentation" of results.
In June, of 2004, the accumulated weight of criticisms led to a retraction by Mann (and Scott Rutherford) in the Journal of Geophysical Research. However, while Mann admits to substantially underestimating historical temperature variations he claimed that it had no effect on his conclusions.
If even one component of Mann and Jones' -- and, consequently, the IPCC's -- temperature reconstruction is in error, then we can't say with any confidence that the 1990s were the warmest decade of the last two millennia, or that 1998 was the warmest year, much less that the last century's rise in temperature is unprecedented.
How interesting! --D. Norris 13:33, Feb 1, 2005 (UTC)
Current
Dear Gentlemen, I hope that many of you will agree that the label ((current)) is gonna be pretty appropriate for this page. ;-) --Lumidek 03:17, 31 Jan 2005 (UTC)
- (William M. Connolley 09:54, 31 Jan 2005 (UTC)) Who knows, it might be one day, but it certainly isn't now.
- Dear William, that's a great strategy to postpone all these things, hide, combine, and recombine your defense. But remember that February only has 28 days. ;-) For others - this "current" was partly a joke, and if it's difficult to keep it, don't waste much time with it. But if they publish MM05 and something starts to happen, the tag may be appropriate. --Lumidek 14:54, 31 Jan 2005 (UTC)
- (William M. Connolley 15:17, 31 Jan 2005 (UTC)) These pages are controversial enough. Wasting time starting joke edit wars is inexcusable.
- I have not started any edit war. As someone who cares about your doomed future, I'm just slowly preparing you for what is gonna happen very soon. ;-) BTW have you heard the sad news plus good news from Virginia that finally they stopped him and I must still wait for my 10,000 dollars? ;-) --Lumidek 15:29, 31 Jan 2005 (UTC)
- (William M. Connolley 16:39, 31 Jan 2005 (UTC)) Hopefully you haven't, but Norris seems to be blindly following you, before realising you were only joking. As for Virginia... what are you talking about?
- I never blindly follow anyone. If I did, I would blindly accept the pronouncements of the IPCC and march in lock step with the 'scientific consensus', but I was trained to question everything, especially the opinions of the herd.
- I happen to think that this topic is going to start changing very rapidly in the next month and warrents a 'current' disclaimer. But I am willing to wait for MM05 to be published.
- Oh, BTW, it is 'Ms Norris' to you, Doctor! But if you are polite, you can call me Denise. --D. Norris 13:40, Feb 1, 2005 (UTC)
- You made a mistake by re-inserting Lumideks joke but are too graceless to admit it. Noted the Ms.
- I agree with Lumidek 100% on the use of 'Current'. Please allow me to refresh your memory as to his exact words:
- For others - this "current" was partly a joke, and if it's difficult to keep it, don't waste much time with it. But if they publish MM05 and something starts to happen, the tag may be appropriate.
- Hmmmm... I admit everything with aplomb and grace. --D. Norris 19:50, Feb 1, 2005 (UTC)
- Perhaps all of you want to read why the "current label" was invented: Template talk:Current. IMHO the number of edits is by far not on the same level as 11 March 2004 Madrid attacks and thus does not justify the current tag. -- mkrohn 00:08, 2 Feb 2005 (UTC)
The introduction
This article needs an introduction. The name "Temperature record of the past 1000 years" implies objective fact - as if there were an actual, documented record of temperatures over the past millenium. However, this article is simply a collection of theories. This needs to be stated clearly and unequivocally in the introduction.--JonGwynne 01:11, 8 Feb 2005 (UTC)
- I tend to agree that the old introduction is not the best one, but I also think that the current one is not better. We don't need to explain what a theory is, better link to theory instead. Also I wonder about the usage of the word "assumption" in your text "the assumption by scientists of what the temperature record may have been". May I ask what your background in science is? As I said changing the current intro is fine with me, but the one your propose sounds strange to me - sorry -- mkrohn 01:50, 8 Feb 2005 (UTC)
- I agree that we don't need to explain what theories are, but we do need to explain that this page is theory and not fact. The title makes it sound like fact and not theory. If you have a better idea for an intro that explains this, then feel free to add it or change mine to suit. But, I'll take another crack at it for now since there has to be one.--JonGwynne 01:55, 8 Feb 2005 (UTC)
Moberg et al.
I am not going to try to make any changes to this or related articles myself, because frankly this battleground is no fun to play in, but I would like to point out the paper by Moberg et al. appearing in today's Nature (Feb. 10, 2005). The link is [4] for those with online access to Nature.
The paper is an attempt to redo the temperature history of the last 1000 years based on what the authors regard as a growing recognition that the methodology of Mann et al. and similar studies systematically surpress centennial scale variability. Their results, using a revised methodology, show both a medieval warm period and a little ice age. They are also consistent with the scale of natural variability reported from borehole measurements, while Mann et al. wasn't. However, they still regard the warming during the last several decades as unprecedented and inconsistent with natural variability.
Dragons flight 18:55, Feb 10, 2005 (UTC)
- For those who have no access to nature: the graph is similar to Image:GlobwarmNH.png with the two exceptions that "Dragons flight" already pointed out: the graph which shows years 0-2000 touches 0 degree about 1000 AD and goes down to about -0.7 degree around 1600 AD. It seems (though I am not an expert) that the graph is a refinement of Mann et al. but tells us qualitatively the same thing. Quote from the paper: "We find no evidence for any earlier periods in the last two millennia with warmer conditions than the post-1990 [...]" -- mkrohn 19:44, 10 Feb 2005 (UTC)
- (William M. Connolley 21:05, 10 Feb 2005 (UTC)) I haven't read it, but probably will tomorrow. I presume its good science - Nature usually is - but I would like to caution that wiki is an encyclopedia; a place for well-considered results; and things just-published don't usually belong (arguably the von S stuff falls into this category, but I know the howls of protest that would occur if I tried to remove it :-).
Hockey Stick disinformation
This article needs attention. The discredited theories of "hockey stick" created by the "hockey team" should be rewritten as a historical discussion of the myths that were believed at various points, and the more correct and up-to-date sources like Moberg et al. and M&M should replace the "mainstream" discussion. Best, Lubos --Lumidek 01:51, 11 Feb 2005 (UTC)
- (William M. Connolley 09:51, 11 Feb 2005 (UTC)) Thank you for making your biases so obvious.
