48
Chapter 3: Climate Spectra and Stochastic Climate Models
3.6 Variability in the Ocean Interior
Few long time series are available in the ocean interior, so that the spectral
characteristics are virtually unknown below the surface at periods longer than
a few years. The longest series of hydrographie stations were made at station
"S" off Bermuda (32 0 N, 64 0 W). The temperature spectra are dominated by
mesoscale eddies because of the proximity of the Gulf Stream, and become
red at very low frequencies (FrankignouI1981). Temperature and salinity on
the interannual time scales are largely independent in the upper layers, and
highly correlated in the thermocline, as expected from vertical advection; a
long-term trend is also seen in the deep layer (Joyce, 1994). It is not known
whether these fluctuations are due to the internal dynamics of the ocean or
are forced by the atmosphere.
On time scales of up to a few years, stochastic atmospheric forcing primarily affects the ocean interior via Ekman pumping. The oceanic response is
dominated by resonant Rossby waves, with an energy input rate that may
be sufficient to maintain the observed eddy field in regions of weak eddy
activity (Frankignoul and Müller, 1979a). The predicted response spectra
are sensitive to friction and other dynamical effects, however, so that the
signatures from the interaction are best seen in the seasonal modulation
and the coherence between oceanic and atmospheric variables (Müller and
Frankignoul, 1981; Brink, 1989). So far, the observations have confirmed the
importance of the stochastic forcing at high frequencies: for periods of up
to a few months, much of the deep ocean subinertial variability is consistent with the barotropic response to stochastic atmospheric forcing, except
near the western boundary currents and their large mid-latitude open-ocean
extension, where instability dominates and the energy level is much higher.
Using observed SST anomaly spectra to specify the buoyancy forcing by
short time scale atmospheric fluctuations, Frankignoul and Müller (1979b)
found negligible effects on time scales of up to 10 2 years. However, this neglects the fresh water flux contribution to the surface buoyancy flux, which
may become dominant on the decadal time scale since surface salinity anomalies may have a longer lifetime than the SST anomalies. Mikolajewicz and
Maier-Reimer (1990) have driven aglobai oceanic GCM with stochastic perturbations in the freshwater flux, using the simplifying so-called "mixed
boundary conditions" (strong SST relaxation onto a prescribed reference temperature, no salinity feedback). The GCM primarily acted as an integrator
to the white-noise forcing, and a decadal mode was identified by Weisse et al.
(1994) in the upper level salinity field of the Labrador sea and the northern
North Atlantic (Figure 3.12). At low frequencies, the internal dynamics of
the ocean model determined the response, causing strong and irregular oscillations and a pronounced 320-year spectral peak weIl above the /-2 level
(Figure 3.13). This variability primarily appeared in the form of an inter mittent "eigenmode" of oscillation of the Atlantic thermohaline overturning cell
which connects the polar regions, involving a positive feedback in deep-water
Chapter 3: Climate Spectra and Stochastic Climate Models
3.6 Variability in the Ocean Interior
Few long time series are available in the ocean interior, so that the spectral
characteristics are virtually unknown below the surface at periods longer than
a few years. The longest series of hydrographie stations were made at station
"S" off Bermuda (32 0 N, 64 0 W). The temperature spectra are dominated by
mesoscale eddies because of the proximity of the Gulf Stream, and become
red at very low frequencies (FrankignouI1981). Temperature and salinity on
the interannual time scales are largely independent in the upper layers, and
highly correlated in the thermocline, as expected from vertical advection; a
long-term trend is also seen in the deep layer (Joyce, 1994). It is not known
whether these fluctuations are due to the internal dynamics of the ocean or
are forced by the atmosphere.
On time scales of up to a few years, stochastic atmospheric forcing primarily affects the ocean interior via Ekman pumping. The oceanic response is
dominated by resonant Rossby waves, with an energy input rate that may
be sufficient to maintain the observed eddy field in regions of weak eddy
activity (Frankignoul and Müller, 1979a). The predicted response spectra
are sensitive to friction and other dynamical effects, however, so that the
signatures from the interaction are best seen in the seasonal modulation
and the coherence between oceanic and atmospheric variables (Müller and
Frankignoul, 1981; Brink, 1989). So far, the observations have confirmed the
importance of the stochastic forcing at high frequencies: for periods of up
to a few months, much of the deep ocean subinertial variability is consistent with the barotropic response to stochastic atmospheric forcing, except
near the western boundary currents and their large mid-latitude open-ocean
extension, where instability dominates and the energy level is much higher.
Using observed SST anomaly spectra to specify the buoyancy forcing by
short time scale atmospheric fluctuations, Frankignoul and Müller (1979b)
found negligible effects on time scales of up to 10 2 years. However, this neglects the fresh water flux contribution to the surface buoyancy flux, which
may become dominant on the decadal time scale since surface salinity anomalies may have a longer lifetime than the SST anomalies. Mikolajewicz and
Maier-Reimer (1990) have driven aglobai oceanic GCM with stochastic perturbations in the freshwater flux, using the simplifying so-called "mixed
boundary conditions" (strong SST relaxation onto a prescribed reference temperature, no salinity feedback). The GCM primarily acted as an integrator
to the white-noise forcing, and a decadal mode was identified by Weisse et al.
(1994) in the upper level salinity field of the Labrador sea and the northern
North Atlantic (Figure 3.12). At low frequencies, the internal dynamics of
the ocean model determined the response, causing strong and irregular oscillations and a pronounced 320-year spectral peak weIl above the /-2 level
(Figure 3.13). This variability primarily appeared in the form of an inter mittent "eigenmode" of oscillation of the Atlantic thermohaline overturning cell
which connects the polar regions, involving a positive feedback in deep-water