During this phenomenon, the instruments of this mooring experienced large vertical motion. Although they could not measure the parameters at the designed depth, depth-time figure (Figure 3) can be plotted using this vertical motion. Vector plot (Figure 3a) shows that this strong flow seems to be at least 300m thick. Salinity plot (Figure 3b) shows that there was low salinity water (＜ 34.6 PSU) advected by this flow above 350m. This water was mixture of the North Pacific Intermediate Water (NPIW, Bingham and Lukas, 1994) and water from the South Pacific because of its potential temperature-salinity relations. Comparison of these figures suggests that some of the North Pacific Water came from southeast.
In order to understand this phenomenon, it is necessary to understand why the North Pacific water come from southeast. Fortunately, we found the low salinity water (＜ 34.5 PSU) having the characteristic of the NPIW along 130°E during June 1996 (Figure 4). Salinity section during this cruise shows that low salinity tongue intrudes from the north of 8°N to 2°N along the isopycnal of σθ=26.5, which is coincide with that of the low salinity water seen in Figure 3b above 350m. Additionally, there was an anticyclonic eddy located at the west of 130°E during the cruises at July 1995 and June 1996 (Figure 5): This was the Halmahera Eddy. The low salinity water observed at the southern mooring probably came from 130°E line advected by the Halmahera Eddy.
During this phenomenon, northeasterly monsoon was dominant in the Philippine Sea (not shown). Then, its curl τ was positive and upwelling is expected from this wind distribution in this region. Masumoto and Yamagata [1991] suggested that Ekman pumping due to positive wind stress curl during boreal winter results in an enlargement of the Mindanao Dome, which is upward transformation of thermocline off the Mindanao. We suggest that the enlargement of the Mindanao Dome let the NECC move southward and results in the southward shift of the Mindanao Eddy. That is, the strong northwestward flow at the southern mooring appears to be a part of the Halmahera Eddy which advected the low salinity water from southeast. Hydrographic observations during boreal winter (1994) and summer (1996) are consistent with this suggestion of the Halmahera Eddy meridional migration between these seasons. The results from the JEGCM also suggest that mean location of the Halmahera Eddy during boreal winter is further south than that during boreal summer (Figure 6).
Similar strong northwestward flow was also seen in March 1994. Considering this result and correspondence between periods of the northeasterly monsoon and strong flow, this phenomenon seems one of the seasonal signals in this region. On the other hand, westward flow with high salinity water (＞34.8PSU) which is originated from the south Pacific was observed above 350m depth during boreal summer. This result means that the south Pacific water enter into the Maluku Sea, and the Maluku Sea is also one of the eastern routes of the Indonesian Throughflow during boreal summer.
Next, we focus intraseasonal variability Oscillations with period of 50 days are clearly seen in the current time series at the northern mooring site (Figure 2a). Although it is not clear in the raw time series at the southern site, band passed time series between 30 days and 120 days (not shown) shows that similar oscillation also occurred there.
As well known, there is also variability with similar period in the tropical atmosphere: Madan-Julian Oscillation (MJO,Madan and Juilan, 1994). In order to check the relation between this variability and 50-day oscillations in current, we analyzed the wind data derived from the numerical forecasting model by Japan Meteorological Agency. This result says that there is a peak in power spectrum of u-wind on 1000 hPa surface with period of 40 - 50 days near the mooring sites,and coherence between wind and current is ＞ 0.4. Because the null hypothesis (correlation = 0) is rejected when its correlation exceeds 0.137 with a confidence level of 95 %, local wind variability associating with MJO may correlate with the 50-day oscillations.