- You may call it "biases", but the more important thing is that it is reality. This article has been kind of completely flawed, and it is necessary to re-make the whole text. --Lumidek 15:01, 11 Feb 2005 (UTC)
I don't feel as strongly as Lumidek apparently does, but I do think that the presentation of this article needs some attention. I am inclined to think that this article would be better served if it moved away from the hockey-stick / not-a-hockey-stick mentality. MBH was a first effort in this field and there have been a variety of improvements and criticisms since then. It would probably be better to focus on what are the range of temperature reconstructions that people are considering today. In particular, MBH shouldn't be the only temperature plot on this page. I realize that there are links to other reconstructions on this page and associated discussion, but by only showing the MBH plot, it gives the implied impression that this is the only important view. I have in mind something like a nicer version of figure 1 from this page [5]. Dragons flight 21:43, Feb 11, 2005 (UTC)
- (William M. Connolley 21:56, 11 Feb 2005 (UTC)) I'm quite happy to de-emphasise the "hockey stick", and thoroughly agree with your "mentality" comment above. I wanted a pic showing all (or most) of the recons but couldn't find one; is the pic you point to usable in wiki? It would be nice if it was.
Moberg et al
(William M. Connolley 21:58, 11 Feb 2005 (UTC)) And if anyone wants my take on Moberg, which I've just read, try: http://mustelid.blogspot.com/2005/02/moberg-et-al-highly-variable-northern.html.
personal attacks
Lumidek, the discussion is already heated enough and it is completely unneccessary to make the situation worse by personal attacking others. Stopping this would be helpful, thanks. -- mkrohn 13:56, 11 Feb 2005 (UTC) (see Wikipedia:Wikiquette and in particular Assume good faith).
- I don't think that the discussion is as intense (and heated) as what I would find appropriate. What the "hockey team" has done is very serious. By the way: This particular article is mostly flawed, it will need a serious reconstruction. I am assuming good faith of Wikipedians, but I am not forced to assume good faith of criminals and the people who don't follow the rules of scientific integrity. --Lumidek 15:04, 11 Feb 2005 (UTC)
- Lumidek, are you aware of the fact that M&M have been wrong (in painfully obvious ways, e.g. the radians/degrees debacle) before? And funny enough, whenever one mistake is pointed out, they use a different method to come to the same conclusion... I would not put much trust into their claimed results before the paper has been discussed for some time. Certainly not enough to accuse Mann et al of being "criminals". --Stephan Schulz 03:01, 13 Feb 2005 (UTC)
- You are aware that Mann et al. seem to be indulging in the same method to defend their hockey stick? Whenever one error is pointed out they show that the hockey stick can be salvaged if you do something slightly different. And I believe that MM have been fairly consistent with their argument about the data-mining nature of the MBH algorithm (red noise in --> hockey stick out) and demonstrating that MBH's hockey stick is the result of a programming error. At any rate, past errors are not relevant - what is relevant is whether the current work has errors in it. -- John Simon (not registered, just watching from a dispassionate distance)
- I am a scientist (and hence read and write a lot of scientific papers), but not a climate scientist (hence much of the specific math is beyond me). However, from what I have read, the older errors in MBH are mostly minor. The McKitrick et al errors, on the other hand, are real whoppers (using the vector norm as an average for temperature (thermodynamic nonsense), using non-absolute temperatures for that (mathematical nonsense), confusing radians and degrees, mistaking correlation and causation in a crass manner, and so on). I used to give them the benefit of the doubt, but after I looked over some of the older McKintrick papers (and commentary) yesterday, I lost all faith in whatever they do. --Stephan Schulz 22:22, 13 Feb 2005 (UTC)
- You are aware that Mann et al. seem to be indulging in the same method to defend their hockey stick? Whenever one error is pointed out they show that the hockey stick can be salvaged if you do something slightly different. And I believe that MM have been fairly consistent with their argument about the data-mining nature of the MBH algorithm (red noise in --> hockey stick out) and demonstrating that MBH's hockey stick is the result of a programming error. At any rate, past errors are not relevant - what is relevant is whether the current work has errors in it. -- John Simon (not registered, just watching from a dispassionate distance)
- Lumidek, are you aware of the fact that M&M have been wrong (in painfully obvious ways, e.g. the radians/degrees debacle) before? And funny enough, whenever one mistake is pointed out, they use a different method to come to the same conclusion... I would not put much trust into their claimed results before the paper has been discussed for some time. Certainly not enough to accuse Mann et al of being "criminals". --Stephan Schulz 03:01, 13 Feb 2005 (UTC)
- (William M. Connolley 21:18, 13 Feb 2005 (UTC)) MBH haven't accepted any of M&M's points as affecting their results, but some of McK's errors have been too obvious for him to fail to accept. And you're wrong about the red-noise stuff: that wasn't in their earlier paper(s?). "programming error" is wrong BTW - if its an error - MBH don't accept it is - its methodological, not a bug. (Disclaimer: I'm part of http://www.realclimate.org).
- Noted. I am also scientifically trained (wouldn't want to claim being a scientist as I am but an economist who happened to do a lot of physics, maths and statistics at grad and undergrad levels) but have been struck by the lack of direct responses to criticism by many participants in the debate. For the most part my comment was reflecting on the passionate (as opposed to dispassionate) nature of much of this malestrom and the playing of the man rather than the ball. I can see how past errors can inform ones view about probabilities, but each idea should be assessed on its own merits. Repeated comments about radians/degrees (which seems to be the preferred soundbite in this area) seem akin to a schoolkid loosing an argument (please note the use of the word akin above before assuming that this is my take on the current imbroglio) and then saying ..."Yeah! Well you can't tie your shoes!" - JS 05:29, 14 Feb 2005 (UTC)
- In my experience, getting a Journal paper published takes anything from 9 months (in the most extraordinary cases) to 2.5 years from the day of submisson, with an average of maybe 18 month. Adding in the time for reading and understanding the criticism, and drafting a reply, do not expect anything published before late 2006 - that is, if Mann et all even feel the necessity and find the time to respond formally. In the meantime, various online resources discuss the various McKittrick papers, incuding some comments by Mann (and co-authors). Try [6].
- One additional remark: I have ideas plenty. Other people have many more. If I asses each idea equally, I will be swamped just checking on nonsense. I take the presentation, the quality of the source (estimated by looking at previous contributions), and even my own qualification in dealing with it into account when deciding where to spend my time and energy. McKitrick does not do well on either of these scales. --Stephan Schulz 14:38, 14 Feb 2005 (UTC)
- (William M. Connolley 10:19, 14 Feb 2005 (UTC)) Errr... you're complaining about a lack of direct response (on the discussions here, or do you mean between the various paper authors?) yet I have directly responded to your points (MBH don't accept M&M's crit; M&M haven't maintained the same crit; "programming error" is wrong) and you have slid away from them. I don't see how schoolkid analogy helps here - if you want to know whats going on, you'll need to look at the details.
