Continuation of a question - question?

Question

I know this isn't a question, but there was not enough space for my comment. So, give me a break and do not report it.

http://answers.yahoo.com/question/index?qid=20140315181904AA5PLeF

pegminer and Hey Dook --

I hope this is useful and answers your questions. If not, drop me a note.

[NOTE: Out of habit - I realize that I mostly refer to tree-rings in the following discussion even though Mann used a multi-proxy data set. It doesn’t matter because the pre-1400 data are dominated by the tree-ring data and the regression analysis uses the tree-ring data (the reason Michael used the tree-data is because it is the only proxy record with annual resolution and he needed that resolution in order to run the regression analysis).]

[NOTE: The biology and math for all of this was established by Hal Fritts. Hal is the only person I know who invented a scientific discipline all by himself. For that reason, I used to call him, “Big Daddy.” http://www.ltrr.arizona.edu/~hal/hal1.html]

Here is a down & dirty version:

The principle that allows us to decompose signals from tree-ring series is based on a model where tree-ring growth for year t, [R(t)] = B(t) + C(t) +D1(t) + D2(t) + E(t); where [B] is the genetic / biological / physiological growth component, [C] is climate, [D1] and [D2] are local and regional disturbances, and [E] is everything else (random…noise…error…whatever).

The challenge is to control / eliminate everything except climate (e.g., precip, temp, drought). Tree stands where climate variables limit growth are typically found at a species’ extreme distributional boundaries (i.e., upper treeline and lower forest border stands on steep slopes with minimal soil development). Mann’s dilemma with the 1,000-year long reconstruction was caused by thin sample depth prior to AD 1400 that was leveraged by a PC dominated by upper elevation bristlecone pine series containing a growth trend beginning in the mid-nineteenth century that is greater than can be explained by temperature, alone.
It has been suggested that this is the result of CO2 precipitating out of the atmosphere at high elevations and increasing the water-use efficiency in some trees. Regardless of what it IS – it IS NOT temperature – and it needs to be removed before building the reconstruction. It needs to be removed because the reconstruction equation is developed using regression techniques over the common time period for both the tree-rings and the climate data – in this case, the historic temperature record.

This is called the calibration period and in Mann’s study it was 1902 – 1980. Using (1) the historic temperature record and (2) the tree-ring record from 1902-1980, you build a regression equation (that describes the climate-tree growth relationship). This equation is called a transfer function because it converts proxy data into temperature data. You then take the full set of tree-ring (proxy) data from 1000-1998 and run it through the regression equation (i.e., transfer function) – and you end up a reconstruction of temperature from 1000-1998.

Now, here is the reason Mann needed to adjust the data. If you leave the CO2 signal in, it gets included in the regression analysis and contributes to defining the parameters in the transfer function. Therefore, your transfer function does not describe the relationship between historic temperature data and proxy temperature date – it describes the relationship between historic temperature data and a proxy record that is some unknown combination of temperature and CO2.

This creates two problems (in addition to the big problem of not knowing what your data represent): (1) because there is a relationship between CO2 and temperature, you are in some sense counting the same thing twice – and, therefore, artificially inflating the correlation in you regression analysis, and more importantly; (2) you are imposing the recent CO2 effect on the whole length of record – which means that, especially for the period from 1000-1400, the reconstruction contains (is contaminated by) a CO2 signal that did not exist from 1000-1400.

Michael addressed the problem by creating a residual series – which contrary to McIntyre’s whining, is not difficult or impossible to figure out – simply by subtracting the proxy data unaffected by CO2 from the PC containing the upper elevation tree-ring data. I’m not sure that I would do it that way, but in Michael’s defense he explicitly states that he is only concerned about the low-frequency signal.

Michael understood what was going on and recognized the problem – something that McIntyre never saw. McIntyre may have some math skills, but he does not know shlt about this stuff – or scientific research, in general.

Answer 1

The biggest problem with tree ring data, is by using or discarding certain tree rings you can get whatever result you want



Edit

Gary what I am saying is there is a big scope for cherry picking, I am not saying it was done, I am saying it could easily be done

Answer 2

> Michael was wrong to use the term “CO2 fertilization” (which is a hypothesis) when the data were tree-ring measurements (which are empirical facts) – and he was incorrect in attributing it to the Graybill and Idso paper.



