
Fig. 1
EOF1 (a, b) and PC1 (c) of monthly SLP anomalies for the cold season (November through March) from 1910 to 2009. The anomalies are constructed only from (a) (c) the first four harmonics over the centennial period (i.e., with periods of 25 yr or longer) and (b) the higher harmonics (i.e., with periods shorter than 25 yr). The fraction (%) of the SLP variance explained by EOF1 is shown in the upper-right corner. EOF1 in each of (a) and (b) is well separated from other EOFs as per the criterion of North et al. (1982). Contour interval is (a) 0.3 hPa and (b) 0.9 hPa, and zero lines are omitted. Stippling and heavy shading in (b) denote the domains used for averaging SLP anomalies for the individual centres of action of AO and AIS, respectively.

Fig. 2
November–March statistics for a 25-yr moving period centred at a given year on the abscissa. (a) AIS index (see text for further detail). Contour intervals are 0.2. Light, moderate and heavy shading indicate significance at the 90%, 95% and 99% confidence levels, respectively, based on one-tailed t-test. (b) AL variability (hPa) defined by 25-yr standard deviation of SLP anomalies averaged over the AL domain (32°–54°N, 162°E–134°W). (c) Percentage of ‘pure AO events’ among all the ‘potential AO events’ with large AO index values regardless of their signs. (d) AL anomaly (hPa) defined by area-averaged SLP anomalies in EOF1 over the AL domain. Dotted domains bounded by solid black lines indicate periods in which EOF1 is not well separated from other EOFs.

Fig. 3
(a) and (b): As in Fig. 2a and 2d, respectively, but based on winter-mean (December—March) anomalies. In (b), the AL anomaly is not plotted when the corresponding EOF1 is not separated well from other EOFs. Asterisks denote the four 25-yr periods centred at 1941, 1956, 1972 and 1997, for each of which EOF1 of winter-mean SLP anomalies is shown in Fig. 4. (c) PC time series for EOF1 of multi-decadal SLP anomalies constructed only from the first four harmonics over the 1910–2009 period (i.e., with periods of 25 yr or longer). The PC values have been averaged within individual winter seasons before plotted. The corresponding SLP anomaly pattern is given in Fig. 1a, whose polarity corresponds to positive PC values.

Fig. 4
EOF1 of winter-mean SLP anomalies over the extratropical Northern Hemisphere (poleward of 20°N) defined for four 25-yr periods, whose central years are labelled in the panels and denoted with asterisks in Fig. 3. The fraction (%) of the SLP variance explained by EOF1 over each of the periods is shown in brackets. Contour interval is 1.5 hPa, and zero lines are omitted. The polarity corresponds to the positive phase of AO. Light and heavy shading indicates the anomalies significant at the 95% and 99% confidence levels, respectively. Each of the periods, EOF1 is well separated from other EOFs.

Fig. 5
Same as in Fig. 4, but for the 25-yr mean SLP anomalies. Contour interval is 0.4 hPa. Zero lines are omitted.