- In response to your comments I said 'noted' to indicate that you had responded to those characterisations I had made and, although I did not say it, that I would investigate them further while reserving judgement. The comment about lack of direct response was not directed at your response but at my impressions of the general debate (both here and in the various blogs and websites that specialise in this stuff) and would be more appropriately directed to the authors and, more so, cheerleaders. The schoolkid analogy was designed to elucidate my view of a debate where frequent references to past errors are made along with other questionable rhetoric. My view of the debate has been that it is commonly on the level of schoolboy debate - hence the analogy. Don't get your knickers in a twist - if I wanted to have a go at you I'd call you an acerebreal cephalopod. Until then assume that I am a reader of the Midnight Star. -JS 12:31, 14 Feb 2005 (UTC)
New figure for article
I have created and uploaded the following image.
Details are included on the image description page. I will let other people decide how to use it in this article. Dragons flight 01:59, Feb 13, 2005 (UTC)
- Wow, this looks great, thankt! Would it be doable to make two more plots with 4 resp. 5 functions in each image? I think this would improve readability, as having all 9 functions in one image is very confusing. -- mkrohn 02:19, 13 Feb 2005 (UTC)
- Can you make the code public so it can be studied and reproduced? Or, so Marko can make readability adjustments. Perhaps on the Image page have explanations and details, so links can reach such details. (SEWilco 21:06, 14 Feb 2005 (UTC))
- The image page already makes it pretty clear what I did. I may release the code (Matlab format) after I have a chance to add comments and clean it up a little. Dragons flight 21:23, Feb 14, 2005 (UTC)
Translation of jargon
Having read MBH 98 I have found that the statistical jargon used in that paper is very different from the statistical jargon to which I am used. Nonetheless, as part of my education, I am trying to translate their methodology into jargon I can understand. This section is, thus, a request for education but also intended as the precursor to a potential technical appendix to the article that might explain the methods used by MBH to those from a different statistical background.
This may be an inapprorpiate place to have this discussion - if so please redirect me (and this section) to the appropriate place, otherwise, answer away.
To illustrate some of my problems they commonly use the term "reconstructive skill" when referring to what I would more commonly call "goodness of fit" statistics e.g. r2. They refer to singular value decompositions (SVD) in forming estimates and discuss each step of the matrix algebra involved, but I would commonly refer merely to the OLS estimate (β=(X'X)-1(X'Y)) rather than bothering with details of the matrix algebra and computation methods that are used to arrive at that solution. That is, SVD is a mathematical technique that can be used to arrive at the OLS estimate (these days it is relegated to a particularly low level of any statistical programing language because matrix inverse functions are readily available along with even higher level functions which directly generate the OLS estimate).
That given, my current understanding of the technique used in MBH 98 is:
1. Instrumental temperature recordings from around the globe are analysed using PCA to generate a set of principal temerature components which they identify with various global temperature trends, e.g. the Northern Hemisphere average, the El Nino deviation and a North Atlantic deviation.
2. They regress each proxy they have individually against these PCs to obtain a Least Squares estimate of the way each individual proxy responds to global temperature fluctuations.
The basis of their method from here on begins to confuse me somewhat...
3. They collect the coefficients from each proxy regression (which is a mapping from temperature PC to proxy) as a collection of new variables and regress the temperature PCs against these coefficients. This yields a way to go from fitted proxy to temperature PC. I find this step confusing because it seems to go back in the opposite direction to step 2 and I am not clear about its purpose.
4. Regardless, step 3 then provides a way to estimate temperature PCs from the proxy data and thus reconstruct temperature during earlier periods when instrumental data is unavailable. (But because I am fuzzy about step 3 this is currently an article of faith for me rather than an expression of full comprehension).
I have some questions that I would appreciate any explanation for as a way of clarifying and expanding on the steps enumerated above:
1. Is this just a complicated way of describing instrumental variables regressions (otherwise known as two-stage least squares)? What labels should then be applied to the various data series in this context? Are the proxies the instruments?
2. Why don't they just regress the proxy series over the 'training' period on the PC temperature series? They have 112 proxies (although there seems to be some discussion about the exact figure) which can be used as explanatory variables for the first (and other) temperature PC which is about 150 (?) annual observations, to obtain a way of predicting the first temperature PC over the full extent of their proxy data.
3. How are non-temperature factors in the proxies dealt with? (Is that the reason for step 3?) For example, there is a non-stationary CO2 factor that undoubtedly affects the tree ring proxies. One will end up wth a spurious or, alternatively, meaningless regression if one uses I(1) RHS variables to explain an I(0) or I(1) LHS variable (unless appropriate correction is made for the integration e.g. first differencing among other techniques)
- JS 10:10, 15 Feb 2005 (UTC)
- (William M. Connolley 22:11, 16 Feb 2005 (UTC)) I've never read the full details, sadly. I think you've missed something: firstly that not all the proxies go all the way back, so there is a "stepwise" technique that uses all that are available in different timeframes; secondly (I think) that PCA/SVD is used to reduce large numbers of proxies down to a few EOFs, when the proxies are dense in certain regions.
- I know I have missed something - hence the request for clarification and contribution from those who know more. At the moment my questions about the method are at the first order level. The stepwise technique is, to my mind, a second order element of the technique that I have not turned my mind to yet - undoubtedly important but not fully ripe for consideration until the foundation is in place. I am confident about their description of step 1 and what it means. They use PCAs to generate the global temperature trend(s) (a small number of series which they identify with, presumably, known temperature patterns) from numerous regional instrumental data series (if they use proxies in this step then that would be news to me and it is not clear from their description). They also use PCAs to reduce some of their proxy sets to lower dimension by extracting the principal trends before using them in steps 2 and 3.
- - JS 22:27, 16 Feb 2005 (UTC)
- JS,
- When I last studied MBH 98, I came to the conclusion that there simply weren't enough precise details to be able to reproduce what they did. You can certainly get the gist of what they are doing, and in fact that may have been enough to convince Nature and the reviewers that their approach was reasonable. However, it has always struck me as scientifically bad form that no where in the paper or the supplmental materials did they provide a detailed elaboration of the methodology and justification for each step.