Had I known this, I probably wouldn't have asked the question. Then you seem to be saying he is adjusting for CO2, which is the main point.



>simply by subtracting the proxy data unaffected by CO2 from the PC containing the upper elevation tree-ring data.



Are you saying he is subtracting out the proxies other than the bristlecones?





Now, here is what I am seeing. Let me know if you agree or disagree.

1) Co2 effect is from sometime in the 1800s on, creating a high growth in the tree-rings(theorized).

Let's say from 1800 on for the remaining points.



2) So, if he is adjusting for CO2, he should change values of the proxy for 1800-1980.



3) MBH98 does not have an adjustment for the bristlecones.



4) MBH99 looks identical to MBH 98 from 1400 on.



5) The bristlecones dominate PC1.



6) The result of the detrend is to change the values not from 1800 on, but from 1000-1400.



7) I can see how this can happen with some sort of anomaly calculation, but 1400-1800 hasn't changed either, as seen from 4).



EDIT:

Wasn't expecting you to agree with all that. I think that is reason for not adjusting as Mann did, and you seem to think there is no problem.



The effect is to increase the pre-1400 values in PC1. That was the point of the charts in my other question,

https://answersrip.com/question/index?qid=20140313081421AAT3IpN

Exhibit B is MBH99, while A is what it would be without the fix for CO2.

If there is an adjustment for CO2, then shouldn't the 1800-1980 portion be reduced?



Don't know if this is what you were talking about, but Mann says the CO2 effect is saturated in the 20th century. Does this sound right to you? Seems to contradict Graybill and Idso.

I should ask, are the Graybill papers still valid, or is there new work that overrides anything in them?

Answer 3

Well, technically, it is an additional detail to a prior question, but there is more informational content (e.g. at least some) here than in a hundred boilerplate cut-and-pasted Wattsup fake questions by deniers.



I cannot decipher "PC" in "leveraged by a PC dominated by upper elevation bristlecone pine" (plot chart?) but I think I get the general drift despite being neither a scientist nor a statistician. The statistically robust calibration period, where there is reliable data on tree rings, temperatures and a bunch of other stuff, is used to develop a model applied to earlier centuries where the only ongoing consistent data are those of tree rings. Through some process I don't quite follow, "proxy data unaffected by CO2" are subtracted "from from the PC containing the upper elevation tree-ring data." I can guess that since the proxy for temperature is tree rings, the "unaffected" tree rings are those early in the 1902-80 calibration period. At any, it is assumed that the proxy has been tweaked in recent years by CO2 (which can be measured for those recent years) AS WELL as climate (which can also be measured). So the model is adjusted so that it incorporates only the climate relationships. Then it is applied to earlier centuries, where the known tree rings, and known or reasonably estimateable precipitation numbers are used to determine or "reconstruct" the unknown temperature record.



Clearly understanding this stuff requires an IQ orders of magnitude above the primate-level Pavlovian denier mentality: Wattsup told me Mann is a dangerous socialist, therefore no lie about him is too stupid to use in the fight to save us from Satanic Hussein ObamaCare.



But, I would appreciate brief feedback re



1) Is my understanding (above) approximately correct or not?

2) Do I also correctly infer that although Mann's reconstruction is based on assumptions more than on hard data, it is roughly corroborated by other estimates (e.g. from ice core bubbles' isotopic ratios) of past temps?

3) Does the "decline" to "hide" have something to do with applying after 1980 a pre 1902 correction to a model where somehow less correction is called for? Can you elaborate or supply a link?

Answer 4

My spouse and i suggest using VPNPower to be able to unblock sites. I've been using them since four years. http://www.vpnpower.net

Answer 5

My goodness. Either you have accurate data or you don't.

Answer 6

My goodness. Either you have accurate data or you don't. If you don't let us face it, you don't and no amount of formula manipulation is going to get you to accuracy.



Besides, I wouldn't trust Mann with an empty sucker stick, let alone data on which legislation depends.