24
10
1;15
.§.. 20
w
a:
::::l
en 30
en w
a:
a.. 40
50
70
0
Figure 12: The two leading EOFs of the temporal correlation matrix constructed from the
time series of 5-month running mean zonal wind anomalies at the 70-, 50-, 40-, 30-, 10-,
15-, and 10-mb levels over the equator based on data for the period May 1956 through
Apri11990. EOF1 is solid and EOF2 is dashed. From Wallace et al. (1993).
Figure 13: As in the lower left panel of Fig. 6, but based on the expansion coeiIicient
(or principal component) time series of the two leading EOF's shown in Fig. 12. From
Wallace et al. (1993)
tified with spatial patterns that appear in quadrature with one another in
the time domain. For example, in representing the annual cycle, the x axis
might represent the July minus January pattern and the y axis the October
minus April pattern. The angle relative to the x axis represents the phase
of the oscillation (e.g., calendar month in the annual cycle relative to an
arbitrary reference time such as I January). The 'climate state', which can
be reconstructed based on a knowledge of PCI and PC2, progresses around
the circular orbit at a uniform rate: i.e., the spatial pattern associated with