- Personally, I think this oversight has a lot to do with the McKitrick & McIntyre debacle. Because MBH weren't more detailed and precise, M&M had to make a variety of assumptions about what MBH had done in their reconstruction. As a result, M&M's "audit" of MBH actually ended up using different methodology than MBH in several key places. Naturally, Mann replies that M&M didn't do things "correctly". M&M replies "but you never said anywhere that you were doing THAT", and a big brouhaha ensues. In general, I believe this has left M&M's audit irrevocably flawed, but at the same time I have considerable sympathy for M&M. They were trying to reconstruct MBH's analysis from published accounts that seem to have been fundementally inadequate for accomplishing that job. MBH deserve to be soundly criticized not providing adequate documentation, but the results they reported do seem to have been reasonable given what was understood at the time and what has subsequently been revealed about their actual methodology. (Note that reasonable is not neccesarily that same thing as being correct. One really has to evaluate their work in the context of subsequent and related work before judging what is correct. But I do not believe that MBH was intending in anyway to be malicious, as some commentators have implied. I should probably also say that I don't have a lot of sympathy for M&M when it comes to the way they have continued to play up their "audit" of MBH and the associated methodological disagreements.)
- Anyway, if you really want to understand what is going on in all of this climate reconstruction business, I would suggest you look at a variety of other papers in the field. [7] has some useful references in this regard. I haven't studied most of them, but I would certainly hope that many of them provide more detailed descriptions of the methodology. IF you really want to dig more deeply into what MBH in particular was doing, you probably ought to read M&M and the ensuing methodological arguments with Mann. An archive of those arguments are provided here [8] and a variety of other discussions appear at RealCimate. Sifting through those arguments can be deeply unpleasant, but as far as I am aware, some of the technical details specific to MBH have only ever been reported as a result of those arguments.
- So far, I haven't really answered your questions though. In order to really give you a thorough answer of what MBH are doing, I would have to study all the materials again, and I don't really have that kind of time. However, in outline form, what MBH did was take instrumental records from many different spatial locations and compare those to proxy data at "nearby" spatial locations. The purpose of the temperature PCs was to first determine what fraction of the variability in local temperature changes could be related to global vs. local climate change. The proxies in a given regions were then used to make PCs for that region which were (more or less) compared to local temperature changes, and (at least in theory) an overall understanding of the contributions of local and global temperature change to those proxies was extracted. Then by weighting the proxies based on how strongly they recorded global temperature changes one could (hopefully) combine them in a way which was an adequate reflection of long-term temperature change, with some error bars based both on the statistics associated with the number of available records and the fraction of variation that was likely to be local. [My apologies if any of this description is factually inaccurate. I am working from memory right now.]
- To my understanding, no one has ever attempted to specifically correct for CO2 variations as a contributor to changes in tree development. However, there are good reasons for believing that most trees growing outside the tropics are water-limited in their growth not CO2 limited.
- Dragons flight 15:45, Feb 17, 2005 (UTC)
- Much obliged. I have made an edit to the method description in the article based on the broadest understanding of their method (and, indeed, my understanding of the method used by this class of studies) which is, I hope, uncontroversial. (Controversy seems to surround the validity of any of their methodological steps.) We'll see how long it takes me to digest the references you have provided (or whether I get a severe case of indigestion and need to take an emetic).
- JS 20:41, 17 Feb 2005 (UTC)
- (William M. Connolley 21:08, 17 Feb 2005 (UTC)) A couple of comments... firstly, thanks for your edit on the page. Second, if you just want to understand one of the reconstructions, and aren't too concerned as to which, then I think the recent Moberg one is fairly simple and fairly well documented. Since I wrote http://www.realclimate.org/index.php?p=122 I certainly think I understand it. Thirdly, you may find some more recent Mann et al papers helpful in understanding MBH98 (or more recent Mann work) (though you may not like the post titles...): http://www.realclimate.org/index.php?p=98 and http://www.realclimate.org/index.php?p=10 (the last provides a link to the Rutherford et al preprint).
- The following is a quote from the Rutherford preprint "After roughly 1960, the trends in the MXD data deviate from those of the co-located instrumental gridbox SAT data for reasons that are not yet understood (Briffa et al., 1998b; 2003; Vaganov et al., 1999). To circumvent this complication, we use only the pre-1960 instrumental record for calibration/cross-validation of this dataset in the CFR experiments." I am concerned that their procedure doesn't so much circumvent the problem as brush it under the carpet. I will provide my thinking below and I ask for comments. At the moment this is just a small detail that jumped out at me and I make no claims that it means anything about anything - I am just asking for explanation of the methodological justification from someone who knows more.
- My thinking: They have observed that for some of their sample the MXD tree ring density data correlate with local temperature measurements, but that after 1960 they don't. Thus, there is a factor other than temperature that is affecting the tree ring densities (or the relationship is highly non-linear or both). It strikes me as more logical to assume that this factor affects the data throughout their sample rather than just after 1960 (which they implicitly do by dropping the data). A consequence of this is that they may suffer from an ommitted variables bias in their pre-1960 calculations - at worst they could have estimated a merely coincidental relationship. Dropping data from a period when their hypothesis does not fit does not circumvent the problem, as suggested above, it just brushes it under the carpet. For those who do not follow this I'll offer an imperfect analogy (as all analogies are): I observe on Monday through Friday that there are many cars on the road in the CBD during daylight hours. I hypothesise that the sunlight causes cars. However, on Saturday and Sunday I notice that there are fewer cars on the road in the CBD and about as many as there are at night. It would be improper scientific practice to discard the Saturday and Sunday observations merely because they do not fit my hypothesis. I should, instead, replace or refine my hypothesis that sunlight causes cars in the CBD. Comments?
- JS 00:58, 18 Feb 2005 (UTC)
- Tree ring width followed temperature from 1880-1960 relatively convincingly and then they diverge dramatically. Briffa et al. 2003 would agree with you that some other variable in the environment must also be important and varying after 1960. They suggest (as do earlier Briffa papers) that the confouding factor is changes in UV flux associated with ozone depletion at high latitudes. To support this they offer some limited evidence that the divergence is most pronounced near the Arctic circle. Because we really only have a UV and ozone measurements for a short time period, it is hard to know if this is the correct interpretation. However, it would give a logical reason for assumming that this confounding factor is only a recent problem, given that ozone depleting air pollution is only a recent problem. So I agree, it is a issue that gets swept under the rug, but they at least offer a plausible reason for doing so.
