Is the IPO Real? What about the PDO?

There has been a surge in interest in the Interdecadal Pacific Oscillation (IPO) as a prominent quasi-periodic variation in the Pacific sea-surface temperature (SST). Through its low-frequency modulation of El Nino-Southern Oscillation (ENSO), the IPO impacts the global-mean temperature on multi-decadal timescales.  The IPO is thought to be “ENSO-like” and therefore its center of action is in the tropical Pacific. The IPO was originally defined by Folland et al. (1999) as the third Empirical Orthogonal Function (EOF) of the 13.3-year low-passed global sea-SST in a record length of 130 years. The original spatial pattern was not “ENSO-like”; it has two centers of action: one in the northern Pacific and a slightly larger one in the North Atlantic.  It later became “ENSO-like” when Parker et al. (2007) let in more of the decadal tail of ENSO spectrum with an 11-year low-pass filter. This ENSO-like pattern is absent if a 15-year low-passed filter is used. The IPO is likely ill-defined statistically due to the small number of degrees of freedom of the low-passed data.  We examine here the effect on the spatial patterns (the eigenvectors) due to the “effectively degenerate eigenvalues” (North et al. 1982) caused by finite sampling errors. While there is undoubtedly some multi-decadal modulation of the higher-frequency ENSO phenomenon in the equatorial Pacific, it is not clear that such modulation is centered in the Pacific. We investigate the effect of the “ENSO-like” multidecadal variation on the global-mean temperature and find it to be weak. We next explore alternative ways of extracting the “ENSO-like” multidecadal pattern.  One procedure used by some is to first extract the Pacific Decadal Oscillation (PDO) using only data in the North Pacific north of 20N, and then regress the global SST onto this leading Principle Component (PC). “ENSO-like” spatial pattern appears in the equatorial Pacific through such a regression but it is associated with the interannual part of the frequency spectrum of the PDO through the influence of the ENSO on the North Pacific. The multi-decadal component of the PDO is weak.