Oceanic precondition and evolution of the 2006 Indian Ocean dipole

3.1. Subsurface Ocean Conditions

[8] The corresponding subsurface evolutions in the eastern equatorial Indian Ocean are shown in Figures 2b–2e. The equatorial subsurface temperature variations are strongly related to the vertical movement of the thermocline and strength of the vertical temperature gradient. TRITON buoy at 1.5°S, 90°E observed quite tight thermocline around the depth of 120 m during the first half of 2006, with gradual spreading of the isotherms toward the end of the time series (Figure 2b). Due to this tight thermocline, the temperature anomalies were positive (negative) at the upper (lower) part of the thermocline. A region of the large negative temperature anomaly was first appeared from the thermocline down to 250 m depth in May. The similar subsurface temperature anomalies were also observed by the TRITON buoy at 5°S, 95°E, although the positive temperature anomaly in the upper thermocline continued until the end of May (Figure 2e).

[9] The zonal currents observed by the TRITON buoy (Figure 2c) and ADCP (Figure 2d) indicated westward current anomalies in May above the thermocline, with the maximum value larger than 50 cm/s. In the climatological variations, semiannual westerly winds in the central equatorial region drive eastward swift currents in the upper-ocean, known as Wyrtki jets [Wyrtki, 1973], during the monsoon transition periods in April/May and October/November (see contours in Figure 2d). The anomalous westward currents in May 2006 were as large as the magnitude of the climatological Wyrtki jet, suggesting abnormally weak eastward currents, which was quite similar to the condition in 1994 [Vinayachandran et al., 1999].

[10] The initiation of the negative subsurface temperature anomalies in May was about three months earlier than that of the negative SSTA associated with the IOD onset in August 2006. Interestingly, the subsurface negative anomalies were “capped” by weak positive temperature anomalies near the surface during May and June at both the TRITON locations. Another anomalous event occurred in August, with rapid shoaling of the thermocline (>30 m), large negative temperature anomaly around the depth of 75 m (<−5°C), and strong westward currents in the upper-layer (>50 cm/s). It was this time of the year that the coherent SSTA pattern appeared and the DMI increased dramatically within a month.

[11] The negative temperature anomalies in the thermocline at 1.5°S, 90°E once weakened by the sudden downward movement of the thermocline in early September, associated with the short period of eastward upper-layer currents. This event, however, lasted only about ten days and the negative anomaly developed again. The strongest negative subsurface anomaly (<−6°C) occurred during November and December 2006, when the surface signatures of the IOD disappeared quickly in concert with the seasonal monsoon reversal. The temperature anomalies at the thermocline depth at 5°S, 95°E (Figure 2e) also demonstrate the peak in late-November to December, but the phase lagged about 15 days behind that at 1.5°S, 90°E.

[12] During the peak of the 2006 IOD from September to November, the current data revealed peculiar subsurface structures. The surface current anomaly at 10 m depth observed by the TRITON buoy were westward (Figure 2c), which was consistent with the direction of the anomalous easterly wind prevailing over the buoy location (Figure 1c). Unlike the similar case of the westward currents in May, there existed eastward current anomalies in the layer between 50 m and 150 m depths from September to November, with slight upward phase shift. This suggests that the negative vertical shear of the zonal current was generated during the second half of the positive IOD period, with the stronger subsurface zonal current and the reversed surface current.

[13] The DMI returned to the normal condition after December 2006, as the negative SSTA in the eastern region became positive and basin-wide warming occurred in the equatorial Indian Ocean (Figures 1d and 2a). The subsurface negative temperature anomalies, on the other hand, continued for more several months at 1.5°S, 90°E, during which the thermocline spread significantly in vertical and the negative anomalies were gradually replaced by the positive anomalies in the lower part of the thermocline. The termination of the negative temperature anomaly in the thermocline at 5°S, 95°E, on the other hand, was characterized by sudden change to the positive anomalies at all the depth from the surface down to 300 m depth, except for the thin layer between 50 m and 100 m depths. The zonal current anomaly at the depth of 10 m switched from westward to eastward in late November (Figure 2c), concurrent with the decay of the positive IOD.

3.2. Zonal Winds and Sea Surface Height Variations Along the Equator

[14] We now argue the large-scale variation in the entire equatorial Indian Ocean by use of the satellite-based data. Longitude-time sections of zonal wind anomalies and sea surface height anomalies (SSHA) are shown in Figure 3, together with the time series of the depth of 20°C isotherm (D20) obtained by the TRITON buoy at 1.5°S, 90°E. The variability of D20 closely corresponds to the SSHA variations at 90°E, with a correlation coefficient of 0.85.