- Dragons flight 01:56, Feb 18, 2005 (UTC)
(William M. Connolley 21:18, 19 Feb 2005 (UTC)) You may find http://www.realclimate.org/index.php?p=121 useful for tracking what is done. And it would probably be a good place to ask questions if needed. Don't be offended by the title :-)
Thanks for the link. I'd already had a read of that post and digested some of it. At the moment my questions are predominantly methodological rather than data-related and the current controversy, as set out in the article, is mostly framed as a data issue (although I suspect that these can overlap significantly - faulty methodology can allow faulty data to distort the results). I have a bit of investigation still to do and may not be in a position to post anything substantive there or ask sufficiently informed questions for a while. The article comments may be closed by the time I work it all out (if hell hasn't frozen over first - or would it be more apropos to say Europe given the current topic?). Nonetheless, I am currently contemplating the possibility that there is an omitted variables bias in these reconstructions coupled with the possibility that, because temperature and CO2 have both trended strongly over the past century, there may be a problem of spurious regressions (see the top of p.363 in particular). It strikes me that such musings cannot be dealt with adequately in a blog and would, if even only partially well founded, ultimately require a correpsondence with the authors - such correspondence couldn't really be commenced until I knew exactly what I was talking about. Regardless, with any luck, Wikipedia should have a nice tight article on it all at the end of it.
JS 01:54, 20 Feb 2005 (UTC)
Links
Let's leave "blog" links (e.g. realclimate.org) and links that require subscription fees to access them (e.g. the nature article) out of wikipedia, shall we?--JonGwynne 01:45, 17 Feb 2005 (UTC)
- I (mostly) disagree. realclimate.org features reasonably well-written and mostly stable articles. It is not a classic blog where one guy dumps his random musings daily until he gets bored. I do not see how the ability of users to discuss articles on realclimate.org devalues it as a source. In fact, given than Mann et al are writing on the web site makes it even a primary source for the discussion. Why don't you want it in? The link to the Nature article is indeed not very useful for most people. I suggest to move it to to the links section and qualifying it with (requires subscription). --Stephan Schulz 01:57, 17 Feb 2005 (UTC)
- You mean: well-written... in your opinion. I'm not saying you aren't entitled to your opinion or that I necessarily disagree with you, but Wikipedia isn't an editorial page, it is an encyclopedia. A blog is a blog - it is a place for people to indulge in chit-chat and is not, therefore, a suitable source for authoritative statements in an encyclopedia. The problem, as I see it, with citing even statements made by famous names in a blog is twofold. First, we have no way of knowing if the person in question really wrote the statement (there is no concept of security or validation in these blogs) - in other words, someone claiming to be Mann could write something and post it. Also, Michael Mann isn't a terribly uncommon name, so someone else with that name could be writing. Second, we have no way of knowing what weight to attribute to the statement. Did the person in question write it seriously after much consideration and then hand it off to others for review and revision before finally publishing it? Or did he come home after several pints at the local pub with his mates and dash off a drunken screed? Or was it somewhere in between? We have no way of knowing.--JonGwynne 13:37, 17 Feb 2005 (UTC)
- By that reason (a blog is a blog), a journal is a journal, and hence Nature is the same thing as the National Enquirer, and we can get rid of all the unreliable scientific stuff in one go. The meaning of blog has shifted very fast. Nowadays, anything with articles which the readers can discuss is called a blog, including things like Slashdot that were around for a long time before the term was even coined. Thus, we have to look at a (so-called) blog as at any other source. Doing so in this case, I feel certain that the Michael Mann on realclimate is the same Associate Professor primarily responsible for the hockey stick reconstruction. Looking at the articles, I find that they are written with reasonable care, and are useful as a source for the ongoing discussion. Yes, this is my opinion, but so is everything else in live beyond pure mathematics. I don't believe in Pink Unicornism, and thus I allow my opinion to influence my actions. --Stephan Schulz 23:49, 17 Feb 2005 (UTC)
- To the best of my knowledge, articles printed in the National Enquirer aren't subject to peer-review as are submissions to Nature. So, comparing the two isn't really a useful exercise. Your personal certainty of the pedigree of certain blog articles isn't really relevant here, is it? You might be right, an article might be written by the Michael E. Mann, but then again, it may not and we'd have no way of knowing, would we? You are right that there is a lot of opinion and assumption in life... let's not add any more than we really have to. What do you think?--JonGwynne 23:58, 18 Feb 2005 (UTC)
- (William M. Connolley 11:39, 17 Feb 2005 (UTC)) The Nature link is very useful to those who have a subscription. As for blogs, its probably best to use them with caution and judge them on their content. AFAIK, the RC Moberg post is the only decent source on the web for a discussion of the Moberg paper (other than mine, which RC obsoletes). If the Nature commentary was freely available, or Moberg had a page explaining his work, I would use that rather than RC.
- To WMC: If you want to include the Nature link, then put it down in the "External Links" section with a note that says "for Nature subscribers only". It isn't really fair to non-subscribers to use it is a primary reference. Don't you agree? (Ooops, I see that Stephan had almost exactly the same suggestion. I was reading from the bottom-up so I didn't see his statements until after I had written mine. See? Stephan can be reasonable and discuss things. I suggest you take a page from his book. ) --JonGwynne 13:37, 17 Feb 2005 (UTC)
- (William M. Connolley 13:47, 17 Feb 2005 (UTC)) The Nature link *is* the primary reference - how could it not be? It takes up very little space, so there is no obvious resaon to remove it. If you follow it, but don't have a subscription, you get enough info to look it up (volume, page etc) in a paper version in your local library.
- It doesn't matter how much space it takes up. Subscription-only links have no place here. Stephan agrees with me - and that should tell you something. If you want to add it to the "External Links" section with an appropriate disclaimer, that's one thing. But if you want to mention the article in the main section, then find a freely-accessible link to it.--JonGwynne 15:24, 17 Feb 2005 (UTC)
- (William M. Connolley 15:31, 17 Feb 2005 (UTC)) If you are implying that there is a *policy* against subs-only links, then please quote/link it. In this case, however, the link itself is useful even if you have no subs, because it tells you the volume # etc - as I said above. I certainly take Stephans opinion seriously, but he is not as settled as you are implying, so I will await his further comments.
- As I said before, I would prefer to have subscription-only links clearly marked - if only for the personal feeling of disappointment I had when I clicked on the link and found no access to the article. Wether it is inline or separate is not that important to me - I can't think of a really elegant solution. That the link should go somewhere is clear - this is the primary reference, after all. --Stephan Schulz 23:33, 17 Feb 2005 (UTC)
- I completely agree with William here. For one there is to my knowledge no policy or rule against external subscription-only links. Wikipedia talk:External_links has some discussion about this, though the discussion was mostly about news related links, where it is easier to get a "free" alternative.
- Not showing subscription-only links would be absurd, as links to books also require buying the book and thus would have to be removed. The rule should be, that we always give the best possible reference. Of course if a free alternative of similar quality exist we should link to that.
- We can discuss about labeling subscription-only links in order to save the reader a click. Unfortunately I see no technical support for adding such labels, thus we should go on with the usual practice. For further discussion about the topic of labeling subscription-only external links Wikipedia talk:External_links probably is making more sense.
