|
Post by icefisher on Nov 6, 2009 18:25:00 GMT
That first posting on how you can always get a hockey stick when you throw out divergent cores from a modern temperature record is really well explained. Political science magic!
|
|
|
Post by socold on Nov 6, 2009 19:11:11 GMT
I see what you are saying, but if the majority do track temperature then I disagree dismissing all of them because one doesn't Perhaps the easiest to understand performing cherry picking without realizing it. rankexploits.com/musings/2009/tricking-yourself-into-cherry-picking/There are well over 1000 posts on this subject on Lucia's blog. After a while it's not difficult to sift out the chaff from the wheat. I agree with that argument only for the case that the tree ring chronologies are random. However even in that case it's easily detected there's a problem because you will get conflicts outside the validation period between different chronologies (ie before 1960 in that case). How well tree ring chronologies represent temperature is outside the scope of the original argument leveled at Briffa, which was that he had ignored a nearby site that showed the opposite trend. Briffa showed that by even including that site the result hardly changed. So we are looking at some different arguments now. It's "moving on" so to speak. The blogs are treating the tree ring chronologies as pure data, which could be red noise, whereas obviously they are not - the schweingruber and other sites agree fairly well pre-1900s. I also suspect there is context involved in the process of compiling them. Until we might hear from dendro experts, it's all a bit of speculation.
|
|
|
Post by icefisher on Nov 6, 2009 19:39:41 GMT
How well tree ring chronologies represent temperature is outside the scope of the original argument leveled at Briffa, which was that he had ignored a nearby site that showed the opposite trend. Briffa showed that by even including that site the result hardly changed. That is simply not true Socold. If tree rings perfectly reflected temperature it would not matter which trees you picked. Its only in the situation where tree ring chronologies fail to be a good proxy for temperature that you can artificially create hockey sticks. It is precisely the issue here. Briffa criticized McIntyre for picking the Khad site only explaining that it was a little bit south of the primary site. But Briffa himself had reached well further away to pick a site to go into his chronology. . . .it was that fact that McIntyre began to wonder why the Khad site was ignored when it was so obviously available as an alternative to reaching much further away. So far Briffa's argument has been it was a "small area" but if thats true McIntyre's area is smaller than Briffa's. He intentionally picked Khad because it was closer and it had been passed over. The question is why. www.climateaudit.org/?p=7656McIntyre does not claim there is no explanation but has noted no explanation has yet been given. We know that confirmation bias might be one explanation, rejecting the divergent alternatives to get so-called temperature sensitive trees. . . .a practice that makes hockey sticks every time. The fact that Briffa can go in and find chronologies that make lesser hockey sticks (he admits they are less than the original) in an area smaller than McIntyre's area still doesn't explain why he went for the bigger stick and the bigger area. Speculating, could it be because he wanted a bigger stick and a more extensive Siberian sampling so as to suggest a more relevant study for summing up regional climate change? And he could get that by gerrymandering his sample? McIntyre's original comments were right on the mark. And the issue of whether tree ring chronologies are a good proxy are right at the heart of this issue. I mean why put the economic future of nations on tiny site in Yamal, rather than say the Polar Urals, which generally have a divergent problem, or McIntyre's compilation. Briffa's new work brings more questions than it answers. Bottom line this is like putting that last jacket in the suitcase, it keeps popping out somewhere. We are all interested in what Briffa's response will be. Maybe he needs to pack a little lighter for the next trip.
|
|
|
Post by socold on Nov 6, 2009 19:46:16 GMT
How well tree ring chronologies represent temperature is outside the scope of the original argument leveled at Briffa, which was that he had ignored a nearby site that showed the opposite trend. Briffa showed that by even including that site the result hardly changed. That is simply not true Socold. If tree rings perfectly reflected temperature it would not matter which trees you picked. Its only in the situation where tree ring chronologies fail to be a good proxy for temperature that you can artificially create hockey sticks I said tree-ring chronologies, not tree-rings. They don't have to perfectly reflect temperature. But they are clearly nothing like red noise.
|
|
|
Post by icefisher on Nov 6, 2009 20:07:50 GMT
I said tree-ring chronologies, not tree-rings. They don't have to perfectly reflect temperature. But they are clearly nothing like red noise. Sorry Socold all proxies that are not perfect operate like red noise. Thats because all imperfect proxies have noise in them. The more accurate they are the less you can get out of it. . . .but remember you can scale your graphs so you can see the hockey stick very clearly. All proxies act like red noise and produce artificial hockey sticks from selective sampling.
|
|
|
Post by socold on Nov 6, 2009 21:37:51 GMT
I said tree-ring chronologies, not tree-rings. They don't have to perfectly reflect temperature. But they are clearly nothing like red noise. Sorry Socold all proxies that are not perfect operate like red noise. Thats because all imperfect proxies have noise in them. The more accurate they are the less you can get out of it. . . .but remember you can scale your graphs so you can see the hockey stick very clearly. All proxies act like red noise and produce artificial hockey sticks from selective sampling. The similar pre-1900s pattern shows they are nothing like red noise
|
|
|
Post by icefisher on Nov 6, 2009 22:24:32 GMT
Sorry Socold all proxies that are not perfect operate like red noise. Thats because all imperfect proxies have noise in them. The more accurate they are the less you can get out of it. . . .but remember you can scale your graphs so you can see the hockey stick very clearly. All proxies act like red noise and produce artificial hockey sticks from selective sampling. The similar pre-1900s pattern shows they are nothing like red noise Keep in mind it was Briffa that skipped over Khad and it was only McIntyre that picked it back up. And if you are not selecting for pre-1900 what you see is line flattening relative to the blade. Which is exactly what you see. You have to go back and reselect for pre-1900 to get a fit to the actual data. Also there is likely more than one factor at work here. As I understand it though I haven't see all the figures but there are few living trees in these chronologies which would tend to slew the most recent measurements from low levels of sampling.
