72
[SSTA ]
~~~:~ =:a:: :g:S::~$~Q:%? ~IT
r\
rI\ ~ Ci'Y\ ~ RES
1900
1920
1940
1960
1980
Figure 31: Time series of (top) global-mean SST anomalies; (middle) expansion coefficient
of the regression field shown in the previous figure , scaled as a least squares best fit to the
global-mean SST time series (FIT) shown together with the expansion coefficient of the
leading EOF of global the SST field (G) ; and (bottom) residual variations in global-mean
SST, obtained by subtracting the fitted curve from the time series of global-mean SST. The
interval between tick-marks is 0.5 K. All time series have been smoothed by successive
applications of 5-month and 9-month running mean filt ers. The spacing between the
curves is arbitrary. After Zhang et al.(1996).
are shown in Fig. 29. Within the tropics, both patterns are reminiscent of
the differences between contrasting polarities of the ENSO cycle as represented in the bottom panels of Figs. 19 and 20. The SLP falls and cooling
over the extratropical North Pacific are also reminiscent of the anomalies
associated with the ENSO cycle, but they are more prominent relative to
the changes in the tropics. Like the ENSO cycle, this interdecadal-scale
change is clearly apparent in the time series of tropical surface air temperature (Fig. 28), and global-mean SST (Fig. 30) .
Whether this feature is in any way unique or extraordinary is difficult
to assess using conventional measures of statistical significance, for the
reasons discussed in section 2.5 of chapter 1. But unique or not, it serves
as a clearly discernible marker in many different time series. A similar shift
occurred in 1958 but the subsequent "warm regime" persisted for only 3-4
years. Mantua and Graham (1996) have shown evidence that yet another
one occurred in 1990. It is only with the benefit of hindsight that it is
possible to distinguish between the short-lived "regimes", whose histories
are most appropriately viewed in the context of the interannual variability,
and the long-lived ones that contribute substantially to the interannual
variability in the climate record. One could also question whether the
apparent abruptness of features such as these might not be illusory: a
product of the superposition of the interannual variations associated with
the ENSO cycle upon more gradual interdecadal variations.
[SSTA ]
~~~:~ =:a:: :g:S::~$~Q:%? ~IT
r\
rI\ ~ Ci'Y\ ~ RES
1900
1920
1940
1960
1980
Figure 31: Time series of (top) global-mean SST anomalies; (middle) expansion coefficient
of the regression field shown in the previous figure , scaled as a least squares best fit to the
global-mean SST time series (FIT) shown together with the expansion coefficient of the
leading EOF of global the SST field (G) ; and (bottom) residual variations in global-mean
SST, obtained by subtracting the fitted curve from the time series of global-mean SST. The
interval between tick-marks is 0.5 K. All time series have been smoothed by successive
applications of 5-month and 9-month running mean filt ers. The spacing between the
curves is arbitrary. After Zhang et al.(1996).
are shown in Fig. 29. Within the tropics, both patterns are reminiscent of
the differences between contrasting polarities of the ENSO cycle as represented in the bottom panels of Figs. 19 and 20. The SLP falls and cooling
over the extratropical North Pacific are also reminiscent of the anomalies
associated with the ENSO cycle, but they are more prominent relative to
the changes in the tropics. Like the ENSO cycle, this interdecadal-scale
change is clearly apparent in the time series of tropical surface air temperature (Fig. 28), and global-mean SST (Fig. 30) .
Whether this feature is in any way unique or extraordinary is difficult
to assess using conventional measures of statistical significance, for the
reasons discussed in section 2.5 of chapter 1. But unique or not, it serves
as a clearly discernible marker in many different time series. A similar shift
occurred in 1958 but the subsequent "warm regime" persisted for only 3-4
years. Mantua and Graham (1996) have shown evidence that yet another
one occurred in 1990. It is only with the benefit of hindsight that it is
possible to distinguish between the short-lived "regimes", whose histories
are most appropriately viewed in the context of the interannual variability,
and the long-lived ones that contribute substantially to the interannual
variability in the climate record. One could also question whether the
apparent abruptness of features such as these might not be illusory: a
product of the superposition of the interannual variations associated with
the ENSO cycle upon more gradual interdecadal variations.