- best regards -- mkrohn 17:07, 17 Feb 2005 (UTC)
(William M. Connolley 21:21, 17 Feb 2005 (UTC)) Thanks Marco. I've made a comment at the Wikipedia_talk:External_links#External_links_to_subscription_services page you found, to see if anyone else has opinions.
Reference #15 is a link to a broken file Acrobat cannot display. 64.105.34.196 (talk) 18:04, 13 July 2009 (UTC)
Deceptive name of article
The name of the article is inherely deceptive. It describes a proxy record as though it were an actual record. I tried to correct this, but WMC won't tolerate any view other than his own - regardless of the truth of it. The article remains hopelessly POV until it is accurately named.--JonGwynne 21:51, 20 Feb 2005 (UTC)
- (William M. Connolley 21:57, 20 Feb 2005 (UTC)) Attempting to push your POV by renaming the article is unacceptable, and great waste of everyones time (including yours). You should not rename a page without discussing it first and gaining a consensus to move it. At the moment, I rather doubt you'll get it.
- What POV? It is a fact that the temperature record of the past 1000 years is a reconstruction. It isn't an actual record. Your attempts to cover up this fact are astonishing. Yes or no William: Is the record of the past 1000 years a reconstruction? --JonGwynne 22:00, 20 Feb 2005 (UTC)
- (William M. Connolley 22:02, 20 Feb 2005 (UTC)) Isn't it about time you answered the RFA against you?
- I still haven't decided if it is worthy of my comment. Your other complaints against me have been so completely lacking in merit and supporting fact that it might well be a waste of my time to comment on this new one. I may change my mind though. Now, isn't it time you answered my question? I'll ask again: Yes or no William: Is the record of the past 1000 years a reconstruction?--JonGwynne 22:14, 20 Feb 2005 (UTC)
- (William M. Connolley 22:35, 20 Feb 2005 (UTC)) I think you should go off and answer your RFA and stop treating wikipedia with disrespect. The answer to your Q is: your page move was an attempt to insert your POV into the article name. Next you'll be trying to move CO2 to "CO2 - increase is only 0.01% of total atmos" and then complaining when people say its a silly name.
- You didn't answer my question. Let's try again: Yes or no William: Is the record of the past 1000 years a reconstruction?--JonGwynne 22:39, 20 Feb 2005 (UTC)
- hint: All you need to do to answer the question is write "yes" or "no".--JonGwynne 22:39, 20 Feb 2005 (UTC)
- (William M. Connolley 22:44, 20 Feb 2005 (UTC)) I've answered your question. Now go answer your RFA.
- No, you haven't answered the question. I asked you if the record of the past 1000 years is a reconstruction. You have not answered that question yet. Actually, to be fair, it is a rhetorical question. We both know the answer is "yes". Of course the record of the last 1000 years is a reconstruction because there are no actual records going back that far. So, the real question is why did you revert the name of this article back to its original and deceptive state? Oh yeah, because you're a narrow-minded, zealot who can't tolerate view which oppose your own.--JonGwynne 23:15, 20 Feb 2005 (UTC)
- I too would like to hear a straight answer to a clear question. The Title of the article implies a factual record of the past 1000 years. I was taken in briefly and believed it until I read on a realized only 150 years are a record. The rest is reconstructive estimates. Why not something like "Global Temperature reconstruction: Past 1000 years" or "Global Temperature Estimates: Past 1000 years". People like to click on a link and look at the pictures or graphs and assume they are what they were led to believe they are. Let's be responsible and truthful up front..--Mike(UnregisteredUser) 03 Mar 2009 —Preceding unsigned comment added by 63.68.128.254 (talk) 00:24, 26 March 2009 (UTC)
Changing the name of the article
It was rude of Jon to rename the article without any discussion, but even so, the name and focus of the article is a point that may be worth discussion. The name Jon proposed, "Reconstruction of temperature record for past 1000 years", strikes me as exceedingly cumbersome. Something like "Reconstructed temperature for the past 1000 years" would accomplish much the same purpose while being only slightly longer than the current title. However, since the article is pretty clear that it isn't talking about direct measurements, I am not really convinced that quibbling over the difference between a "record" and a "reconstruction" is all that important. But, just to offer an observation, the current page might nearly as well be called "temperature reconstruction controversy", since that is almost the entirety of its current focus.
However, I would like to offer a more general suggestion for feedback from this community. How about dropping the "1000 years" part and just calling it "Temperature reconstructions" or something similar? There is nothing particularly magical about 1000 years after all, and there are at least 4 annually resolved temperature reconstuctions (of the type currently discussed in the article) that go back farther than 1000 AD. If that was decided to be desirable, it would probably need to be part of a broader rewrite to add information on longer time scales. Some brief comments about changes during the ice ages and over geological time scales could be useful for perspective.
Anyway, just a suggestion, and one that obviously requires some feedback before any further action is taken. Dragons flight 00:52, Feb 21, 2005 (UTC)
- I didn't mean to be rude. The reason I unilaterally changed it is because the problem was so obvious and the solution seemed a simple one.
Do you notice that you imply that everybody else working on the article is either stupid (for not recognizing the obvious problem and the simple solution), lazy (for recognizing it, but not acting), or part of a giant conspiracy? That is rude. Something as major as a wholesale renaming of a page (especially one with a lot of controversial editing going on) should never be done without a discussion.--Stephan Schulz 13:09, 21 Feb 2005 (UTC)
- I was willing to do all the work to edit all the articles that referenced it, but WMC interrupted me halfway through and still hasn't explained why. I take your point about trying to keep the name from becoming cumbersome though. However, there are certain words that have to all be in the title for it to be accurate. In no particular order:
- Temperature
- Record
- Reconstruction
- Further, I think there should be some mention of the time frame or maybe just a generic "Historical" would do. What do you think?
- How about "Historical Temperature Record Reconstruction"? Your idea of using the word "controversy" in the title is interesting, but I doubt William would ever allow it. --JonGwynne 01:03, 21 Feb 2005 (UTC)
- I'm not sure I see the need to use both "record" and "reconstruction", but generally, I'd view "historical temperature record reconstruction" as a very reasonable title. I'm also slightly biased against "historical" because as I said it might be nice to have a few comments about longer term (i.e. pre-historical) climate change for perspective, but unless there is a ground swell of support for including such a perspective, I would be willing to waive that objection. Other people's opinions? Dragons flight 03:47, Feb 21, 2005 (UTC)
- (William M. Connolley 09:42, 21 Feb 2005 (UTC)) At the moment there is a (sensible) separation between temperature records from the satellite/balloon record; the historical temperature record; and the proxy record - the last being the one we're talking about. The most problematic bit of the name is "1000 y" as DF has pointed out. "Reconstruction" isn't such a great word though, "proxy" is better. All temperature records are reconstructions (well all are proxies too, though that doesn't mean much). "Historical" is not good, since historical temperature record exists.
