Keep up the good work.

Are we still talking about, and operating on, these data? (Presumably referred to as MBH98).

I use a method of analysis completely different from your elegant work, making no serious attempt to derive a decent quantitative estimate of the “uncertainties”, although I can, under various assumptions, produce one.

The method I use provides a grand overview of the data. It avoids setting up an hypothesised model, such as the very commonly used and accepted linear one, which for all we know might well be totally unrelated to the reality behind the origins of the numerical data.

The technique derives the grand patterns in the data, and enables the behaviour of any chosen subgroups of the data to be compared directly. This yields outcomes that I find very instructive, and which incidentally seem to refute completely any suggestion of “hockey stick” behaviour, save for a few isolated single site series.

If current thinking is that there might be a “more reliable” data set or sets, would it be possible to get a direct link to them?

I’d much appreciate any thoughts you (or anyone else) may have.

Robin

It turns out that the remark by LB is actually more or less a direct quote from an article written by Mann, although that wasn’t immediately obvious.

The article is ‘Climate Over the Past Two Millenia’, the Annual Review of Earth and Planetary Sciences, vol35, 111-136, publication date May 2007. The quote comes from the bottom of p113.

http://arjournals.annualreviews.org/doi/abs/10.1146/annurev.earth.35.031306.140042@climate.2007.1.issue-1

The script is still not working for me. Hard coding the scale.method object took care of the first object not found error, but now I get the object not found error referencing an object named Sxxinv. I have since updated my copy of R, but no joy still when using your script.

Thanks for the R education and refreshing my statistics skills.

Steve: Sorry bout that. I need to re-run this not in a session that’s already open. Sxxinv should be solve(Sxx) .

I think the ’1500 CE’ refers to the Luterbacher et al. 2004 proxies only, not the whole study.

Take a look at Craig Loehle’s study for something a bit more interesting.

“Mann et al. (1998, 1999) used a network of 415 annually resolved proxy data … to reconstruct temperature patterns over the past thousand years. Zhang et al. (Z. Zhang, M. Mann, S. Rutherford, R. Bradley, M. Hughes et al., manuscript in preparation) have more recently assembled a much larger network of 1232 annually resolved proxy data consisting of tree rings, corals and sclerosponge series, ice cores, lake sediments, and speleothems combined with reconstructions of European seasonal surface temperatures back to 1500 CE based on a composite of proxy, historical, and early instrumental data (Luterbacher et al. 2004) … The additional inclusion of non-annually resolved, but still relatively high (e.g., decadal), resolution proxies (e.g., nonlaminated lake and ocean sediments) with high enough resolution and accurate enough age models to calibrate at decadal resolution leads to an even larger network of 1302 proxy series.

I make that 887 “new” proxies.”

I wondered if you Steve, or any one else was aware of this new paper?

Steve: Nope. But the “415 proxies” are the same old MBH98 proxies, Graybuill bristlecones and all. They will probably add in the Briffa MXD network used in Rutherford et al. but I’m not sure what else will be in it. But always keep in mind that the vast majority of these records are very short and do not affect things before 1400. The issue will be what’s in the MWP network – bristlecones. And there will be a lot of Graybill bristlecone chronologies, that’s for sure.