|
|
|
Post by socold on Nov 6, 2009 23:49:48 GMT
The similar pre-1900s pattern shows they are nothing like red noise Keep in mind it was Briffa that skipped over Khad and it was only McIntyre that picked it back up. And if you are not selecting for pre-1900 what you see is line flattening relative to the blade. Which is exactly what you see. If you are not selecting for pre-1900 it could show anything. What you wouldn't expect, which is the case, is that all the sites show such agreement pre-1900. which looks more like a climatic influence affecting all sites.
|
|
|
Post by sigurdur on Nov 7, 2009 0:28:07 GMT
Or cosmic rays, which I guess are climatic influences.
|
|
|
Post by icefisher on Nov 7, 2009 3:14:21 GMT
If you are not selecting for pre-1900 it could show anything. What you wouldn't expect, which is the case, is that all the sites show such agreement pre-1900. which looks more like a climatic influence affecting all sites. What you are looking at Socold is an average of tree rings for that period. When you have a confirmation bias for the modern selections the blade on the end will grow higher not being averaged out with the lower values. So what you have is some correlation with temperature but an inaccurate comparison between the modern period and non-biased older period. Simple enough huh?
|
|
|
Post by socold on Nov 7, 2009 15:00:13 GMT
|
|
|
Post by icefisher on Nov 7, 2009 15:58:45 GMT
We know what the temperature should be in the modern period, if a proxy diverges from temperature in the past 50 years then it seems to me to be valid to disregard it rather than average it. Thats the wrong answer. Trees diverge in their response to temperature change. Since the modern instrument record is used to scale the change of the past, its incorrect to reject divergent samples as you cannot go back and reject the divergent historical samples unless you already know what the temperature changes were back then. . . .ending up with the reconstruction you exactly wanted to end up with. I detect you are beginning to sense the shifting ground you are standing on as you flip flop through this discussion. You cannot do this kind of anaylsis without using statistical analysis and the underlying assumption of all statistics is you have homogenus populations and that every tree ring/chronology has an equal chance of being selected and that you have a sample of sufficient size. Violate any principle and your work is garbage. Briffa's defense variously moves from one area of shifting ground to another. He is doing a jig! The main steps in his dance is to move on and move away from how the most radical possible reconstruction he was able to come up with was used for 9 years in just about every temperature reconstruction in the record. I grant he did avoid cherry picking every tree that went in to the work leaving the barest veneer of science in Briffa 2000. What one needs to understand is this is the modus operandi of the warmistas. Get stuff published then get it referenced. Build a momentum that becomes difficult to reverse.
|
|
|
Post by socold on Nov 7, 2009 16:38:46 GMT
Every tree ring/chronology shouldn't have an equal chance of being selected. Selection, or weighting, should depend on how well the chronology acts as a temperature proxy.
If a chronology fails miserably to follow recent temperature then confidence that it is a good proxy for temperature over the past 1000 years has to be considerably lower than a chronology that does follow recent temperature. Sure a chronology that follows recent temperature well might not do so in the past, but if it has tracked temperature well recently it at least has a better chance of doing so.
There's no reason to think that the set of chronologies that follow recent temperature well will bias past temperature lower (if anything it would be chronologies that show little modern increase that would be perhaps susceptible to an inability to increase)
Obviously there can be error, chronologies that validate recently still may diverge at points in the past. I think the spread in chronologies going back into the past should give an idea of the error involved.
Certainly if all chronologies that validate with recent warming fail to show a MWP 2C warmer than present I take that as evidence that the MWP likely wasn't 2C warmer than present.
As for Briffa's defense, it stands up to me. He addressed 2 specific points in one response. He would need to respond multiple times before he could even possibly be accused of "jigging" around surely.
|
|
|
Post by icefisher on Nov 8, 2009 2:22:02 GMT
Every tree ring/chronology shouldn't have an equal chance of being selected. Selection, or weighting, should depend on how well the chronology acts as a temperature proxy. Well since the test is whether rings get wider rather than narrower when it gets warmer; I would suggest that every tree has a 50% chance of just randomly passing the test. Then when you consider there are probably thousands of reasons why a tree might grow well or poorly and only one being temperature. . . .I have a little trouble picking forests of fossil trees because living trees that did nothing but get on average wider rings when temperature was going up. Now if the rings followed closely to annual temperature fluctuations, every year, now that might be a decent test. Like maybe set as a standard that it has to track temperature better than sunspot data tracks temperature since the warmistas reject that correlation all the time. Now maybe Briffa did such an analysis and rejected Khad on that basis. . . .but it is that analysis that McIntyre asked Briffa for.
|
|
|
Post by socold on Nov 8, 2009 13:59:28 GMT
Briffa didn't consider khad, but if he had the results wouldn't have significantly changed.
|
|