Quantitative vs qualitative
Noticing the recent edit war and vaguely remembering previous edit war(s) I think there is a lot of confusion and aggro created by these two words. When I first read the article I found the use of the words noticeable. There seems to be an implicit sense that quantitative is somehow better than qualitative and this is reinforced with each repetition of these words. I think the impression is subtle but, nonetheless, real. However, I think that their continued use is making a distinction that is not helpful and only creates controversy to no real gain for the article. I have three main points about these two words:
1) Numbers can lie just as well as words can - quantitative is no better than qualitative in terms of objectivity.
2) The 'quantitiative' reconstruction of Moberg et al. doesn't conform to the general slow trend down followed by sharp upturn (I would think a W (sort of) would be a better description of it).
3) The 'qualitative' reconstructions mentioned are actually quite quantitative. Grape harvest times are very quantitiative; records of harbour freezing are also quantitiative; so is the physical extent of farming (e.g. Greenland) which isn't mentioned. If my great-great-great uncle Harry recorded in his diary that "It was cold today", that would be qualitative. If he recorded that winter-crop harvests were low because of unseasonable weather you are starting to stray to quantitative and if he actually recorded crop yields then you are completely quantitative.
I think the actual distinction that is suggested is how comprehensive and representative the data are. But that is a different issue.
Thus, 'I do not think it means what you think it means'.
- JS 02:48, 21 Feb 2005 (UTC)
- I agree. The quantitative / qualitative, especially in the opening bit, seems problematic. As you say, this is especially true in terms of quantitative reconstructions, like Moberg, that seems to be agreeing more with what is referred to in the article as the "qualitative" picture of climate change during the last millenia. Though I still think there is a meaningful distinction to be made between records that tell you whether it is hot or cold in a given place and records that (hopefully) tell you something meaningful about the numerical temperature. Perhaps this belongs in a different spot? I would also note that even quantitative records may only provide qualitative information about temperature if there is no practical way to convert the data to temperature measurements.
- Dragons flight 03:18, Feb 21, 2005 (UTC)
I was hoping for some more responses, particularly from WMC given that he introduced it all way back when. However, absent further responses, I will look to doing some editing shortly. Silence will be taken as agreement.
- JS 03:03, 23 Feb 2005 (UTC)
- Normally, that might be ok. But you do know that Wikipedia was partially down (and still does not do watchlists for me) due to a power failure? I'd wait until things are back to normal. --Stephan Schulz 16:14, 23 Feb 2005 (UTC)
- WILCO - JS 20:27, 23 Feb 2005 (UTC)
Recent Edits
In attempting the minor change to remove the quantitative vs qualitative split discussed in the section above I found that more significant changes were needed to make the article flow properly. I have recast the split as between those showing minimal prior variability and those showing more variability. I have also removed a lot of stuff that gives this issue its heat, but which I judged to be non-central. That is, I have attempted to keep the article tightly focused on the temperature record and removed many comments related to AGW - reducing it to saying that these reconstructions have implications for that debate.
I didn't do much to the second half of the article because I ran out of steam - these sections are still a bit of a hodge-podge. Also, if this recasting of the article sticks, there needs to be a bit of work putting correct links in and fleshing a few things out (e.g. references to those grape harvest papers). Some more cross-referencing to other Wikipedia articles might be in order because I deleted a fair few of those references in trying to keep the article tightly focused.
As a general request, could people not do a wholesale revert but re-edit sections or just revert sections? I think that the previous article needed some work. I've started the process but it is not finished. However, reasonable minds can disagree on this.
- JS 22:28, 26 Feb 2005 (UTC)
- (William M. Connolley 22:47, 26 Feb 2005 (UTC)) I really wish you hadn't done that. You've lost a major distinction that the previous version had, which is the distinction between the history-type stuff (LIA/MWP; frost fairs etc etc) and what the quantative recons show; and you have, I think, invented an inaccurate distinction between the various quantiatve recons.
- I felt that the previous distinction was confusing and misleading. I do not claim that the current distinction is perfect and there is certainly work to be done on the explanation of regional estimates. I have just a stub for those at the moment. As argued above, the grape harvest ones etc. are no less quantitative than the qualitative ones - but they are regional rather than global. I would think that the distinction that everyone argues about really is how variable the previous climate was. I think that is a valid and important distinction to highlight. I welcome edits to improve this but think that the previous distinction was not right - a regional/global distinction certainly, a distinction based on the degree of variability (which is what leads to all the arguments about whether current warming is unprecedented) certianly, quantitative/qualitative - I don't think that is the most important or most accurate. - JS 00:54, 27 Feb 2005 (UTC)
- The problem with the present beginning is that it is focused on a current climate dispute. A distinction similar to the previous one is better for explaining types of records, based upon the methods and tools for producing the records. I think the article should begin with a focus on the end results: temperature records. Then explain the types and methods of records. Various interpretation and analysis can follow. Much of the present variability distinctions should be in a section which compares various records ("this is what is similar and different in the above"), with yet other sections oriented toward interest in certain characteristics ("range of variation is of interest"). (SEWilco 09:26, 27 Feb 2005 (UTC))
- That seems like a reasonable comment. On reflection it is probably too focussed on current issues. I think an article starting with the methods and leading to the results might work but I objected to the quantitative/qualitative split (for reasons mentioned above). Nonetheless, a lot of the reason this work is interesting and relevant (and subject to controversy) is very much the question of how large is natural variability. This will not change even when the current controversy dies down. Thus, I think that the range of variability identified is a significant element of what these records bring to the scientific table and should be included in the article. I presume your edits are already in train. - JS 11:03, 27 Feb 2005 (UTC)
Subject of article
Some ambiguity exists in this article. The subject should be defined clearly. General study of past temperature probably belongs in Historical temperature record. Repetition in this article can be reduced with references to the more general article.
What is the subject of this article?
- MBH study with a range of 1000 years
- Recent temperature record (Why 1000 years? Anything significant happen 1200 or 12000 years ago?)
- Detailed temperature record
- Climate variation
- Historical temperature record
(SEWilco 18:48, 27 Feb 2005 (UTC))
- (William M. Connolley 19:35, 27 Feb 2005 (UTC)) Hmmm... there are problems with all this. When I created this page, it was 1000 years because that was the longest multi-proxy record. When I created HTR, I really meant *instrumental* temperature record. Probably, there should be:
- Temperature record - all of it, with sub pages:
- Satellite/balloon record - last 25/50 years - satellite temperature record
- Instrumental record - mostly, last 150 years; some discussion of early instrumental (back to 1750 ish). Contents: whats on HTR, up to section "Proxies: tree rings, ice cores: the last 1000 years".
