they weight the proxies according to their correlation with NH temperature

This doesn’t make any sense physically since trees respond to local climate and not global climate. Perhaps they should have weighted them by the r^2 of local temperature. I wonder if this would cause the LIA and MWP to reappear?

]]>Before they get to their Total Least Squares step, they weight the proxies according to their correlation with NH temperature – I’ve written about Partial Least Squares regression previously and , of course, this is Partial Least Squares. I’ve shown that the Mannian regression methodology reduces to PArtial Least Squares in the early steps – of course they didn’t realize it. So here we have two climate reconstrucions both using Partial Least Squares, both supposedly as novel inventions.

Mann made a PArtial Least Squares estimator and re-scaled it to match the variance of the temperature PC1. Hegerl et al re-scale their PLS estimator a little differently – by a Total Least Squares regression on NH temperature.

The Hegerl proxies are, as I predicted, virtually identical to the Osborn and Briffa *“independent”* reconstruction: PC1, foxtails, Yamal, Tornetrask, Yang composite, etc. It’s pretty funny.

#83. bender, speaking of bin-and-pin, have you read Mann’s discussion of endpoint smoothing – it was a Comment by Willie Soon and Reply by Mann in GRL a couple of years ago. Should be up on Mann’s website. I didn’t try to sort it out and, since it’s Mann-speak, it always takes time.

The reply by Mann in GRL compared three methods. Here’s his description of the methods:

To approximate the “minimum norm’ constraint, one pads the series with the long-term mean beyond the boundaries (up to at least one filter width) prior to smoothing.

To approximate the “minimum slope’ constraint, one pads the series with the values within one filter width of the boundary reflected about the time boundary. This leads the smooth towards zero slope as it approaches the boundary.

Finally, to approximate the “minimum roughness’ constraint, one pads the series with the values within one filter width of the boundary reflected about the time boundary, and reflected vertically (i.e., about the “”y” axis) relative to the final value. This tends to impose a point of inflection at the boundary, and leads the smooth towards the boundary with constant slope.” (Mann, 2004)

He recommends using the “minimum roughness” constraint … apparently without noticing that it pins the endpoints.

I wrote a reply to GRL pointing this out, and advocating another method than one of those three, but they declined to publish it. I’m resubmitting it.

w.

]]>