166 Mojib Latif, Axei Timmermann, Anselm Grotzner, Christian Eckert, Reinhard Voss
ocean and the atmosphere is simulated at a period of about 35 years. We shall refer
this type of interdecadal variability to as the 'interdecadal mode'. Since enhanced
variability is simulated in both media at the 35-year time scale, our working
hypothesis is (as for the quasi-decadal variability) that the variability arises from
ocean-atmosphere interactions (see Timmermann (1996) and Timmermann et al.
(1998) for further details).
The 35-year period is seen nicely in the spectrum ofthe overtuming index (Fig.
9.10b). We used therefore the index of the meridional overtuming (Fig. 9.9c) to
describe the evolution of the interdecadal mode by computing lagged regres sion
patterns of selected quantities (SLP, SST, fresh water flux, SSS, surface currents,
and convection). In order to highlight the interdecadal variability, we band-pass filtered the model data prior to the regression analyses retaining variability with time
scales of 25-45 years. The anomalous SST pattern ten years prior to the maximum
overtuming (lag - 10 years) is characterized by negative anomalies which cover the
entire North Atlantic Ocean, with maximum cold anomalies near 40
0
N (Fig.
9.l4b). The associated SLP anomaly pattern is the North Atlantic Oscillation
(NAO) in its reverse polarity (Fig. 9.14a). Important to the subsequent phase reversaI are the positive salinity anomalies which develop primarily east of Newfoundland (Fig. 9.l4d). As shown in Timmermann et al. (1998), this positive SSS
anomaly is created by anomalous fresh water input (primarily through enhanced
evaporation, Fig. 9 .14c) and anomalous salt transport by anomalous Ekman currents (Fig. 9.14e). The convection at this time is near normal (Fig. 9.l4t). The SSS
anomaly has grown and expanded in area five years later (lag -5 years), as shown in
Fig. 9.l5d. At this time, the SLP and SST anomaly patterns are not well developed
and in a kind of 'transition stage' (Figs. 9 .15a and 9 .15b). The convection, however,
shows a clear intensification in the sinking region south of Greenland (Fig. 9.15t),
in response to the enhanced density which is due to the increased sea surf ace salinity in this region. This willlead to a strengthened thermohaline circulation and subsequently to an enhanced poleward heat transport, which will in turn reverse the
SST tendendy. This is visualized in Fig. 9.l6b which shows positive SST anomalies that cover most ofthe North Atlantic at the time ofthe maximum overtuming
(lag O). The SLP anomaly pattern has also reversed, with an intensified Icelandic
low and Azorian high (Fig. 9.16a). The signals in the anomalous salinity and convection are now relatively weak (Figs. 9.l6d and 9.16t). The anomalous SLP and
SST pattern will grow further, until half a cycle is completed at lag + 5 years (not
shown). A negative SSS anomaly develops through air-sea interactions at lag + 5
years which will eventually reduce the density in the sinking region and weaken
the convection. This willlead to a weakend thermohaline circulation, reduced poleward heat transport and a change towards anomalous cold temperatures in the
North Atlantic, which completes a full cycle (not shown).
Thus as for the quasi-decadal variability, large-scale air-sea interactions are
important in generating the interdecadal variability. The atmospheric response to
mid-Iatitudinal SST anomalies seems to be a key process for the generation of midlatitudinal low-frequency variability on the time scales of decades. While the
ocean and the atmosphere is simulated at a period of about 35 years. We shall refer
this type of interdecadal variability to as the 'interdecadal mode'. Since enhanced
variability is simulated in both media at the 35-year time scale, our working
hypothesis is (as for the quasi-decadal variability) that the variability arises from
ocean-atmosphere interactions (see Timmermann (1996) and Timmermann et al.
(1998) for further details).
The 35-year period is seen nicely in the spectrum ofthe overtuming index (Fig.
9.10b). We used therefore the index of the meridional overtuming (Fig. 9.9c) to
describe the evolution of the interdecadal mode by computing lagged regres sion
patterns of selected quantities (SLP, SST, fresh water flux, SSS, surface currents,
and convection). In order to highlight the interdecadal variability, we band-pass filtered the model data prior to the regression analyses retaining variability with time
scales of 25-45 years. The anomalous SST pattern ten years prior to the maximum
overtuming (lag - 10 years) is characterized by negative anomalies which cover the
entire North Atlantic Ocean, with maximum cold anomalies near 40
0
N (Fig.
9.l4b). The associated SLP anomaly pattern is the North Atlantic Oscillation
(NAO) in its reverse polarity (Fig. 9.14a). Important to the subsequent phase reversaI are the positive salinity anomalies which develop primarily east of Newfoundland (Fig. 9.l4d). As shown in Timmermann et al. (1998), this positive SSS
anomaly is created by anomalous fresh water input (primarily through enhanced
evaporation, Fig. 9 .14c) and anomalous salt transport by anomalous Ekman currents (Fig. 9.14e). The convection at this time is near normal (Fig. 9.l4t). The SSS
anomaly has grown and expanded in area five years later (lag -5 years), as shown in
Fig. 9.l5d. At this time, the SLP and SST anomaly patterns are not well developed
and in a kind of 'transition stage' (Figs. 9 .15a and 9 .15b). The convection, however,
shows a clear intensification in the sinking region south of Greenland (Fig. 9.15t),
in response to the enhanced density which is due to the increased sea surf ace salinity in this region. This willlead to a strengthened thermohaline circulation and subsequently to an enhanced poleward heat transport, which will in turn reverse the
SST tendendy. This is visualized in Fig. 9.l6b which shows positive SST anomalies that cover most ofthe North Atlantic at the time ofthe maximum overtuming
(lag O). The SLP anomaly pattern has also reversed, with an intensified Icelandic
low and Azorian high (Fig. 9.16a). The signals in the anomalous salinity and convection are now relatively weak (Figs. 9.l6d and 9.16t). The anomalous SLP and
SST pattern will grow further, until half a cycle is completed at lag + 5 years (not
shown). A negative SSS anomaly develops through air-sea interactions at lag + 5
years which will eventually reduce the density in the sinking region and weaken
the convection. This willlead to a weakend thermohaline circulation, reduced poleward heat transport and a change towards anomalous cold temperatures in the
North Atlantic, which completes a full cycle (not shown).
Thus as for the quasi-decadal variability, large-scale air-sea interactions are
important in generating the interdecadal variability. The atmospheric response to
mid-Iatitudinal SST anomalies seems to be a key process for the generation of midlatitudinal low-frequency variability on the time scales of decades. While the