If you are attempting to obtain an overall trend for the continent

Given the fact Antarctica has significantly different climate zones with divergent climate trends, doesn’t Steig’s methods and/or PCA of any number give inherently incorrect results? Don’t the climate zones need to be classed with a k-PCA at a minimum, with special handling to prevent the low variability and the small number of interior Antarctic sites being disregarded in the transformations?

But if you use lower order PCs, then surely Steig should have considered relevant literature like Smith et al 1996, which warns against the precise situation that occurs. Smith et al 1996 is even cited by coauthor Mann in MBH98 and Smith’s methods are discussed at length in Schneider 2001. This is not an esoteric reference. It’s the sort of thing that maybe he should discuss in his classes on PCS.

I’m not so sure that Smith’s SST data for the 19th century is any better than Antarctic station data in the 1960s. I’d be surprised if it was. The enterprises are similar.

Steig promoted his calculation as follows:

“People were calculating with their heads instead of actually doing the math,” Steig said. “What we did is interpolate carefully instead of just using the back of an envelope. While other interpolations had been done previously, no one had really taken advantage of the satellite data, which provide crucial information about spatial patterns of temperature change.”

And you say that his calculation is OK for the “back of the envelope”.

Admittedly, Steig has to deal with having much less data and many fewer locations than Smith et al did. For a back-of-the-envelope estimate, what Steig did is reasonable. But because the rough estimate doesn’t show dramatically different results, it shouldn’t have been published and something more detailed should have been done.