Fig. 6 The fractional composition of the throughflow in each of the entrance channels to the archipelago given by (2) is contoured as a function of the streamfunction on Sulawesi and at the edge of the Pacific western boundary layer assuming the streamfunction on Halmahera has nondimensional value 0.6. Regimes in which the throughflow fraction is wholly NP are shaded light grey, and SP dark grey.
Returning to Fig. 1 of the change in salinity observed over the last 30 years, the contrast between changes in the Maluku and Halmahera Seas suggests another alternative, which is that most of the SEC flowing through the Halmahera Sea has been diverted to the Maluku passages leaving local effects to determine the salinity within the Sea. The interpretation is consistent with ARLlNDO's finding of insignificant transport through the Halamahera Sea, Gordon (pers. comm. 1998).
CONCLUDING REMARKS
Observations of subsurface salinity, and analysis of wind stresses combined with a simple model, suggest decadal-scale variability in gyre closure in the western equatorial Pacific Ocean yielding a fresher equatorial Pacific and saltier (and possibly warmer) throughflow in recent times. It will be interesting to see whether GCMs with their more complex 3-D flows show this variability. It is noteworthy that the result is very sensitive to the wind stress data used to force the model. In many ways the finding is not surprising, and likely a consequence of the decadal-scale variability in the Pacific system noted by many other authors. The impact on the Indian Ocean and climate is difficult to assess, and requires further modelling studies.
Regarding several observational programs, which are in the planning stages, more direct measurements of cross-equatorial heat transport in the Indian Ocean would be desirable. Monitoring throughflow transport appears to be more important than monitoring its T, S properties. Though a TRITON-like buoy in the Banda Sea would give us useful information on shifts in the contribution from the SEC source.