- Multi-proxy records - about reconstructions of up to the last 2000 years. This current page. Should focus on the various reconstructions, ideally describe how they are done, limiatations, etc.
- Longer term records: ice cores (800 kyr); ocean cores (Myr but vaguer); other stuff I don't know about...
- Temperature record - all of it, with sub pages:
- As to the name... most recons (e.g. the recent Moberg; or von S) still only go back 1000 yr, so the name could stay for now I think.
- Moberg goes back to 1 AD.
- (William M. Connolley 20:53, 27 Feb 2005 (UTC)) Oops you're right, I was thinking of your graph.
- 3 other of the temperature comparisons in the plot extend before 1000 AD. I would strongly favor abandoning the 1000 year part of the name. This has already been in discussion above. If we change the name of the HTR to Instrumental, as you suggest, it would remove the principal objection to adding historical to the name of this page. Perhaps something like "Historical Proxy Temperature Records" would fit. I'm not convinced we have found the right name for this page yet, but dropping the 1000 years part is something that I think needs to happen. I would be happy to donate a new figure that shows the full range of existing reconstructions, if there is a suitable page to put it on. Dragons flight 20:21, Feb 27, 2005 (UTC)
- (William M. Connolley 20:53, 27 Feb 2005 (UTC)) OK, now you've pointed out Moberg, I agree with the 1000. I don't like "historical" though... because to me that means the Frost-fairs type stuff. And... if we call it proxy TR, then the Frost fairs stuff won't belong. OTOH that stuff could fit into the MWP/LIA pages perhaps. How about "Millenial temperature record"?
- Few will understand that, and it is kind on meaningless. Maybe Past Temperature Record"? --Stephan Schulz 01:24, 28 Feb 2005 (UTC)
- (William M. Connolley 09:41, 28 Feb 2005 (UTC)) But Past TR doesn't distinguish it from any of the others! Some indication of timescale would be useful.
- Why do we need the "record" part? Much of it is not recorded. Can we systematically name the articles "Global Temperature during the last X years" and start the article with a description of the different methods?--Stephan Schulz 22:24, 28 Feb 2005 (UTC)
- (William M. Connolley 23:11, 28 Feb 2005 (UTC)) I wouldn't put in "global" - it wouldn't fit, e.g., Moberg (and many others) which are NH only. "Temperature during the last X years"? Possible, but not very update-safe (as "past 1000 years" shows). Could "Temperature during past millenia" do? That implies "a few", which fits 1kyr or 2. Ideally the title should be proof against extension of the record back to 3 kyr. Or we could just live with that when it happens and pick "Temperature of the last 2000 years"?
- (William M. Connolley 09:41, 28 Feb 2005 (UTC)) But Past TR doesn't distinguish it from any of the others! Some indication of timescale would be useful.
- WMC, I think we have a semantic disagreement with "historical" and "proxy". Perhaps it is just my background, but I tend to view "historical" as everything occuring since the rise of written histories, ie. 6000 years or so. Which I think you would agree is a reasonable time frame for the article. Similarly, I count as "proxies" everything from which we estimate temperatures without the benefit of a thermometer, which to my mind includes frost fairs, agricultural records, frozen rivers, and all the qualitative stuff in addition to the tree rings and more directly quantifiable stuff. So apparently we have different preferences for interpreting the same phrases, even though I think we both have a similar image of what we want the article to be about. For the record, I'm not wild about "millenial" since that just conjures up the 1000 years again for me. Since we are changing the name of the article, I would like to include "indirect", "proxy", "reconstruction", or some other term that indicates that these are not direct thermometry measurements. Dragons flight 14:42, Feb 28, 2005 (UTC)
- (William M. Connolley 16:32, 28 Feb 2005 (UTC)) OK, we do indeed have a semantic problem, as yet with no satisfactory solution.
- Perhaps the confusion is because there is a continuum of proxies, at one end of which is a simple timeline of variations in one isotope proxy within a single core, at the other end are newspaper reports of when the Thames froze over. The first is an uninterrupted measure of a physical variation, the latter are vague and spotty reports of long events around 0°C. In between are proxies such as tree rings, which can be measured precisely and assembled in long records but there is some ambiguity as to the meaning of the data and several reasons for variations. Also less ambiguous than Thames ice news reports are long grape frost records. (SEWilco 22:26, 28 Feb 2005 (UTC))
- (William M. Connolley 23:11, 28 Feb 2005 (UTC)) To me proxy has a more limited meaning, and excludes Frost Fairs. But if it means other things to other people, then perhaps we could avoid it in the page title.
- If people agree on this then... HTR needs mv to instrumental TR; new overall page TR needs creation; and some content moves need to occur. *If* people agree here, then we could post notes on those pages *prior* to doing any major alteration.
- (William M. Connolley 16:32, 28 Feb 2005 (UTC)) No-one has disagreed with the basic proposal above (though we still have a problem with the 1-2 millenia period name). So I'll put notes on the HTR page.
- 100 years misses the start of the recent rise. 1000 years lands in Medieval Optimum. 5000 misses Holocene Optimum. Can we agree that "the last glaciation" is the oldest time to consider for current temperatures? The current temperatures seem to begin at the end of the deglaciation, around the Younger Dryas. The deglaciation period has things of interest such as temp change flowing from south to north hemisphere, but there isn't enough ice for such dramatic activity now. The instrumented record based on european thermometer technology is rather well defined. The period between YD and the thermometer is not as well defined, but too large to ignore. What other distinctions are in that period? (SEWilco 22:26, 28 Feb 2005 (UTC))
- (William M. Connolley 23:11, 28 Feb 2005 (UTC)) I don't think they should be *designed* to fit various time intervals, for the reasons you give. The split by type-of-record is better based: instrumental (globally to 1850 odd; sporadically further); "calibrated proxy" back 2000y ish; longer term (ice/ocean cores; misc).
- What is your definition of "calibrated"? Is this good enough: "At the time the oceans formed, the Earth's temperature was between 0°C and 100°C". (Sorry, I don't have a source handy for whether the oceans formed.) — SEWilco 21:28, 1 Mar 2005 (UTC)
- Now, now SEWilco, don't forget that the boiling and freezing point can change as a function of salinity and atmospheric pressure so one can only say that it was only approximately between 0-100°C when the oceans formed. :-) Dragons flight 21:38, Mar 1, 2005 (UTC)