Regime shifts

Fig. 1. Regime shifts in (top) March-April MEI and (bottom) October-December SOI, as obtained using the regime shift detector with the following parameters: Target significance level p* = 0.2, regime length l = 20 years, and Huber parameter h = 1.

Due to strong interannual variability in ENSO, the regime shift analysis of climatic variables in the equatorial Pacific is very difficult. Nevertheless, the regime shift detector was able to identify statistically significant shifts in some of the variables. For example, the March-April Multivariate ENSO Index (MEI) experienced regime shifts in 1977 and 1999, which were statistically significant at p < 0.001 and p < 0.01 respectively (Fig. 1, top). The same shifts were found in the March-April Southern Oscillation Index (SOI, not shown). In addition, the SOI experienced regime shifts in the time series for the prewinter (October-December) season (Fig. 1, bottom). For monthly SOI values, the regime shifts in 1976 and 1998 are detectable even at the lower target significance level of p* = 0.01 (Fig. 2). The actual significance levels for both these shifts p < 0.00000001.

The regime shift in the late 1990s was also observed in the North Pacific (a review for that region is coming soon). It may serve as an additional confirmation of the shift in ENSO indices, since some of the decadal variability in the Tropics may have a subtropical or extratropical origin (see Power et al., 2006 for a review).

Fig. 2. As in Fig.1, except for monthly SOI values, with p* = 0.01 and l = 250 mos.