Applicability of Oscillatory Climate-Mode Indices for the Diagnosis and Prognosis of Interannual and Longer Time Scale Climate Variability of the Northeastern Pacific and Western North America

For a number of reasons, oscillatory climate-mode or teleconnection indices—like ENSO, PDO, AMO, AO (and all the other O's) are receiving a lot of attention in the analysis of both the modern and paleo climatic variability of the northeastern Pacific and western North America. First, although generally defined using a single variable (e.g. SSTs) they often organize multiple climatic variables into a set of recognizable and recurrent patterns, and consequently can greatly reduce the dimensionality of climate. This reduction also provides a small number of indices to which the more applied aspects of regional climatic responses (e.g. wildfires, drought, etc.) can be correlated. Second, because some climate-mode indices vary slowly relative to the timescales of the data with which they are defined or compared, they carry with them the promise of predictability, or at the least the ability to organize regional responses into "regimes." Third, the temporal and spatial patterns of climate modes may provide clues as to the physical mechanisms that generate the climatic variations that the modes represent. Some simple and multiple time series models fit to observed climate-mode indices illustrate that: 1) the apparent quasi-periodic behavior of some indices (like the PDO) is not inconsistent with simple "red-noise" models, or models in which persistence or memory can generate quasi-periodic variations; 2) spurious decadal-scale periodicity can easily be induced by simple data-analytical treatments like smoothing; and 3) there are probably only two principal modes of variability in North Pacific climate-mode time series, ENSO and global warming. Consequently, while they may be useful for the diagnosis of interannual and longer timescale variability, the applicability of oscillatory climate-mode indices for the prognosis (i.e. prediction) of such variability is limited.