160 Mojib Latif, Axei Timmennann, Anselm Gr6tzner, Christian Eckert, Reinhard Voss
and Reynolds (1983), Sutton and Allen (1997), Saravanan and McWilliams (1997),
and Weisse et al. (1997). The latter attribute the dynamics ofthe Antarctic Circumpolar Wave (ACW, White and Peterson (1996)) to such a stochastic climate model
with advection, while Sutton and Allen (1997) and Saravanan and McWilliams
(1997) explain part ofthe North Atlantic low-frequency variability by this mechamsm.
In summary, the stochastic forcing of the climate system by the noise inherent to
it is an important mechanism in generating climate variability. The shape of the
spectra derived from observations (or model simulations) may provide important
indications for the underlying dynamics leading to the variability considered. If the
variability is consistent with the simple vers ion (1) of the stochastic climate model,
one may as sume that ocean dynamics are not important, and that the variability
arises from the integration of the atmospheric weather noise by the oceanic mixed
layer. If (statistically significant) peaks are found, this points generally towards a
prominent role of the ocean dynamics in producing the variability at decadal time
scales. We shall show two such examples below which were both derived from a
multi-century simulation with a coupled ocean-atmosphere general circulation
model.
9.3 Quasi-decadal variability in the North Atlantic
We turn now to the integration with our coupled ocean-atmosphere general circulation model (ECHAM3/LSG). Details of the coupled model and the integration
can be found in Voss et al. (1998), Timmermann (1996) and Timmermann et al.
(1998). The coupled model was integrated for 2000 years, and the analyses shown
below were performed using annualy averaged values. The coupled model simulates reasonably well the low-frequency variability in the North Atlantic, which is
shown by the comparison of the standard deviations of annually averaged SSTs as
computed from observations and the coupled model simulation (Fig. 9.8).
One can separate conceptually the ocean circulation in the North Atlantic into a
'wind-driven' and a 'thermohaline' part. The former is forced by the surface wind
stress and associated with horizontal circulations in the upper ocean (subtropical
and subpolar gyres), while the latter is associated with a meridional (north-south)
circulation and deep convection in the northem North Atlantic. We shall show that
variations in both types of circulation systems can lead to interdecadal variability.
Since the adjustment times of the gyre circulations are considerably shorter than
that ofthe thermohaline circulation, variations associated with the gyres have typical time scales of about 10-20 years (quasi-decadal), while those linked to the thermohaline circulation have typical time scales of several decades (interdecadal).
Interestingly, both types ofvariability seem to co-exist in the coupled model simulation, as shown below, and there are some indications from observations that this
is the case in the real world too (e.g. Deser and Blackmon (1993)).
and Reynolds (1983), Sutton and Allen (1997), Saravanan and McWilliams (1997),
and Weisse et al. (1997). The latter attribute the dynamics ofthe Antarctic Circumpolar Wave (ACW, White and Peterson (1996)) to such a stochastic climate model
with advection, while Sutton and Allen (1997) and Saravanan and McWilliams
(1997) explain part ofthe North Atlantic low-frequency variability by this mechamsm.
In summary, the stochastic forcing of the climate system by the noise inherent to
it is an important mechanism in generating climate variability. The shape of the
spectra derived from observations (or model simulations) may provide important
indications for the underlying dynamics leading to the variability considered. If the
variability is consistent with the simple vers ion (1) of the stochastic climate model,
one may as sume that ocean dynamics are not important, and that the variability
arises from the integration of the atmospheric weather noise by the oceanic mixed
layer. If (statistically significant) peaks are found, this points generally towards a
prominent role of the ocean dynamics in producing the variability at decadal time
scales. We shall show two such examples below which were both derived from a
multi-century simulation with a coupled ocean-atmosphere general circulation
model.
9.3 Quasi-decadal variability in the North Atlantic
We turn now to the integration with our coupled ocean-atmosphere general circulation model (ECHAM3/LSG). Details of the coupled model and the integration
can be found in Voss et al. (1998), Timmermann (1996) and Timmermann et al.
(1998). The coupled model was integrated for 2000 years, and the analyses shown
below were performed using annualy averaged values. The coupled model simulates reasonably well the low-frequency variability in the North Atlantic, which is
shown by the comparison of the standard deviations of annually averaged SSTs as
computed from observations and the coupled model simulation (Fig. 9.8).
One can separate conceptually the ocean circulation in the North Atlantic into a
'wind-driven' and a 'thermohaline' part. The former is forced by the surface wind
stress and associated with horizontal circulations in the upper ocean (subtropical
and subpolar gyres), while the latter is associated with a meridional (north-south)
circulation and deep convection in the northem North Atlantic. We shall show that
variations in both types of circulation systems can lead to interdecadal variability.
Since the adjustment times of the gyre circulations are considerably shorter than
that ofthe thermohaline circulation, variations associated with the gyres have typical time scales of about 10-20 years (quasi-decadal), while those linked to the thermohaline circulation have typical time scales of several decades (interdecadal).
Interestingly, both types ofvariability seem to co-exist in the coupled model simulation, as shown below, and there are some indications from observations that this
is the case in the real world too (e.g. Deser and Blackmon (1993)).