(2) you have not responded to the false MBH claim of robustness to the presence/absence of all dendroclimatic indicators when their reconstruction is not robust to the prsence/absence of bristlecones

(3) you have not repsonded to MBH withholding the CENSORED results without bristlecones;

(4) "The fact that the end of the reconstructed proxy yields such a good resemblance to the instrumental data". The cross-validation R2 of the 15th century proxy network is about ~0 - hardly a good resemblance. In the AD1820 step, MBH uses 112 proxies including actual instrumental data to "predict" 79 years of data, so a decent R2 is almost unavoidable.

(5) you have not addressed the MBH withholding of adverse cross-validation R2 data.

(6) the Milankowitch point is not argued in MBH and provides no evidence whatever;

(7) the "agreement" with other reconstructions is not particularly close; but the other studies are not independent. There is extensive overlap of proxies - 14 of 17 series used in Jones et al [1998] are used by MBH. Bristlecones, Polar Urals, Tornetrask are used over and over. The coauthors overlap. So similarity to these other studies means virtually nothing.

(8) You say: “When I do a Google on ‘Bristlecone temperature’, I get 27500 hits, but when I include ‘flawed’ in the searxch string, I only get 297 (your name keeps cropping up…). Why such a difference if your claim about their quality were true?” What a ridiculous point. Most of the hits have nothing to do with bristlecones as a proxy but with travel information or the like.

(9) MBH claimed that the proxies were carefully selected and people propose policies based on this. The onus is on them to show that the proxies are valid. I have raised questions from specialist literature. What is particularly disquieting is that MBH withheld information about the bristlecones and adverse cross-validation statistics, so previous scientists were not necessarily aware of the problems.

‘A response of plant growth to increased atmospheric carbon dioxide, which has been anticipated from laboratory data, may now have been detected in the annual rings of subalpine conifers growing in the western United States. Experimental evidence shows that carbon dioxide can be an important limiting factor in the growth of plants in this high-altitude environment. The greatly increased tree growth rates observed since the mid-19th century exceed those expected from climatic trends but are consistent in magnitude with global trends in carbon dioxide, especially in recent decades. If correctly interpreted, these findings have important implications for climate studies involving tree ring observations and for models of the global carbon dioxide budget.’

Graybill & Idso:

‘The growth-promoting effects of the historical increase in the air’s CO2 content are not yet evident in tree-ring records where yearly biomass additions are apportioned among all plant parts. When almost all new biomass goes into cambial enlargement, however, a growth increase of 60% or more is observed over the past two centuries. As a result, calibration of tree-ring records of this nature with instrumental climate records may not be feasible because of such growth changes. However, climate signals prior to about the mid-19th century may yet be discovered by calibrating such tree-ring series with independently derived proxy climate records for those times.’

These seem to concern CO2 and tree-rings rather than temperature and tree rings. Isn’t there a very big leap to say that this implies that tree rings are not of use for temperature construction? You need to prove each step of this chain of logic, before making strong arguments about the Bristle cone data. I mean, the fact that there are tree rings, is a consequence of seasonally varying temperature (precipitation), after all? Please eluciadate us…

Rasmus

Perhaps the null hypothesis should be that the document frequency of “flaw”, “flawed”, etc is ~1/1000.

Cheers, — Jo

If it is true as you say, that the Bristlecone is not correlated with temperature or whatever predictor, then a multiple regression wouldn’t attach much weight to it. Of course, previous claims about the Bristlecone may be wrong. Moreover, a stepwise multiple regression ought to pick up the ’signal’ and discard the noise. The fact that the end of the reconstructed proxy yields such a good resemblance to the instrumental data, and the agreement between long slow decline before that and the Milankovitch theory, point provide independent evidence for the curve’s validity. Now, these large-scale features in the MHB-curve are also in good agreemetn with other reconstructions, that is within reasonable bounds of uncertainty. (These paleoclimatic reconstructions are generally associated with a high degree of uncertainty compard to instrumental data, due to sparse information.) When I do a Google on ‘Bristlecone temperature’, I get 27500 hits, but when I include ‘flawed’ in the searxch string, I only get 297 (your name keeps cropping up…). Why such a difference if your claim about their quality were true?

Cheers!

Rasmus

you say: “PCA can sometimes give you different first modes if the leading eigenvalue is not well-separated from the next.” This is true, but not relevant here. The pattern of the bristlecones is quite well-separated.

But you’re missing the point:

1. Mann et al. said that their PC1 was the “dominant component of variance” – they didn’t say that it was just the bristlecones.

2. Mann et al. said that their reconstruction was robust to dendroclimatic indicators in total. A fortiori, it should be robust to the presence/absence of bristlecones.

3. The bristlecones are stated not to be a temperature proxy by the specialist authors and a”mystery” by coauthor Hughes. The MBH reconstruction should NOT depend on such controversial and tainted proxies.

4. Mann et al. should have disclosed all this and not made the claim about robustness to the presense/absence of all dendroclimatic indicators.

You can’t just do another data mining step at the regression step as Mann et al. do. If the bristlecones are not a tmeperature proxy, then you have to check for spurious regression results. Did MBH check the Durbin-Watson statistics as one way of minimizing chances of spurious regression? If they did, they ignored the results.

I don’t have time right now to parse IPCC application of the hockeystick, but I disagree with the suggestion that the HS was presented as an incidental argument in IPCC and that its usage was consistently nuanced and qualified.

The only thing that can do that is a comparison of the local temperature from the bristlecone sites, and the growth rates of the bristlecones. Graybill did this study, and published it, as M&M reference. MBH accept this to be true – see MBH’99.

Where we know the relationship of temperature to bristlecone growth, the relationship is poor. How can data where we don’t know the temperature validate that ?

what is more, I am bemused by your approach to Graybill’s data. Graybill published; his data and interpretations are a matter of public record. You can go look at that data and form your own view. McIntyre has paraphrased what Graybill said- what is wrong with agreeing with the authors ?

I am suprised you do not wish to form a view on the relationship between temperature and bristlecone growth.
yours
per

Again, it’s not the single PC shapes that matter, but the final regression results. I think it is easy to get confused by focusing on just ONE PC. PCA can sometimes give you different first modes if the leading eigenvalue is not well-separated from the next. Thus, by removing some series can result in a different leading principal component, and there is nothing suspicious about that. The regresion step finds the best wieghted combination of N PCs, and the important question is therefore how many PCs were included and what fraction of the variance these account for.

Most of the recent upturn is due to instrumental data. The real ‘hockeystick has an upturn at the end, but this is small compared with the superimposed instrumental data. The fact that the end of the ‘hockey stick’ matches the beginning of the warming seen in the instrumental data so well, and the fact that the slow decline before this matches the orbital parameter theory, is interesting (this too is an independent validation…). In fact, the hockey stick may serve as a counter evidence to your proposition – that the bristle cone data are not representative of the temperature, is false.

But, the way I see it, all paleoclimatic proxies are uncertain, and may be influenced by other factors (eg moisture, light, etc), and one must be careful to put too much confidence in such ‘reconstructions’. IPCC TAR states only that it is LIKELY (66-90% chance) that the 1990s have been the warmest dacade and 1998 the warmest year of the millennium (p. 3). I think this statement is fair, but also that this does not pu too much importance on the proxy.