Two different time scales appear to contribute to the low-frequency fluctuations in the wintertime SST over the NW Pacific. In SAFZ (Fig. 4a), strong decadal fluctuations are super-imposed on the multi-decadal variability, manifested as the relatively warm 1950's〜1960's and a cool period since mid-1970's (Minobe 1997). Yet, this signal is not apparent in the marginal seas (Figs. 4c〜d). The interdecadal warming trend in surface air temperature over China since the late 1960's is in contrast with the cooling trend in SAFZ.
DISCUSSION
Our second mode seems to be consistent with the scenario suggested in the previous studies, in a sense that the tropical SST anomalies could change the extratropical atmospheric circulation and generate SST anomalies with the opposing sign in STFZ. Unlike suggested in the previous studies, however, the atmospheric anomaly pattern aloft engaged in this process is not the classical PNA but a meridional seesaw over the eastern Pacific. At the surface, it is the subtropical high but not the Aleutian low that changes its intensit. It is suggested in Fig. 3 that positive (negative) SST anomalies in STFZ may be reinforced by the suppressed (enhanced) surface westerlies and intensified (weakened) Trades in the presence of the stronger (weaker) subtropical high. The oceanic temperature anomalies generated around the subtropical front may be transported along an isopycnal surface in the interior upper ocean and then later alter the thermal structure in the tropical Pacific (Deser et al. 1996; Gu and Philander 1997), although this process may be too slow to explain the rather quick transition in the tropical and subtropical Pacific in the 1970's (Fig. 4b). In fact, a simulation by McCreary et al. (1998) suggests that changes in the intensity of a shallow overturning cell between the tropics and subtropics due to the anomalous Trades could lead to the decadal SST fluctuations similar to those in our second mode.
Yet, on the decadal scale, tropical influence via the atmospheric teleconnection seems limited over the subtropics. Seemingly, the northern North Pacific is not just a slave of the tropics on that time scale. In fact, we have shown that the cooling in SAFZ occurred 2 years in advance of the tropical warming in 1977. Moreover, a SST difference map by Koide and Kodera (1997) based on the adjacent 5-year periods before and after 1977 resembles our second EOF, while their another 5-year difference map before and after 1989 resembles our first EOF. The concentra-tion of decadal SST anomalies in SAFZ and STFZ in spite of broader distribution of the associated surface wind anomalies is suggestive of the atmospheric forcing via anomalous frontgenetic effects by the Ekman current.
