We also plotted the unsmoothed SST anomalies averaged over several key regions where these modes exhibit significant signals. Our first PC is similar to the counterpart of NLY97 based on the COADS data, but ours traces the decadal SST variability in SAFZ even better than theirs (Fig. 4a). Our first PC races the variability in KOIZ also well, although changes in KOIZ tend to occur slightly earlier than in SAFZ. Unlike the “1988/89 shift”, the “1976/77 climate shift” did not manifest itself clearly in SAFZ (Miller et al. 1994; NLY97). As pointed out by NLY97 and consistent with Watanabe and Mizuno (1994), cooling in SAFZ in the mid-1970's (Fig. 4a) preceded the warming by 2 years that occurred in 1977 over the equatorial Pacific (Fig. 4b) concomitantly with the cooling in STFZ (NLY97) and nearly so with the warming along the Kuroshio off Japan (Fig. 4b). On decadal time scales, the SST in the northern Sea of Japan varied more or less coherently with that in SAFZ, except around 1970 when the warmth in SAFZ had no counterpart in the Sea of Japan. The decadal variability in the southern Sea of Japan looks more coherent with that in the East China Sea, both of which are represented reasonably well by the third PC (Fig. 4c). This PC exhibit marked positive correla-tion with those in surface air temperature over China, once a secular trend is removed from the latter (Fig. 4d), which is consistent with Fig. 3j.
