Thermohaline stratification of the Indonesian Seas: Model and observations

The Indonesian Throughflow, weaving through complex topography. drawing wat er from near the division of the North Pacific and South Pacific water mass fields, represents a severe challenge to modeling efforts. Thermohaline ob servations within the Indonesian seas in August 1993 (southeast monsoon) an d February 1994 (northwest monsoon) offer an opportunity to compare observa tions to model output for these periods. The simulation used in these compa risons is the Los Alamos Parallel Ocean Program (POP) 1/6 drg ton average) global model, forced by ECMWF wind stresses for the period 1985 through 199 5. The model temperature structure shows discrepancies from the observed pr ofiles, such as between 200 and 1200 dbar where the model temperature is as much as 3 degrees C warmer than the observed temperature. Within the 5 deg rees-28 degrees C temperature interval, the model salinity is excessive, of ten by more than 0.2. The model density, dominated by the temperature profi le. is lower than the observed density between 200 and 1200 dbar, and is de nser at other depths. In the model Makassar Strait, North Pacific waters ar e found dawn to about 250 dbar, in agreement with observations. The model s ill depth in the Makassar Strait of 200 m, rather than the observed 550-m s ill depth, shields the model Flares Sea from Makassar Strait lower thermocl ine water, causing the Flores lower thermocline to be dominated by salty wa ter from the Banda Sea. In the Maluku, Scram, and Banda Seas the model ther mocline is far too salty, due to excessive amounts of South Pacific water. Observations show that the bulk of the Makassar throughflow turns eastward into the Flores and Banda Seas, before exiting the Indonesian seas near Tim er. In the model, South Pacific thermocline water spreads uninhibited into the Banda, Flores, and Timer Seas and ultimately into the Indian Ocean. The model throughflow transport is about 7.0 Sv (Sv = 10(6) m(3) s(-1)) in Aug ust 1993 and 0.6 Sv in February 1994, which is low compared to observationa lly based estimates. However. during the prolonged EI Nino of the early 199 0s the throughflow is suspected to be lower than average and, indeed, the m odel transports for the non-El Nino months of August 1988 and February 1989 are larger It is likely that aspects of the model bathymetry, particularly that of the Torres Strait, which is too open to the South Pacific, and the Makassar Strait, which is too restrictive, may be the cause of the discrep ancies between observations and model.