363
moisture transport might cause a significant variation in ocean freshwater
forcing. Clearly, the simple model has reached the limit of its usefulness
here, in particular since we consider a single basin only and do not include
water exchange between the catchment areas of different oceans. The need
for a careful examination with idealised 3-dimensional oceanic and atmospheric models is evident.
5 Flux adjustments and climate drifts
NSM and MS have demonstrated that erroneous transports in either subsystem lead to incorrect climate sensitivity and stability in a coupled model
as used here, even when flux adjustments are applied to simulate the current equilibrium climate. Complementary discussions of the dynamical
consequences of flux adjustments in coupled models have been given in
Neelin and Dijkstra (1995) and Schneider (1996). Here, the focus is on
the relationship between flux adjustments and - applied flux adjustments
notwithstanding - drifts of the simulated 'current' climate, which have occurred in the models of the United Kingdom Meteorological Office (UKMO,
Murphy, 1995) and the Max Planck Institute for Meteorology (MPI) in
Hamburg (Santer et al., 1994).
In both the UKMO and the MPI models the ocean part was spun up
separately before coupling. The UKMO ocean model was run for 150 years
using restoring boundary conditions on both T and S, with a relaxation
timescale of 15 days. The MPI ocean model was run for 5000 years using
restoring boundary conditions on both T and S; then the implied surface
freshwater fluxes were diagnosed and used as boundary conditions for another 2000 years (Cubasch et al., 1992). 75 years after coupling, the UKMO
model showed pronounced warming in the upper 500 m of the Antarctic
Circumpolar Current (ACC) region, of typically 1° to 2°C at the sea surface; below 500 m cooling occurred. Simultaneously, salinity increased in
the upper 500 m and decreases below, with a resulting increase in convection and meridional overturning (from 5 Sv to 15 Sv). The MPI model
SST showed, 100 years after coupling, cooling in the Arctic of 5°C and
warming in the Ross Sea of 6°C. The drift had added to it an oscillatory
component in the Northern Hemisphere (Santer et al., 1994).
Only speculations concerning the origin of the drift were given in the
cited papers. However, Santer et al. (1994) note that the drift was largest
where the flux adjustment fields were largest as well. Murphy (1995) con-
moisture transport might cause a significant variation in ocean freshwater
forcing. Clearly, the simple model has reached the limit of its usefulness
here, in particular since we consider a single basin only and do not include
water exchange between the catchment areas of different oceans. The need
for a careful examination with idealised 3-dimensional oceanic and atmospheric models is evident.
5 Flux adjustments and climate drifts
NSM and MS have demonstrated that erroneous transports in either subsystem lead to incorrect climate sensitivity and stability in a coupled model
as used here, even when flux adjustments are applied to simulate the current equilibrium climate. Complementary discussions of the dynamical
consequences of flux adjustments in coupled models have been given in
Neelin and Dijkstra (1995) and Schneider (1996). Here, the focus is on
the relationship between flux adjustments and - applied flux adjustments
notwithstanding - drifts of the simulated 'current' climate, which have occurred in the models of the United Kingdom Meteorological Office (UKMO,
Murphy, 1995) and the Max Planck Institute for Meteorology (MPI) in
Hamburg (Santer et al., 1994).
In both the UKMO and the MPI models the ocean part was spun up
separately before coupling. The UKMO ocean model was run for 150 years
using restoring boundary conditions on both T and S, with a relaxation
timescale of 15 days. The MPI ocean model was run for 5000 years using
restoring boundary conditions on both T and S; then the implied surface
freshwater fluxes were diagnosed and used as boundary conditions for another 2000 years (Cubasch et al., 1992). 75 years after coupling, the UKMO
model showed pronounced warming in the upper 500 m of the Antarctic
Circumpolar Current (ACC) region, of typically 1° to 2°C at the sea surface; below 500 m cooling occurred. Simultaneously, salinity increased in
the upper 500 m and decreases below, with a resulting increase in convection and meridional overturning (from 5 Sv to 15 Sv). The MPI model
SST showed, 100 years after coupling, cooling in the Arctic of 5°C and
warming in the Ross Sea of 6°C. The drift had added to it an oscillatory
component in the Northern Hemisphere (Santer et al., 1994).
Only speculations concerning the origin of the drift were given in the
cited papers. However, Santer et al. (1994) note that the drift was largest
where the flux adjustment fields were largest as well. Murphy (1995) con-