synthesizing diverse ocean observations scattered
in space and time into a dynamically consistent
picture, can be tackled successfully only on the
basis of models that faithfully represent the essential mechanisms of the circulation and its response
to atmospheric flux variations.
It is beyond the scope of this chapter to give a
comprehensive discussion of the host of issues
important in ocean model development or to
attempt an assessment of the state of ocean modelling of relevance for the spectrum of applications.
We feel it important, however, to reflect upon the
motivations and implications of one of the most
basic modelling choices – integration time – and
the associated distinction between solutions in
equilibrium and non-equilibrium with the thermohaline forcing (Section 2.2.3). Our main interest
here is on the dynamical aspects of the wind- and
thermohaline-driven circulation in high-resolution
model studies, i.e. on the forced and spontaneous
variations obtained under realistic forcing, in comparison to actual (WOCE) observations. We feel it
particularly instructive to contrast the behaviour
and simulation capability of present ocean circulation models in different dynamical categories:
variability in response to wind forcing on intraseasonal to interannual time scales (Section 2.2.4.1);
eddy variability (Section 2.2.4.2); and aspects of
the decadal-scale response to variations in highlatitude buoyancy forcing (Section 2.2.4.3). Since
other chapters in this volume specifically deal with
dynamical and modelling issues relating to the
Southern (Rintoul et al., Chapter 4.6) and Tropical (Godfrey et al., Chapter 4.3; Liu and Philander, Chapter 4.4) Oceans, the discussion here will
mainly draw from examples from the northern
mid-latitude oceans.
Discussions of many questions of model formulation, including numerical model concepts,
sub-grid-scale parameterization and the forcing of
ocean circulation models, are covered in Chassignet
and Verron (1998). A review of developments and
the quasi-equilibrium solution behaviour of global
ocean models has been given by McWilliams
(1996, 1998).
2.2.2 Historical perspective
The development of numerical models able to
describe faithfully the dynamical behaviour of the
complex, non-linear and turbulent system of the
ocean’s general circulation inextricably faces a
number of obstacles. One reason why the development of realistic ocean models has considerably
lagged behind that of their atmospheric counterparts has certainly been the societal need for
weather prediction, which focused attention on the
latter. However, a key impediment in ocean model
development has also been in the small-scale
nature of many of the essential features of the general ocean circulation. The demands for both computing power and storage capacity from including
energetic, Rossby radius-scale features in ocean
models exceed the demands of comparable atmospheric models by two orders of magnitude. This is
compounded by the need for much longer integration periods due to the slow adjustment of deep
ocean properties in response to surface fluxes.
As demonstrated with the first near-equilibrium
solution for a three-dimensional model with wind
and thermal forcing, obtained for an idealized box
ocean by Bryan and Cox (1968), the adjustment of
the deep density fields to the imposed surface flux
conditions is governed by the long time scales
associated with the weak vertical diffusion and
weak advective flushing of the deep ocean basins.
Hence, due to the need of extremely long integration times, modelling aimed at the goal of simulating the global circulation in equilibrium with the
surface fluxes has to compromise in the spatial,
particularly horizontal, resolution, and hence in
the representation of the actual structure of ocean
current features.
The development of three-dimensional models
in which both density and velocity are predicted
simultaneously received its primary incentive from
the need to include an interactive dynamical ocean
component into climate simulation (see also Wood
and Bryan, Chapter 2.3). Incorporation of oceanic
feedbacks in model simulations of climate change
had gone through a number of stages. Early
attempts to study the CO 2 -induced climate change
by the use of a general circulation model of the
atmosphere included the ocean merely as a saturated surface interacting with the atmosphere (e.g.
Manabe and Wetherald, 1975). In a next step, the
ocean’s capacity for seasonal heat storage was
taken into account, by including a shallow mixedlayer ocean with constant thickness of about 50 m
(Manabe and Stouffer, 1979, 1980; Washington
and Meehl, 1984). Up to this stage the discussion
of climate change focused upon the difference
SECTION 2 OBSERVATIONS AND MODELS
60
in space and time into a dynamically consistent
picture, can be tackled successfully only on the
basis of models that faithfully represent the essential mechanisms of the circulation and its response
to atmospheric flux variations.
