174 Mojib Latif, Axei Timmermann, Anselm Gr6tzner, Christian Eckert, Reinhard Voss
Ensemble Tralectories
6 r--r---.---,-----,
4
-2
-4
-6 ~2~1- 0~-- ~~ . 0-~zro~--~ 2~
2
o
-2
-4
Ume
~~ __ ----~~~--~-J
660
Urne
6r-.---r-~__r-____,
4
2
O
-2
-4
~ L-6~20 -:-~-::S3:c0:----~,-L-~---l650
time
2
o ~NR~
-2
-4
-6 L-~ ~~0--~2~30-----2~L---~250
time
Norm. Ensemble Var.ance
2 .0 r-..,---r----.,.-...o-,
1.5
1.0
0.5
2~
Urne
2.0 r---,-----r-~---.----_,
1.5
1.0
0.5
0.0 ..... 630 ~----=~-'-:-~-650~--~660
tlme
2.0 "-",---,,,,-~---, _ _ _,
1.5
t .O
0.5
0.0 ........ _ _ ""-~~ ____ ....J
630
tirne
~
240
650
1 !
250
Fig. 9.19 Individual ensemble trajectories and ensemble mean (thick line) of projections
onto the first EOF of the North Atlantic SST anomalies for the four different forecast
ensembles (left) and time series ofthe respective normali sed ensemble variances (right). The
thick dashed lines show the temporal behavior of fitted first order autoregressive processes.
From Gr6tzner et al. (1999).