284
a)
r=0.91
2.0
0.0
-2.0
-4.0 '----_~ _ _ --L-_ _ ~ _ ___' _ _ ~ _ _ _"_ _ _ ~ _ _ _'___________"
20
40
60
80
100
Time [a]
Figure 10: Canonical Correlation Analysis (CCA) of low-pass filtered (retaining variability
with time scales longer than 5 years) North Atlantic SST and 500 hPa height anomalies
simulated by the CGCM. The SST and height anomalies were projected onto the first five
EOFs prior to the analysis. a) canonical time series.
with a period of the order of about 15 years. However, the observational
record is certainly much too short, in order to prove the existence of a
decadal climate cycle in the North Atlantic.
We investigated therefore additionally the variability simulated by our
coupled ocean-atmosphere general circulation model "ECHO" in the multidecadal integration described above. In order to derive the major modes
of ocean-atmosphere co-variability, we performed a Canonical Correlation
Analysis (CCA, see Barnett and Preisendorfer (1987)) between low-pass
filtered simulated North Atlantic SST and 500 hPa height anomalies. The
coupled run shows a rather stable decadal cycle in the North Atlantic, as
can be inferred from the time series of the leading CCA mode (Fig. lOa).
The canonical correlation amounts to about 0.9, and the period of the
decadal variations is of the order of 17 years, which is consistent with the
observations of Levitus et al. (1994) shown in Fig. 9.
a)
r=0.91
2.0
0.0
-2.0
-4.0 '----_~ _ _ --L-_ _ ~ _ ___' _ _ ~ _ _ _"_ _ _ ~ _ _ _'___________"
20
40
60
80
100
Time [a]
Figure 10: Canonical Correlation Analysis (CCA) of low-pass filtered (retaining variability
with time scales longer than 5 years) North Atlantic SST and 500 hPa height anomalies
simulated by the CGCM. The SST and height anomalies were projected onto the first five
EOFs prior to the analysis. a) canonical time series.
with a period of the order of about 15 years. However, the observational
record is certainly much too short, in order to prove the existence of a
decadal climate cycle in the North Atlantic.
We investigated therefore additionally the variability simulated by our
coupled ocean-atmosphere general circulation model "ECHO" in the multidecadal integration described above. In order to derive the major modes
of ocean-atmosphere co-variability, we performed a Canonical Correlation
Analysis (CCA, see Barnett and Preisendorfer (1987)) between low-pass
filtered simulated North Atlantic SST and 500 hPa height anomalies. The
coupled run shows a rather stable decadal cycle in the North Atlantic, as
can be inferred from the time series of the leading CCA mode (Fig. lOa).
The canonical correlation amounts to about 0.9, and the period of the
decadal variations is of the order of 17 years, which is consistent with the
observations of Levitus et al. (1994) shown in Fig. 9.
