While I was reading about rotated varimax PCA in connection with Rob Wilson’s article, I came across R.W. Houghton and Y.M. Tourre, 1992, Characteristics of Low-Frequency Sea Surface Temperature Fluctuations in the Tropical Atlantic, Journal of Climate
Volume 5, Issue 7 (July 1992) pp. 765—772 url. They observed that a PC analysis applied to Atlantic SSTs yielded a dipole in the 2nd EOF. Although this article is not discussed in Vimont and Kossin’s discussion of the Atlantic Multidecadal Mode, both articles seem to be probing similar data – with the Atlantic Multidecadal Mode looking very comparable to the dipole of Houghton and Toure.
Houghton and Toure carried out a rotated varimax on their PCs. The rotated EOFs had very different characteristics. One rotated EOF was concentrated north of the ITCZ and the 2nd rotated EOF was located south of the ITCZ.
Since the series north of the ITCZ was orthogonal to the series south of the ITCZ, the unrotated EOFs more or less corresponded to the sum and the difference of the two rotated EOFs. Houghton and Toure concluded that there was no physical significance to the dipole – other than being the empirical difference of two uncorrelated series.
This seems like an issue well worth reflecting on if one is discussing the Atlantic Multidecadal Mode in a hurricane context.