146
However, this does not imply that the observed surface warming is
not associated with increases in CO2. To see this, consider first a more
parochial argument. If it is anomalously warm over London, then the
chances are that the winds have a more southerly than normal component.
If it is persistently warmer over London then the chances are the winds are
persistently more southerly. This mayor may not be due to enhanced CO2.
However, if one wants to know whether the fact that London is warmer is
due to enhanced CO2 then one must ask whether enhanced. CO2 will cause
the wind to be more southerly.
Hence, similar to the 'London' argument, if one wants to know whether
the fact that the hemispheric mean temperature anomaly has been persistently positive is due to enhanced CO2 , then one should ask whether
enhanced CO2 will increase the probability of the regimes in Fig lla,c being observed. The question then arises; how could such an increase in the
PDF of these regimes be most effectively achieved? Certainly, the adjoint
analysis that we have discussed in this paper suggests that the most effective way to force an increase in the frequency of a regime is not achieved
by a forcing which projects directly on the regime's geographical structure.
Precisely this question arose in section 4.2 in trying to understand what
processes model systematic error was most sensitive to. One way to study
this is through the sensitivity analysis described in section 3.5. We can
imagine taking an atmospheric model trajectory over, let's say, a decade.
For every trajectory segment say of order 10 days or so, we can compute the
sensitivity of the positive PNA regime by integrating the regime backwards
with the adjoint model. Each trajectory segment will have its own sensitivity pattern. By compositing over many different trajectory segments we
get a mean sensitivity for the regime of interest.
Such a calculation is currently in progress (Susanna Corti, personal
communication). However, the adjoint eigenmode results shown in Fig 20
already gives us an approximate idea of the difference between the sensitivity structure and the regime structure. In particular, Fig 20 suggests that
extratropical quasi-stationary Rossby wave patterns are likely to be sensitive to forcing in the Indonesian area. This result is consistent with the SST
anomaly experiment shown in Fig 21, where a localised SST anomaly of
about lK could induce a substantial extratropical response, not altogether
different from the positive PNA regime patterns shown in Fig II.
Hence, even though we might only be interested in anthropogenic changes
However, this does not imply that the observed surface warming is
not associated with increases in CO2. To see this, consider first a more
parochial argument. If it is anomalously warm over London, then the
chances are that the winds have a more southerly than normal component.
If it is persistently warmer over London then the chances are the winds are
persistently more southerly. This mayor may not be due to enhanced CO2.
However, if one wants to know whether the fact that London is warmer is
due to enhanced CO2 then one must ask whether enhanced. CO2 will cause
the wind to be more southerly.
Hence, similar to the 'London' argument, if one wants to know whether
the fact that the hemispheric mean temperature anomaly has been persistently positive is due to enhanced CO2 , then one should ask whether
enhanced CO2 will increase the probability of the regimes in Fig lla,c being observed. The question then arises; how could such an increase in the
PDF of these regimes be most effectively achieved? Certainly, the adjoint
analysis that we have discussed in this paper suggests that the most effective way to force an increase in the frequency of a regime is not achieved
by a forcing which projects directly on the regime's geographical structure.
Precisely this question arose in section 4.2 in trying to understand what
processes model systematic error was most sensitive to. One way to study
this is through the sensitivity analysis described in section 3.5. We can
imagine taking an atmospheric model trajectory over, let's say, a decade.
For every trajectory segment say of order 10 days or so, we can compute the
sensitivity of the positive PNA regime by integrating the regime backwards
with the adjoint model. Each trajectory segment will have its own sensitivity pattern. By compositing over many different trajectory segments we
get a mean sensitivity for the regime of interest.
Such a calculation is currently in progress (Susanna Corti, personal
communication). However, the adjoint eigenmode results shown in Fig 20
already gives us an approximate idea of the difference between the sensitivity structure and the regime structure. In particular, Fig 20 suggests that
extratropical quasi-stationary Rossby wave patterns are likely to be sensitive to forcing in the Indonesian area. This result is consistent with the SST
anomaly experiment shown in Fig 21, where a localised SST anomaly of
about lK could induce a substantial extratropical response, not altogether
different from the positive PNA regime patterns shown in Fig II.
Hence, even though we might only be interested in anthropogenic changes