144
of the system. An imposed time-invariant forcing would have the effect of
biasing the outset of these saddles towards one or more of the regimes.
Fig 27 shows the geographical distribution of change in surface temperature between the 1950s and 1980s, as reported in IPCC (1992). The
increase in surface temperature is largest in winter, where it is concentrated over North America and northern Eurasia, and is partially offset by
cooling over the North Atlantic and the North Pacific.
The tendency for hemispheric-mean lower tropospheric temperatures to
be warm during periods in which the PNA index is positive has already
been noted by Gutzler et al (1988). For example, Fig 28 shows the 1000500hPa thickness anomalies associated with the PN A 500hPa height teleconnection patterns. At the ~urface, the temperature anomalies associated
with these thickness patterns will tend to be accentuated over land relative to the ocean, because of the smaller heat capacity of the active land
surface. These positive land surface temperature anomalies will therefore
dominate over the negative SST anomalies if a zonal mean temperature
anomaly is calculated.
More particularly, the observed change in (northern winter) mean temperature is consistent with an increase in the frequency of the regimes
shown on the left hand side of Fig 13, as suggested by the nonlinear
paradigm outlined in section 4. One further consequence of this is that,
just as the horizontal structure of decadal change should reflect the horizontal structure of the regime centroids, so also the vertical structure of
decadal change should reflect the vertical structure ofthe regimes. Now, to
some degree of approximation, the regimes themselves have an equivalent
barotropic structure, with the equivalent barotropic level at about 300hPa.
From the hydrostatic relationship, if there is a tendency for warming below the equivalent barotropic level, then there should be a tendency for
cooling above this level. This is what Angell observes, as shown in Fig
26. (It can be noted that this figure is constructed from radiosonde data
which has a land bias. However, according to our discussion above, most
of the observed warming has occurred over land regions. Hence it is likely
that, according to our nonlinear paradigm, the temperature in the upper
troposphere will be cooler in a radiosonde-based estimate of hemispheric
mean temperature, than in a satellite-based estimate with its more uniform
coverage.)
If this interpretation of the upper tropospheric cooling is correct, then
one can ask the following question.
of the system. An imposed time-invariant forcing would have the effect of
biasing the outset of these saddles towards one or more of the regimes.
Fig 27 shows the geographical distribution of change in surface temperature between the 1950s and 1980s, as reported in IPCC (1992). The
increase in surface temperature is largest in winter, where it is concentrated over North America and northern Eurasia, and is partially offset by
cooling over the North Atlantic and the North Pacific.
The tendency for hemispheric-mean lower tropospheric temperatures to
be warm during periods in which the PNA index is positive has already
been noted by Gutzler et al (1988). For example, Fig 28 shows the 1000500hPa thickness anomalies associated with the PN A 500hPa height teleconnection patterns. At the ~urface, the temperature anomalies associated
with these thickness patterns will tend to be accentuated over land relative to the ocean, because of the smaller heat capacity of the active land
surface. These positive land surface temperature anomalies will therefore
dominate over the negative SST anomalies if a zonal mean temperature
anomaly is calculated.
More particularly, the observed change in (northern winter) mean temperature is consistent with an increase in the frequency of the regimes
shown on the left hand side of Fig 13, as suggested by the nonlinear
paradigm outlined in section 4. One further consequence of this is that,
just as the horizontal structure of decadal change should reflect the horizontal structure of the regime centroids, so also the vertical structure of
decadal change should reflect the vertical structure ofthe regimes. Now, to
some degree of approximation, the regimes themselves have an equivalent
barotropic structure, with the equivalent barotropic level at about 300hPa.
From the hydrostatic relationship, if there is a tendency for warming below the equivalent barotropic level, then there should be a tendency for
cooling above this level. This is what Angell observes, as shown in Fig
26. (It can be noted that this figure is constructed from radiosonde data
which has a land bias. However, according to our discussion above, most
of the observed warming has occurred over land regions. Hence it is likely
that, according to our nonlinear paradigm, the temperature in the upper
troposphere will be cooler in a radiosonde-based estimate of hemispheric
mean temperature, than in a satellite-based estimate with its more uniform
coverage.)
If this interpretation of the upper tropospheric cooling is correct, then
one can ask the following question.