By concentrating on the salinity minimum of the Antarctic and North Pacific Intermediate Waters, they found clear signatures indicating freshening in the outcrop regions of these waters. Correspondingly, they found salinity increases along neutral surfaces at shallow depths in tropical latitudes, suggesting that the excess of evaporation over precipitation has increased in these regions.
Use of expendable and mechanical bathythermograph data:
Ridgway and Godfrey (1996) examined XBT data in the southwestern Pacific from recent decades, after the fashion of Levitus (1989). The relatively large amount of data permits statistically significant results to be found despite the lack of salinity data. Temperature at 450m has cooled over 1℃ near the centre of the South Pacific subtropical gyre, from 1975-1979 to 1985-1989; a gyre-like pattern of change accounted for 48% of the interpentadal variance. The changes have the same vertical signature as those found across eddies, suggesting that the changes are primarily due to a weakening of the gyre. Their results indicate that the inflow and outflow to the East Australian Current, generally about 25 Sv, changed by 10 Sv or more from the late 1970's to the late 1980's. A second EOF (26% of interpentadal variance) is concentrated near the south end of the East Australian Current. Holbrook and Bindoff (1997) undertook a more extensive survey, using MBT's as well as XBT'S, thus extending their coverage back to 1955. They did not restrict themselves to interpentadal variability, so that their data had substantially more degrees of freedom. They found a wide variety of modes of variability. None of these individually accounted for more than 6% of the variance, though several were statistically significant. Some of these (particularly tropical modes) correlated with the Southern Oscillation Index.
Model studies of multidecadal temperature change in the southern Indian Ocean:
SST has risen particularly sharply over the period 1900-1983, in the southern Indian Ocean; this has potential implications for trends in Australian climate. Allan et al. (1995) have analysed COADS data for the Indian Ocean, by 21-year blocks in this time period; data analysed were wind stresses, SST, cloud cover and sea level pressure. The stresses have been used to model SST and circulation anomalies by Reason et al. (1996a). The model omits some potentially important thermodynamic effects, but it demonstrates that changes in circulation induced by the observed wind changes may be a significant contributor to the observed interdecadal SST variability. Reason et al. (1996b) then performed similar runs with a global ocean model; they found that strengthening Pacific winds produced more realistic spatial patterns of SST change in the south Indian Ocean, through an increase in the Indonesian Throughflow magnitude.
Interdecadal variation in coastal sea level:
Finally, it is worth noting that tide gauge records are potentially a valuable source of information on intercadal variations in ocean circulation. Clarke and Liu (1994) formed interannual time series of sea level for northwestern and western Australia, over 1955-1988; the two are strongly correlated with one another, and with the Southern Oscillation Index. They also formed an index of the Indonesian Throughflow, which also correlates strongly with the SOI. However, they did not find interdecadal variations over that period which were significant, over the large noise associated with ENSO variations.
