On North Atlantic Intedecadal Variability: A Stochastic View
155
variability, such as the tropical and North Pacific regions. Latifand Bamett (1994),
for instance, postulate the existence of an interdecadal cycle in the North Pacific,
and Gu and Philander (1997) put forward the idea of a coupled tropical PacificNorth Pacific interdecadal cyc1e. It is beyond the scope ofthis paper to address aU
the different aspects of interdecadal variability. The reader is referred to the observational papers of e.g. Folland et al. (1986), Dickson et al. (1988), Mysak et al.
(1990), Deser and Blackmon (1993), Kushnir (1994), Trenberth and Hurrell
(1994), Mann and Park (1994), Levitus and Antonov (1995), Zhang et al. (1997),
Mantua et al. (1997) and references therein for further infonnation on the observational aspects of interdecadal variability. A fairly comprehensive overview of the
different aspects of interdecadal variability (including theoretical, modeling, and
observational aspects) can be found in the recent book published by Anderson and
Willebrand (1996), while an overview of the interdecadal variability simulated in
coupled ocean-atmosphere models is given in the review artic1e by Latif(1998).
The paper is organized as follows. We describe briefly the stochastic c1imate
model concept and show some observational results supporting it in section 9.2.
Section 9.3 deals with the quasi-decadal variability that arises from variations in
the North Atlantic subtropical gyre circulation, as simulated by our coupled oceanatmosphere model. We present the interdecadal variability associated with the
North Atlantic thennohaline circulation ofthe same model in section 9.4. The predictability of interdecadal changes, as derived by c1assical predictability experiments, is addressed is section 9.5. The paper is conc1uded with a discussion in
section 9.6.
9.2 The stochastic cUmate model
In contrast to the atmosphere, which has variations with typical time scales of a
few days, the ocean is much more inert. Anomalies of the sea surface temperature
(SST), for instance, persist typically for several months, while the characteristic
time scale for the deep ocean is ofthe order ofthousand years. Many aspects ofthe
interactions between climate-subsystems with considerably different time scales
can be described analogous to the Brownian motion in statistical physics. The picture is that oceanic anomalies (such as SST anomalies) re suit from the integration
of many statistically independent atmospheric events, e.g. the passage of high and
low pressure systems. One may view the SST variability, for instance, as the re suit
of the integration of the surf ace heat flux anomalies within the oceanic mixed layer.
This kind of interaction was introduced by Hasselmann (1976) and is referred to as
the 'stochastic c1imate model'. The stochastic climate model scenario is a relatively
simple concept which may be regarded as a kind of 'null hypothesis' for the generation of climate variability in general and decadal variability in particular.
The time evolution of a typical oceanic quantity y (such as SST) may be
described by a Langevinequation
155
variability, such as the tropical and North Pacific regions. Latifand Bamett (1994),
for instance, postulate the existence of an interdecadal cycle in the North Pacific,
and Gu and Philander (1997) put forward the idea of a coupled tropical PacificNorth Pacific interdecadal cyc1e. It is beyond the scope ofthis paper to address aU
the different aspects of interdecadal variability. The reader is referred to the observational papers of e.g. Folland et al. (1986), Dickson et al. (1988), Mysak et al.
(1990), Deser and Blackmon (1993), Kushnir (1994), Trenberth and Hurrell
(1994), Mann and Park (1994), Levitus and Antonov (1995), Zhang et al. (1997),
Mantua et al. (1997) and references therein for further infonnation on the observational aspects of interdecadal variability. A fairly comprehensive overview of the
different aspects of interdecadal variability (including theoretical, modeling, and
observational aspects) can be found in the recent book published by Anderson and
Willebrand (1996), while an overview of the interdecadal variability simulated in
coupled ocean-atmosphere models is given in the review artic1e by Latif(1998).
The paper is organized as follows. We describe briefly the stochastic c1imate
model concept and show some observational results supporting it in section 9.2.
Section 9.3 deals with the quasi-decadal variability that arises from variations in
the North Atlantic subtropical gyre circulation, as simulated by our coupled oceanatmosphere model. We present the interdecadal variability associated with the
North Atlantic thennohaline circulation ofthe same model in section 9.4. The predictability of interdecadal changes, as derived by c1assical predictability experiments, is addressed is section 9.5. The paper is conc1uded with a discussion in
section 9.6.
9.2 The stochastic cUmate model
In contrast to the atmosphere, which has variations with typical time scales of a
few days, the ocean is much more inert. Anomalies of the sea surface temperature
(SST), for instance, persist typically for several months, while the characteristic
time scale for the deep ocean is ofthe order ofthousand years. Many aspects ofthe
interactions between climate-subsystems with considerably different time scales
can be described analogous to the Brownian motion in statistical physics. The picture is that oceanic anomalies (such as SST anomalies) re suit from the integration
of many statistically independent atmospheric events, e.g. the passage of high and
low pressure systems. One may view the SST variability, for instance, as the re suit
of the integration of the surf ace heat flux anomalies within the oceanic mixed layer.
This kind of interaction was introduced by Hasselmann (1976) and is referred to as
the 'stochastic c1imate model'. The stochastic climate model scenario is a relatively
simple concept which may be regarded as a kind of 'null hypothesis' for the generation of climate variability in general and decadal variability in particular.
The time evolution of a typical oceanic quantity y (such as SST) may be
described by a Langevinequation