Reconstructing and Modelling the Last Glacial Maximum
689
concluded that the last interglacial was extremely
similar to the current climate, with warmer northern summers due to increased summer insolation.
There is no doubt about the great value of
CLIMAP's achievements. However, many newer
data are now available and CLIMAP members and
other authors have pointed out some problems with
the CLIMAP-LGM reconstruction. Prell (1985) indicates that the Pacific SST equations may be biased toward warm values. New data show stronger
coo lings of the lower latitudes for the SSTs, e.g.
Anderson and Webb (1994) and Guilderson et al.
(1994), as well as for the land masses, e.g. Rind
and Peteet (1985). Tushingham and Peltier (1991)
recommend changes of ice-sheet thicknesses.
According to there construction of Haupt et al.
(1994), Samthein et al. (1995), and Weinelt et al.
(1996), which is based on the SIMMAX technique
developed by Pflaumann et al. (1996), it seems that
the northern North Atlantic up to Svalbard has been
free of sea ice during summer, in contrast to the
CLIMAP reconstruction. Nevertheless, the
CLIMAP data set is still the most important and
widely distributed source of glacial SSTs which are
used in general circulation models (GCMs) to define the boundary conditions for climate simulations.
Within the International Marine Global Change
Study (IMAGES) of Past Global Changes
(PAGES) - a core project of the International
Geosphere-Biosphere Program (IGBP) - there is
an effort to collect all available new data on glacial
SSTs. This will finally lead to a new global map,
which can then be used by modellers to produce a
probably more realistic set of glacial simulations.
We feel that modelling aspects should enter at
an early stage in the preparation of a data set which
is also intended to be used to drive and to verify
climate models. From a modelling point of view
such a data set - we preliminarily call it "CLIMAP
2000" - should also contain the sea surface salinity
(SSS) which is used together with the SST to define the upper boundary conditions of ocean models. Besides the need for data to define boundary
conditions to drive models, modellers also need data
for model verification. Depending on the model type
the quantities which may be part of the CLIMAP
2000 data set can either be used for defming boundary conditions or to validate model output; see
Table 2. Of special interest for CLIMAP 2000 is
the inverse modelling approach, which has the advantage that boundary conditions need not necessarily be provided completely and that any available
data for a quantity which is modelled can be incorporated to constrain the model output.
In the following sections we describe modelling
activities at the University of Bremen to simulate
the climate ofthe LGM and relate the models used
to the suggested CLIMAP 2000. The climate system (see Fig. 1) is too complex to be simulated by
a single model. Therefore, the separate simulation
of atmosphere and ocean with three-dimensional
GCMs will be discussed in the sections Atmospheric modelling and Ocean modelling, respectively. In the section Coupled atmosphere - ocean
modelling, the problems arising from coupled modelling of ocean and atmosphere will be explained.
Models to investigate the evolution of ice-sheets are
presented in the section Ice-sheet modelling. A
model which is able to assess the distributions of
biogeochemical oceanic compounds is described in
the section Biogeochemical modelling. In the
section Inverse modelling of the ocean, a threedimensional inverse model of the Atlantic is presented and in the section Reconstruction of
vegetion zones an inverse approach which combines wind fields with pollen count data to reconstruct vegetation zones is suggested. Finally, the
various modelling approaches are discussed with
respect to their relation to CLIMAP 2000.
Modelling Climate System Components
Atmospheric Modelling
A global atmospheric model has to be provided with
the upper and lower boundary conditions at the top
and the bottom of the atmosphere, respectively
(Table 2). While the former, the insolation, can be
calculated for any time slice up to several million
years ago (Berger 1978; Berger and Loutre 1991),
the latter boundary conditions at the bottom ofthe
atmosphere (SST I sea ice, topography Iglaciers
and vegetation I albedo, cf. Table 2) have to be reconstructed from proxy data. Given the prescribed
689
concluded that the last interglacial was extremely
similar to the current climate, with warmer northern summers due to increased summer insolation.
There is no doubt about the great value of
CLIMAP's achievements. However, many newer
data are now available and CLIMAP members and
other authors have pointed out some problems with
the CLIMAP-LGM reconstruction. Prell (1985) indicates that the Pacific SST equations may be biased toward warm values. New data show stronger
coo lings of the lower latitudes for the SSTs, e.g.
Anderson and Webb (1994) and Guilderson et al.
(1994), as well as for the land masses, e.g. Rind
and Peteet (1985). Tushingham and Peltier (1991)
recommend changes of ice-sheet thicknesses.
According to there construction of Haupt et al.
(1994), Samthein et al. (1995), and Weinelt et al.
(1996), which is based on the SIMMAX technique
developed by Pflaumann et al. (1996), it seems that
the northern North Atlantic up to Svalbard has been
free of sea ice during summer, in contrast to the
CLIMAP reconstruction. Nevertheless, the
CLIMAP data set is still the most important and
widely distributed source of glacial SSTs which are
used in general circulation models (GCMs) to define the boundary conditions for climate simulations.
Within the International Marine Global Change
Study (IMAGES) of Past Global Changes
(PAGES) - a core project of the International
Geosphere-Biosphere Program (IGBP) - there is
an effort to collect all available new data on glacial
SSTs. This will finally lead to a new global map,
which can then be used by modellers to produce a
probably more realistic set of glacial simulations.
We feel that modelling aspects should enter at
an early stage in the preparation of a data set which
is also intended to be used to drive and to verify
climate models. From a modelling point of view
such a data set - we preliminarily call it "CLIMAP
2000" - should also contain the sea surface salinity
(SSS) which is used together with the SST to define the upper boundary conditions of ocean models. Besides the need for data to define boundary
conditions to drive models, modellers also need data
for model verification. Depending on the model type
the quantities which may be part of the CLIMAP
2000 data set can either be used for defming boundary conditions or to validate model output; see
Table 2. Of special interest for CLIMAP 2000 is
the inverse modelling approach, which has the advantage that boundary conditions need not necessarily be provided completely and that any available
data for a quantity which is modelled can be incorporated to constrain the model output.
In the following sections we describe modelling
activities at the University of Bremen to simulate
the climate ofthe LGM and relate the models used
to the suggested CLIMAP 2000. The climate system (see Fig. 1) is too complex to be simulated by
a single model. Therefore, the separate simulation
of atmosphere and ocean with three-dimensional
GCMs will be discussed in the sections Atmospheric modelling and Ocean modelling, respectively. In the section Coupled atmosphere - ocean
modelling, the problems arising from coupled modelling of ocean and atmosphere will be explained.
Models to investigate the evolution of ice-sheets are
presented in the section Ice-sheet modelling. A
model which is able to assess the distributions of
biogeochemical oceanic compounds is described in
the section Biogeochemical modelling. In the
section Inverse modelling of the ocean, a threedimensional inverse model of the Atlantic is presented and in the section Reconstruction of
vegetion zones an inverse approach which combines wind fields with pollen count data to reconstruct vegetation zones is suggested. Finally, the
various modelling approaches are discussed with
respect to their relation to CLIMAP 2000.
Modelling Climate System Components
Atmospheric Modelling
A global atmospheric model has to be provided with
the upper and lower boundary conditions at the top
and the bottom of the atmosphere, respectively
(Table 2). While the former, the insolation, can be
calculated for any time slice up to several million
years ago (Berger 1978; Berger and Loutre 1991),
the latter boundary conditions at the bottom ofthe
atmosphere (SST I sea ice, topography Iglaciers
and vegetation I albedo, cf. Table 2) have to be reconstructed from proxy data. Given the prescribed