JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 98, NO. C7, PAGES 12,501-12,509, JULY 15, 1993 Simulated Seasonal Circulation in the Indonesian Seas YUKIO MASUMOTO AND TOSHIO YAMAGATA Department of Earth and Planetary Physics, Faculty of Science, University of Tokyo, Tokyo Seasonal circulation in the Indonesian Seas is investigated in detail by use of a high-resolution oceanic general circulation model forced by the monthly mean climatological wind stress. Although the net barotropic through flow between the Pacific and Indian oceans is not permitted, the model appears to be successful in predicting the observed transport through the Lombok Strait. The modeled transport varies from almost nothing in January to 5 Sv in September, with a local minimum in June; this seasonal variation is regulatedby means of remote actions of the monsoonalwinds in the tropical Indian Ocean as well as local and remote action of the monsoonal winds over the western Pacific. The drainage into the South China Sea through the Java Sea during the boreal summer may explain the apparent local minimum of the modeled Lombok through flow. The root cause of the semiannual aspectof the Lombok through flow, however, lies in the Wyrtki jets of the Indian Ocean, which reduce the positive pressure gradient across the Lombok Strait twice a year. 1. INTRODUCTION The interbasin mass exchange between the Pacific and Indian Oceans through the Indonesian Seas (Figure 1) is believed to play a crucial role in the global ocean thermo- haline circulation [cf. Broecker, 1991]. At least, the Pacific- Indian through flow may be crucial in accounting for the heat and freshwater balance in the upper layer around Australasia [cf. Wijffels et al., 1992]. Since the 1960s, there have been several attempts to estimate the transport of the through flow by usinga variety of indirect methods[Wyrtki, 1961; Godfrey and Golding, 1981; Piola and Gordon, 1984; Fine, 1985; Fu, 1986; Gordon, 1986; Godfrey, 1989]. Those estimates dis- perse significantly from a negligible value to 16 Sv. As demonstrated by Fu [1986], some estimates are highly vul- nerable to the commondifficulty of estimating the barotropic component of ocean currents. According to the only one direct measurement reported so far [Murray and Ariel, 1988], the 1985 annual mean transport through the Lombok Strait was only 1.7 Sv, with a seasonal root-mean-square (rms) deviation of 1.2 Sv. In addition, Murray and Ariel [1988] showed that about 80% of the net transport is con- tained in the upper 200 m, in good agreement with Wyrtki's conclusion [1987] that the lateral pressure gradient is con- centrated in the upper 200 m. Several attempts have been made to address the through flow by using numerical models. Semtner and Chervin [ 1988] reported an annual mean transport of about 16 Sv when they used the global oceanic general circulation model (OGCM) forced by the climatological wind stress. This high value is close to the early estimate made by using OGCMs [Cox, 1975; Takano, 1975]. However, their model geometry is rather coarse, especially in the Indonesian seas. In contrast, trading off the fine vertical resolution, Kindle et al. [1989] developed a reduced-gravity model resolving rather detailed geometry. Their calculation yields a net transport of 7.5 Sv with almost the same amount of seasonal variability. M. Inoue and S. E. Welsh (personal communication, 1991) adopted a similar reduced-gravity model with fine lateral resolution. Although their model domain is limited to the Copyright 1993 by the American Geophysical Union. Paper number 93JC01025. 0148-0227/93/93 JC-01025505.00 tropical Pacific and eastern Indian Oceans, the use of the open boundary conditions leads to an annual mean net transport of 9.8 Sv with almost the same amount of seasonal variability. Despite those efforts, the annual mean of the through flow and the annual variation remain controversial. In the present analysis, made by using a high-resolution OGCM with fine geometry and bottom topography, we permit only a purely baroclinic component with regard to the Pacific-Indian Ocean through flow. In other words, the net barotropic through flow, including the circulation around Australasia, is prohibited in the present model. This ad hoc measure is expected to be useful in order to demonstrate the effect of local circulations around islands in the Indonesian seas on the through flow. In particular, it allows us to elucidate a contribution to the Lombok through flow from the local barotropic circulation around the Lesser Sunda Islands. The present model is assumed to give a realistic picture of upper layer circulations in the Pacific and Indian oceans, since it is the baroclinic component associated with a strong surface 20NI_ . I• I I ß .. PAC I F I C OCEAN SOUTH /• CHINA /.t _ •1_• [ .• SEA / .... • I ND I AN OCEAN 100E 120E 140E 150E Fig. 1. Chart showing the Indonesian archipelago resolved by the present Indo-Pacific model. Dashed lines correspond to sections across which depth-integrated mass transport is calculated. 12,501