GREENHOUSE GASES EFFECTS 431
from inadequate spatial resolution, problems in parameter-
izing sub-grid-scale motions, and in estimating the fluxes of
heat, moisture and momentum across the air/sea interface.
When forced with observed surface temperatures, salini-
ties and wind stresses, ocean models have been moderately
successful in simulating the observed large-scale circulation
and mass distribution, but most models underestimate the
meridional heat flux and make the thermocline too deep, dif-
fuse and too warm.
The deeper ocean is also driven, in part, by fluxes of
radiant heat, momentum, and of fresh water derived from
precipitation, river run-off and melting ice, but measure-
ments of all these are difficult and very sparse at the pres-
ent time. Different models show considerable differences in
their simulations of the deep ocean circulation, but identifi-
cation of systematic errors is hardly possible because of the
paucity of observations. The distribution of temperature and
salinity are the primary sources of information for check-
ing model simulations, but it is very difficult to simulate the
salinity field because the distribution of sources and sinks of
fresh water at the surface is so complex.
Perhaps the most effective way of checking ocean
models on decadal time scales is to see how well they simu-
late the horizontal spread and vertical diffusion of transient
tracers such as tritium/He^3 and C^14 produced in nuclear
bomb tests. Current models simulate quite well their shal-
low penetration in the equatorial ocean and deep penetration
in high latitudes but fail to reproduce the deep penetration
at 30–50N, probably because of inadequate resolution of
the Gulf Stream and its interaction with the North Atlantic
current. The computed poleward transport of heat and the
transport across other designated vertical sections can be
checked against hydrographic measurements being made
from research ships as part of the World Ocean Circulation
Experiment, as described in Mason (1993). Some detailed
measurements are also being made on the seasonal variation
in the depth of the mixed ocean layer and the thermocline
that can be compared with the model simulations.Coupled Atmosphere—Deep Ocean ModelsThe UKMO has developed a deep global ocean model coupled
to its global atmospheric model to carry out long-period cli-
mate simulations and to make realistic predictions of climate
changes produced by gradual increases of atmospheric CO 2
until it reaches double the present value. The results of the
first of these enhanced CO 2 experiments, and of similar ones
conducted elsewhere, are described in the following section.
Here we summarise the structure and operation of the cou-
pled model, its problems and deficiencies, and the research in
progress to overcome them. A more detailed analysis of the
first version is given by Murphy (1995). In the latest version,
the model atmosphere is divided into 19 layers (20 pressure
levels) between the surface and 50 km with 5 levels in the
surface boundary layer (lowest 1 km) to allow calculation of
the surface fluxes of heat, moisture and momentum. There are
also four levels in the soil to calculate the heat flux and hencethe surface temperature. The variables listed in the previous
section Simulation of the Present Climate are calculated on
a spherical grid with mesh 2.5 lat 3.75 long, about 7,000
points at each level.
The incoming solar radiation is calculated as a function
of latitude and season, and diurnal variations are included.
Calculations of radiative fluxes at each model level use four
wavebands in the solar radiation and six bands in the long-wave
infra-red, allowing for absorbtion and emission by water vapour,
carbon dioxide, ozone and clouds. Sub-grid-scale convection is
represented by a simple cloud model that treats the compensat-
ing subsidence and detrainment of air and the evaporation of
precipitation. Precipitation is calculated in terms of the water
and ice content of the cloud; cooling of the atmosphere by
evaporation of precipitation is allowed for. Reduction in wind
speed caused by the aerodynamic drag of mountains, oceans
waves, and by the breaking of ororgraphically-induced gravity
waves are computed. In calculating changes in the extent and
thickness of sea ice, drifting of the ice by wind-driven ocean
currents is taken into account.
In the land surface model the different soil types and
their differing albedos are specified, as are the different
types of vegetation, their seasonal changes and their effects
on evaporation, albedo, aerodynamic drag.
The ocean model computes the current, potential temper-
ature, salinity, density and the transports of heat and salt at
20 unequally-spaced levels (depths) in the ocean, eight of
these being in the top 120 m in order to simulate better the
physics and dynamics in the active, well-mixed layer, its sea-
sonal variation, and the surface exchanges of heat, moisture
and momentum with the atmosphere. The vertical veloc-
ity at the sea floor is computed assuming flow parallel to
the slope of the bottom topography specified on a 1 1
data set. The horizontal grid, 2.5 3.75, the same as that
of the atmospheric model, is too coarse to resolve oceanic
meso-scale eddies of scale 100 km which contain much of
the total kinetic energy, but are crudely represented by sub-
grid-scale turbulent diffusion and viscosity. The latter has to
be kept artificially high to preserve computational stability
with the penalty that the simulated currents, such as the Gulf
Stream, are too weak. Lateral diffusion of heat and salt take
place along ispycnal (constant density) surfaces using diffu-
sion coefficients that decrease exponentially with increasing
depth. The coefficients of vertical diffusion are specified as
functions of the local Richardson number, which allows for
increased mixing when the local current shear is large.
Coupling with the atmosphere is accomplished in three
stages. The atmospheric model, starting from an initial state
based on observations, is run on its own until it reaches an
equilibrium climate. The ocean model, starting from rest
and uniform temperature and salinity is also run separately,
driven by the wind stresses, heat and fresh-water fluxes pro-
vided by the atmospheric model. This spin-up phase of the
ocean takes place over 150 years (restricted by available
computer time) during which a steady state is achieved in
the upper layers of the ocean as they come into equilibrium
with the atmospheric forcing. Finally, the ocean is coupled
to the atmosphere, sea-ice and land-surface componentsC007_002_r03.indd 431C007_002_r03.indd 431 11/18/2005 10:28:00 AM11/18/2005 10:28:00 AM