It is beyond the scope of this chapter to give a
comprehensive discussion of the host of issues
important in ocean model development or to
attempt an assessment of the state of ocean modelling of relevance for the spectrum of applications.
We feel it important, however, to reflect upon the
motivations and implications of one of the most
basic modelling choices – integration time – and
the associated distinction between solutions in
equilibrium and non-equilibrium with the thermohaline forcing (Section 2.2.3). Our main interest
here is on the dynamical aspects of the wind- and
thermohaline-driven circulation in high-resolution
model studies, i.e. on the forced and spontaneous
variations obtained under realistic forcing, in comparison to actual (WOCE) observations. We feel it
particularly instructive to contrast the behaviour
and simulation capability of present ocean circulation models in different dynamical categories:
variability in response to wind forcing on intraseasonal to interannual time scales (Section 2.2.4.1);
eddy variability (Section 2.2.4.2); and aspects of
the decadal-scale response to variations in highlatitude buoyancy forcing (Section 2.2.4.3). Since
other chapters in this volume specifically deal with
dynamical and modelling issues relating to the
Southern (Rintoul et al., Chapter 4.6) and Tropical (Godfrey et al., Chapter 4.3; Liu and Philander, Chapter 4.4) Oceans, the discussion here will
mainly draw from examples from the northern
mid-latitude oceans.
Discussions of many questions of model formulation, including numerical model concepts,
sub-grid-scale parameterization and the forcing of
ocean circulation models, are covered in Chassignet
and Verron (1998). A review of developments and
the quasi-equilibrium solution behaviour of global
ocean models has been given by McWilliams
(1996, 1998).
2.2.2 Historical perspective
The development of numerical models able to
describe faithfully the dynamical behaviour of the
complex, non-linear and turbulent system of the
ocean’s general circulation inextricably faces a
number of obstacles. One reason why the development of realistic ocean models has considerably
lagged behind that of their atmospheric counterparts has certainly been the societal need for
weather prediction, which focused attention on the
latter. However, a key impediment in ocean model
development has also been in the small-scale
nature of many of the essential features of the general ocean circulation. The demands for both computing power and storage capacity from including
energetic, Rossby radius-scale features in ocean
models exceed the demands of comparable atmospheric models by two orders of magnitude. This is
compounded by the need for much longer integration periods due to the slow adjustment of deep
ocean properties in response to surface fluxes.
As demonstrated with the first near-equilibrium
solution for a three-dimensional model with wind
and thermal forcing, obtained for an idealized box
ocean by Bryan and Cox (1968), the adjustment of
the deep density fields to the imposed surface flux
conditions is governed by the long time scales
associated with the weak vertical diffusion and
weak advective flushing of the deep ocean basins.
Hence, due to the need of extremely long integration times, modelling aimed at the goal of simulating the global circulation in equilibrium with the
surface fluxes has to compromise in the spatial,
particularly horizontal, resolution, and hence in
the representation of the actual structure of ocean
current features.
The development of three-dimensional models
in which both density and velocity are predicted
simultaneously received its primary incentive from
the need to include an interactive dynamical ocean
component into climate simulation (see also Wood
and Bryan, Chapter 2.3). Incorporation of oceanic
feedbacks in model simulations of climate change
had gone through a number of stages. Early
attempts to study the CO 2 -induced climate change
by the use of a general circulation model of the
atmosphere included the ocean merely as a saturated surface interacting with the atmosphere (e.g.
Manabe and Wetherald, 1975). In a next step, the
ocean’s capacity for seasonal heat storage was
taken into account, by including a shallow mixedlayer ocean with constant thickness of about 50 m
(Manabe and Stouffer, 1979, 1980; Washington
and Meehl, 1984). Up to this stage the discussion
of climate change focused upon the difference
SECTION 2 OBSERVATIONS AND MODELS
60