[15] A strong negative SSHA in a region east of about 85°E and large positive anomalies to the west between September and December were associated with the 2006 positive IOD. From January to September 2006, prior to and beginning of the positive IOD, there were several eastward propagating SSHA signals along the equator, with the phase speed of about 2.5 m/s that corresponds to the value of the gravest equatorial Kelvin wave. Such equatorial Kelvin waves can be excited by both the zonal wind stress along the equator and reflections of the equatorial Rossby waves at the western boundary of the basin. For example, the upwelling Kelvin wave first appeared in February (a dashed line marked with ‘A’ in Figure 3b), in concurrence with strong easterly wind anomalies in the central Indian Ocean in early February. Since the negative SSHA occupied in the western region from the beginning of January, it is likely that this Kelvin wave was partly excited by western boundary reflection of the upwelling Rossby waves generated in the western off-equatorial Indian Ocean in late 2005 [Luo et al., 2007]. However, this event in February caused only the minor negative temperature perturbation at the lower part of the thermocline (Figure 2b). It seems that the positive temperature anomalies in the shallower part of the thermocline were not affected by the wave.

[16] The initiation of the coherent subsurface negative temperature and the zonal current anomalies in May (Figure 2) was coincident with the basin-wide equatorial easterly wind anomalies (Figure 3a) and the eastward propagation of another negative SSHA (‘B’ in Figure 3b), suggesting an important role of the upwelling Kelvin wave on the subsurface anomalous conditions. In late June to July, the propagation of a downwelling Kelvin wave was observed (‘C’ in Figure 3b), which was associated with westerly wind anomalies around 60°E to 70°E. While there were the easterly wind anomalies in the eastern region that could reduce the deepening tendency of the thermocline, the downwelling Kelvin wave reaches to 90°E, resulting in 20 m depression of D20 in July. The unique subsurface structure in July, i.e. the negative anomalies in the upper thermocline appeared with the positive anomalies below the depth of 120 m (Figure 2b), could possibly be produced by the mixed influences of the remote and local anomalous wind forcing.

[17] Another event of the basin-wide easterly wind anomalies occurred subsequently in late-July and August. The ocean responses again showed the upwelling Kelvin wave signal (‘D’ in Figure 3b) with the strong westward current anomalies down to 200 m depth (Figure 2d). A sequence of the above mentioned intraseasonal events of the equatorial waves contributed to the overall shallowing tendency of D20 during the first half of 2006, therefore, to a precondition for the IOD event. The atmospheric forcing in late-July to August is believed to be crucial for the occurrence of the negative SSTA in the eastern Indian Ocean, hence the onset of the positive IOD in 2006.

[18] After the intraseasonal westerly wind event in mid-September, the basin-wide easterly wind anomalies revived and continued about three months until early December with the maximum value of more than 7 m/s, causing to reversal of the climatological westerly winds. The positive and negative SSHA to the west and east of around 85°E, respectively, also persisted from October to December. Together with the observed surface westward current and the negative temperature anomalies in the upper layer at 90°E, typical atmosphere and ocean conditions for the positive IOD was established during this particular period in 2006.

[19] During the termination period in December 2006 and January 2007, the easterly wind anomalies weakened and the positive SSHA appeared in the far eastern basin. Unlike the SSTA change, the SSHA and subsurface temperature anomalies associated with the positive IOD event continued several more months at least to the end of February 2007 (Figures 2 and 3). This different evolution in the surface and the subsurface anomalies suggests that the heat exchanges by the surface heat flux are important during the termination phase of the positive IOD [Behera et al., 1999; Tokinaga and Tanimoto, 2004]. In addition, upward phase shift of subsurface eastward current anomalies from September to December (Figure 2d) might play significant role in the termination of the IOD, possibly through changes in the zonal heat advection.

[20] It should be recalled here that the strong ocean-atmosphere feedback processes appeared from August. A key factor that determines the timing of this IOD onset is the background conditions associated with the monsoonal winds. In usual August, strong southeasterly monsoonal wind prevails in the eastern Indian Ocean. It generates strong upwelling along the coasts of Sumatra and Java, by which the thermocline is uplifted and the subsurface conditions are brought to the ocean surface. The SST in this region, therefore, is sensitive to the subsurface ocean conditions and the wind forcing after summer monsoon onset [Annamalai et al., 2003]. In addition, as anomalous southeasterly winds intensify the climatological wind speed in August, surface cooling due to latent heat flux is enhanced and, therefore, the negative SSTA is generated [Tokinaga and Tanimoto, 2004]. Both factors can contribute to the rapid growth of the negative SSTA and to the development of the positive IOD in August 2006. Recently, Vinayachandran et al. [2007] reported the evolution processes of the 2006 IOD reproduced in their ocean general circulation model forced by QuikSCAT wind. Their heat budget analysis also supported that the vertical process became significant in August.