As a validation of zonally varying current structures and associated thermocline depth and salinity distribution around the equator, we compared simulated fields with observational ones (Fig.2 of Wyrtki and Kilonsky, 1984).

　

　

Simulated annual means of the zonal current velocity, isotherm and salinity distributions are given in the upper, middle and lower panels in Figure 4. Those fields given in Figure 4 are the annual mean at tenth model year and no substantial changes were noticed in the following 40 years. The horizontal and vertical extents of the equatorial undercurrent are in good agreement with the observation and the maximum velocity appeared around the depth of 150 m reaches 80 cm/s, which is a little bit weaker than the observed value. The flow patterns extending to the northern and southern flanks of the undercurrent are also well simulated. The thermocline structure given in the middle panel, especially its accentuated contour intervals is very close to the observed structure except for the magnitude of meridional gradient around 5°N〜10°North. As to salinity field in the bottom panel, we noticed that the overall distribution pattern does not deviate considerably from the observational one. A closer inspection shows, however, that just the north of the equator there exists a thin lower salinity tongue extending downward at the depth around 200〜250 m and we are now checking whether it is the outcome of numerical noises or not. In spite of the existence of those detailed structures, we think that our simulation is successful in reproducing realistic vertical profiles of velocity, temperature and salinity fields and that the success is due not only to the employed high horizontal resolution but to high vertical resolution.

　

Realistic meridional heat transport is difficult to simulate with low horizontal resolution model in which mesoscale eddy parameterization is not incorporated (Bning et al., 1995). Figure 5 shows the annual mean meridional heat transport calculated from our eddy-resolving model. The maximum poleward heat transport in the northern and southern hemispheres turned out to be about 1.5 PW between 10 - 30°N and about 0.7 PW between 10 - 15°S respectively. Thus simulated meridional heat transport pattern is not symmetric to the equator, while direct estimates based on hydrographic observations (Hastenrath, 1982; Talley, 1984) and the indirect one using satellite radiation and atmospheric reanalysis data (Trenberth and Solomon, 1994) also suggest an anti-symmetric pattern but with a lesser degree. This discrepancy is possibly due to the underestimation of the southward heat transport in the South Pacific and/or the Southern Indian Ocean (Figs.5c, 5d), though the southward heat transport in both the Pacific and the Indian Ocean derived from the model is within the range of observed values.

　

The large latitudinal change of the heat transport can be seen in 8-10°S in the Pacific and the Indian Ocean (Figs.5c, 5d). The amplitude reaches to about 0.9 PW, which is comparable to the northward heat transport by the Gulf Stream in the North Atlantic (Fig.4b). This indicates that the Indonesian Throughflow carries such a huge amount of heat from the Pacific to the Indian Ocean and possibly affects the climate over the Indian Ocean. In the North Atlantic, the meridional heat transport between 20 - 36°N is approximately 1 PW, which is similar to the estimation from the hydrographic observation (Fig.4b). Such a large heat transport is attributed mostly to the northward transport of the warm water by the Gulf Stream/the North Atlantic Current and to the southward transport of cold North Atlantic Deep Water forming the Deep Western Boundary Current. Bning et al. (1995) showed that the horizontal resolution is one of the crucial factors for a reliable estimate of the meridional heat transport. Actually, heat transport in that region simulated with a coarse horizontal resolution of 1°turned out to be only 0.7 PW, while the same model with mesoscale eddy parameterization (e.g. Gent and McWilliams, 1990) can reproduce the northward heat transport of about 1 PW, which is comparable to the one obtained in our eddy-resolving simulation.