Comment on 'Regime shifts and red noise in the North Pacific'

摘要

A recent paper of ours (Rudnick and Davis, 2003, hereafter RD) has generated debate on the nature of interdecadal climate and ecosystem variability. At issue is whether this variability is best characterized as a stationary random process, or as a sequence of distinct "regimes" with different statistics. RD reiterated a point previously made by many others (e.g. Wunsch, 1999) that the shortness of the available time series affects the reliability of statistics than can be calculated. In particular, RD examined a composite analysis of time series (Ebbesmeyer et al., 1991; Hare and Mantua, 2000) that purported to identify regime shifts. Using time series with stationary statistics and plausible frequency content, RD demonstrated that the composite analysis had an uncomfortably high rate of false positives. RD did not conclude whether or not true regimes exist, or physical and biological processes are coupled. The fundamental result of RD was that a previously widely accepted analysis was questionable.