374
transport and air temperatures, which cannot be fully understood by the
more local analysis of convective effects. An analysis of a coupled GeM
can be aided considerably by the conceptual framework developed in this
paper. The effects described here must act in the complex model as well;
if they are unimportant in the GeM, this begs for an explanation of what
processes do govern the GeM dynamics. We see the recurrent theme in
dynamical meteorology and oceanography that simple models are needed
to build up a complete understanding.
The sensitivity experiments of section 3 have made extensive use of a
general strategy to analyze various feedback processes independently. By
spinning up a coupled model and keeping some transports of the equilibrium state fixed, feedbacks can be turned off at will, providing the most
rigorous way of assessing the importance of competing effects. The same
strategy can be used in a coupled GeM; there is, however, the possibility that upon a switch in degree of coupling, convection patterns do
become unconditionally unstable, as is the assumption underlying Rahmstorf's (1995a) analysis. Whether our approach is feasible for GeMs can
only be determined by applying it.
The analysis of land effects suggests the need for idealised coupled GeM
studies, in which the zonal transports are modelled explicitly rather than
parameterized. The simple model offers guidance for how to conduct these
experiments; for example, it can be analyzed how transport efficiencies depend on the size of the ocean basin with active thermohaline circulation;
in particular, whether and how a relatively small ocean like the North
Atlantic can influence the overall strength of the Jet Stream and the atmospheric eddies it produces through baroclinic instability. Again, the
results of Manabe and Stouffer (1988) indicate that, at least on very long
timescales, the North Atlantic does playa major active role.
To conclude, I firmly believe that a simple model as the one used here
can provide important insights into the dynamics of the coupled climate
system. It provides a starting point for further analyses or experimentation,
which might then corroborate or reject the ideas formulated here.
Acknowledgements
I wish to thank David Anderson and Jiirgen Willebrand for organising
this exciting School, and for giving me the opportunity to present this
material. A number of people read drafts of the manuscript and made
useful suggestions; I am particularly indebted to Stefan Rahmstorf, Jeff
transport and air temperatures, which cannot be fully understood by the
more local analysis of convective effects. An analysis of a coupled GeM
can be aided considerably by the conceptual framework developed in this
paper. The effects described here must act in the complex model as well;
if they are unimportant in the GeM, this begs for an explanation of what
processes do govern the GeM dynamics. We see the recurrent theme in
dynamical meteorology and oceanography that simple models are needed
to build up a complete understanding.
The sensitivity experiments of section 3 have made extensive use of a
general strategy to analyze various feedback processes independently. By
spinning up a coupled model and keeping some transports of the equilibrium state fixed, feedbacks can be turned off at will, providing the most
rigorous way of assessing the importance of competing effects. The same
strategy can be used in a coupled GeM; there is, however, the possibility that upon a switch in degree of coupling, convection patterns do
become unconditionally unstable, as is the assumption underlying Rahmstorf's (1995a) analysis. Whether our approach is feasible for GeMs can
only be determined by applying it.
The analysis of land effects suggests the need for idealised coupled GeM
studies, in which the zonal transports are modelled explicitly rather than
parameterized. The simple model offers guidance for how to conduct these
experiments; for example, it can be analyzed how transport efficiencies depend on the size of the ocean basin with active thermohaline circulation;
in particular, whether and how a relatively small ocean like the North
Atlantic can influence the overall strength of the Jet Stream and the atmospheric eddies it produces through baroclinic instability. Again, the
results of Manabe and Stouffer (1988) indicate that, at least on very long
timescales, the North Atlantic does playa major active role.
To conclude, I firmly believe that a simple model as the one used here
can provide important insights into the dynamics of the coupled climate
system. It provides a starting point for further analyses or experimentation,
which might then corroborate or reject the ideas formulated here.
Acknowledgements
I wish to thank David Anderson and Jiirgen Willebrand for organising
this exciting School, and for giving me the opportunity to present this
material. A number of people read drafts of the manuscript and made
useful suggestions; I am particularly indebted to Stefan Rahmstorf, Jeff