700
Herterich et al.
gen.erated during previous synchronously coupled
penods (Fig. 7). Figs. 8 and 9 show results from a
first long-term experiment with the coupled model
under modem boundary conditions. This run has
been performed without flux adjustments over a
simulation period of 1600 years with synchronously
coupled periods of 15 months. During the first 230
years of simulation time the ocean-only phases
were successively extended, starting with a length
of3 years at the beginning ofthe simulation, up to
a length of 50 years. The time series of ocean layer
temperatures (Fig. 8) mark a strong drift of the
model towards a cooler climate during the first 100
years of coupling, which is mainly the result ofthe
different representation of surface fluxes in the
submodels. The small peaks of the upper layer temperatures show the onsets of the synchronously
coupled phases, when the atmosphere responds to
the new distribution of SSTs and sea ice, which
were calculated during the former ocean-only
phase without feedback from the atmosphere
model. The deep ocean temperatures and salinities
(Fig. 8) indicate that an equilibrium ocean circulation has not yet been reached. The meridional
stream function of the ocean model spin-up run
(Fig. 9, top) shows a reasonable circulation with the
formation of North Atlantic Deep Water and export into the Southern Ocean. The long-term integration (over 1600 years) of the coupled model
without flux adjustments exhibits also a conveyortype thermohaline circulation, with a weaker formation ofNADW but lacking any AABW (Fig. 9,
bottom).
This coupled model is now fast enough to perform a complete series of experiments with an integration time of several thousand model years for
each run. The objective is to determine the required
minimum flux adjustment which might be unavoidable to prevent this unrealistic climate drift. If an
adjustment can be found which depends on systematic deficiencies of the model components but is
mostly independent of the simulated climate, then
the model is able to simulate paleoclimate states like
the LGM which may strongly deviate from the
modem one. The resultant global data sets at the
atmosphere-ocean interface would be a valuable
contribution to CLIMAP 2000.
Ice-Sheet Modelling
The continental ice-sheets represent the third largest component of the climate system. The aim of
ice-sheet modelling is to understand their physical
behaviour and interaction with atmosphere ocean
lithosphere and other subsystems (Hans:n et al:
1996). The ice preserves information about
paleoclimatic conditions, and its volume is one of
the most important climate indicators.
The three-dimensional ice-sheet model
SICOPOLIS takes into account the difference between cold ice with a temperature below the pressure melting point and less viscous temperate ice
having a temperature exactly on the pressure melt~
ing point and containing small amounts of liquid
water (Greve 1995). Mountain glaciers and some
parts of Greenland and Antarctica near the bottom
are observed to be temperate. Difficulties arise as
the applied "shallow ice approximation" is not
really valid at domes and coastal margins, where
the mass loss by ice calving happens, and an iceshelf model is not yet included. Another problem
remains concerning the flow law which does not
take into account the anisotropic structure of ice
crystals. This may have an evident effect on the
deformation and ice flow.
The evolution ofthe extent and height of an icesheet strongly depends on its boundary conditions
(Table 2). As the measured data basis on Antarctica will remain sparse, output fields of these quantities resulting from atmospheric models are valuable. Studies with varied geothermal heat flux reveal the ice volume reacting very sensitively to the
heating from below (Hansen and Greve 1995).
Currently, this quantity has to be set to a constant
value (in space and time) due to a lack of data.
A comparison between the observed modem
state and the result of a steady state model run for
the Antarctic ice-sheet is shown in Fig. 10. Although
the model reproduces features as e.g. the positions
of domes quite well, improvements by a more detailed and accurate snow input field would be desirable. Here, precipitation output ofthe ECHAM
3 (DKRZ Modellbetreuungsgruppe 1994) serves
as atmospheric forcing. Obviously the modelled icesheet is too high, which may be a consequence of
the inaccurate boundary conditions. But it could also
Herterich et al.
gen.erated during previous synchronously coupled
penods (Fig. 7). Figs. 8 and 9 show results from a
first long-term experiment with the coupled model
under modem boundary conditions. This run has
been performed without flux adjustments over a
simulation period of 1600 years with synchronously
coupled periods of 15 months. During the first 230
years of simulation time the ocean-only phases
were successively extended, starting with a length
of3 years at the beginning ofthe simulation, up to
a length of 50 years. The time series of ocean layer
temperatures (Fig. 8) mark a strong drift of the
model towards a cooler climate during the first 100
years of coupling, which is mainly the result ofthe
different representation of surface fluxes in the
submodels. The small peaks of the upper layer temperatures show the onsets of the synchronously
coupled phases, when the atmosphere responds to
the new distribution of SSTs and sea ice, which
were calculated during the former ocean-only
phase without feedback from the atmosphere
model. The deep ocean temperatures and salinities
(Fig. 8) indicate that an equilibrium ocean circulation has not yet been reached. The meridional
stream function of the ocean model spin-up run
(Fig. 9, top) shows a reasonable circulation with the
formation of North Atlantic Deep Water and export into the Southern Ocean. The long-term integration (over 1600 years) of the coupled model
without flux adjustments exhibits also a conveyortype thermohaline circulation, with a weaker formation ofNADW but lacking any AABW (Fig. 9,
bottom).
This coupled model is now fast enough to perform a complete series of experiments with an integration time of several thousand model years for
each run. The objective is to determine the required
minimum flux adjustment which might be unavoidable to prevent this unrealistic climate drift. If an
adjustment can be found which depends on systematic deficiencies of the model components but is
mostly independent of the simulated climate, then
the model is able to simulate paleoclimate states like
the LGM which may strongly deviate from the
modem one. The resultant global data sets at the
atmosphere-ocean interface would be a valuable
contribution to CLIMAP 2000.
Ice-Sheet Modelling
The continental ice-sheets represent the third largest component of the climate system. The aim of
ice-sheet modelling is to understand their physical
behaviour and interaction with atmosphere ocean
lithosphere and other subsystems (Hans:n et al:
1996). The ice preserves information about
paleoclimatic conditions, and its volume is one of
the most important climate indicators.
The three-dimensional ice-sheet model
SICOPOLIS takes into account the difference between cold ice with a temperature below the pressure melting point and less viscous temperate ice
having a temperature exactly on the pressure melt~
ing point and containing small amounts of liquid
water (Greve 1995). Mountain glaciers and some
parts of Greenland and Antarctica near the bottom
are observed to be temperate. Difficulties arise as
the applied "shallow ice approximation" is not
really valid at domes and coastal margins, where
the mass loss by ice calving happens, and an iceshelf model is not yet included. Another problem
remains concerning the flow law which does not
take into account the anisotropic structure of ice
crystals. This may have an evident effect on the
deformation and ice flow.
The evolution ofthe extent and height of an icesheet strongly depends on its boundary conditions
(Table 2). As the measured data basis on Antarctica will remain sparse, output fields of these quantities resulting from atmospheric models are valuable. Studies with varied geothermal heat flux reveal the ice volume reacting very sensitively to the
heating from below (Hansen and Greve 1995).
Currently, this quantity has to be set to a constant
value (in space and time) due to a lack of data.
A comparison between the observed modem
state and the result of a steady state model run for
the Antarctic ice-sheet is shown in Fig. 10. Although
the model reproduces features as e.g. the positions
of domes quite well, improvements by a more detailed and accurate snow input field would be desirable. Here, precipitation output ofthe ECHAM
3 (DKRZ Modellbetreuungsgruppe 1994) serves
as atmospheric forcing. Obviously the modelled icesheet is too high, which may be a consequence of
the inaccurate boundary conditions. But it could also