The continuous migration in the Pacific sector is also observed in the latitude-time cross section of the 10-30-year band-pass filtered SLPs, which have higher temporal resolution than the MWF.
We also examined the SST and LAT fields. The gridded SST data are produced from GOADS and newly digitized Kobe collection (Manabe 1999), and have significantly improved data coverage than the SST data produced only from COADS in the early few decades in the 20th century. The SST analysis strongly suggests that the boundary between the oval and horseshoe SSTs shifted toward the south compared between the early and late few decades in the 20th century. The LAT analysis has indicated that the high-latitudes BO in the LAT field continued through the 20th century but its penetration into the mid-latitudes was limited from 1930-1970. These SST and SAT results are generally consistent with the southward BO migration in the SLP field in the Pacific sector.
The spatial structure changes of the BO have substantial implications on the mechanism of the BO. Several papers has proposed that the BO arises from the air-sea interaction in the Pacific Ocean (Latif and Barnett, 1994, 1996; Gu and Philander 1997; Jin 1997; White and Cayan 1998; Yukimoto et al., 1999; Knutson and Manabe 1998; Weng and Neelin 1999; Talley 1999; Cessi 2000). It is noteworthy that all the mechanisms studied by these papers are more or less delay-negative feedback, though the physical processes for the delay are different from a model to another. In a delay-negative feedback model, the oscillation period is generally proportional to the delay time, and hence the wider ocean basin, which can involves a longer delay time, may be related with the longer oscillation period. In the early of the 20th century, the bidecadal oscillation has a shorter timescale with mid-latitude contribution in the North Atlantic, whereas after the middle of the 20th century the mid-latitude contribution is located on the Pacific basin with the longer BO timescale. Thus, the authors would like to point out the possibility that the changes of the spatial structure and timescale of the BO is related with the changes of the relative contributions between the Pacific Ocean and Atlantic Ocean. This hypothesis is, of course, at a crude speculative stage, but would be worth to be examined using air-sea coupled models in future.
Another interesting physical implication of the BO migration is for the resonance between the BO and the PPO over the North Pacific. Although the three climatic regime shifts in the 1920s, 1940s and 1970s are quite robust so that independent studies detected them (Minobe 1997; Mantua et al., 1997), it is not clear what was occurred around 1900. Minobe (1997) proposed that another regime shift occurred around 1890 based on air-temperature over mid-latitude western North America. However, the occurrence of the regime shift was not so clear, and different assessment pointed out the possibility of the regime shift around 1900 (S. Hare personal communication). The lack of a prominent regime shift around 1890-1990 may be related with the lack of the energetic BO over the central North Pacific.
From the aspect of the methodology, the present results have shown that the MWF is a useful tool to know the spatial structure change corresponding to the frequency structure change as far as maximal scale dilation parameters are continuously detected as a function of time. On the other hand, when maximal scale dilation parameters are discontinuous, MWF do not give physically meaningful information. However, it is not difficult to distinguish whether scale dilation parameters at the maximal wavelet amplitudes are continuous or not, when one calculates wavelet transform for a representative time series. Because wavelet transform is rapidly getting popular in studies for the ocean and atmosphere, the authors expect continuous changes of the scale-dilation parameters at the maximal wavelet amplitudes are sometimes observed in applications of the wavelet transform to climate researches.
