The equatorial Indian Ocean (EIO) exhibited anomalous conditions characteristic of an Indian Ocean dipole (IOD) during 2006. The eastern EIO had cold sea surface temperature anomalies (SSTA), lower sea level, shallow thermocline and higher chlorophyll than normal. The anomalies in the east, restricted to the south of the equator, were highest off Sumatra. The western pole of the IOD was marked by warm SSTA and deeper thermocline with maxima on either side of the equator. An ocean general circulation model of the Indian Ocean forced by QuikSCAT winds reproduces the IOD of 2006 remarkably well. The switch over to cooling in the east and warming in the west happened during May and July respectively. In the east, air-sea heat flux initiated cold SSTA in the model which were sustained later by oceanic processes. In the west, surface heat fluxes and horizontal advection caused warm SSTA and contribution by the latter decreased after August.
 The thermal structure of the equatorial Indian Ocean (EIO) is characterized by warmer (cooler) temperature in the east (west) consequent to the upwelling along the western boundary and strong eastward equatorial jets that transport warm upper layer water to the east. During certain years, this pattern switches to anomalous conditions known as the Indian Ocean Dipole (IOD) [Saji et al., 1999] or Zonal Mode [Webster et al., 1999; Annamalai et al., 2003]. Positive IOD events are associated with cooler (warmer) than normal sea surface temperature (SST) in the eastern (western) EIO, accompanied by shallower (deeper) thermocline and mixed layer. Easterly wind anomalies in the EIO and alongshore wind anomalies off Sumatra are prominent during the IOD. The wind anomalies are accompanied by suppression (enhancement) of convection [Gadgil et al., 2004] in the east (west). IOD affects the climate of not only in the Indo-Pacific region but also in places far away from its oceanic focal point [Saji and Yamagata, 2003]. IOD has occurred in concert with El Niño as well as independently [Yamagata et al., 2004] suggesting that air-sea coupling local to the Indian Ocean are capable of generating the sequence of events leading to the formation of IOD. Significant positive IOD events have occurred during 1961, 1963, 1967, 1972, 1977, 1982, 1994 and 1997 [Yamagata et al., 2002] and here we show that one such event occurred during 2006.
 The Indian Ocean model used in this study is based on the Geophysical Fluid Dynamics Laboratory, Modular Ocean Model Version 4 [Griffies et al., 2004]. The model domain covers the Indian Ocean between 30°S–30°N and 30°–120°E, with a horizontal grid spacing of 0.25° and 40 vertical levels. The model topography is based on 5 minute resolution ETOPO5 dataset with the minimum depth of the ocean as 30 m. At the land boundaries on the western and northern side of the model domain a no-slip boundary condition has been applied. Sponge layers of width 3.5° have been applied on the open boundaries at south and east where temperature and salinity are restored to climatological values [Locarnini et al., 2006; Antonov et al., 2006] with a time scale of 30 days. Vertical and horizontal mixing are based on those of Large et al.  and Chassignet and Garraffo  respectively. Sensible and latent heat fluxes as well as upward longwave radiation are calculated within the model as a function of model SST. Penetrative shortwave radiation is calculated according to Morel and Antoine  and the river runoff is discharged into the upper 15 m of the model. Further details of the model configuration as well as the evaluation of its performance are given by Kurian and Vinayachandran [2006, 2007] and Vinayachandran and Kurian .
 First, the model was spun-up for a period of seven years using climatological forcing. The climatological simulation by the model, particularly the SST is found to be in excellent agreement with observations [Kurian and Vinayachandran, 2007]. The results presented here are from the QuikSCAT run (QR) for the period 24 July 1999 to 31 December 2006. For a smooth transition from climatology to QuikSCAT period, the model was started from the eighth year climatology and run from March 1998 to 23 July 1999, forced by daily winds, downward shortwave and longwave radiations, air temperature and specific humidity from NCEP [Kalnay et al., 1996], weekly chlorophyll from SeaWiFS, and rainfall from TRMM. The NCEP winds were replaced by QuikSCAT winds in the QR run. All anomalies presented here are calculated with respect to the mean for the seven year period 2000–2006.
3. Anomalous Conditions in 2006
 The southeasterly trades in the southern Indian Ocean were unusually strong during the month of February particularly to the south of Sumatra which weakened by March and there were westerly wind anomalies to the west of Sumatra and to the east of about 80°E during March and April (Figure 1a). Easterly wind anomalies, typical of the IOD, which appeared along the EIO during May were observed till the end of the year, although with varying intensities (Figures 1c and 1d). There were upwelling favorable alongshore winds off the coast of Sumatra beginning from the summer monsoon month of June. During the Fall, the wind pattern was dominated by the easterly anomalies on either side of the equator (Figure 1c).
 Cold SST anomalies (SSTA) of about 0.5–1°C appeared to the south of Java during June. This anomaly patch grew in size and spread northward along the coast of Sumatra during the July–September period (Figures 2b, 2d and 2f). The maximum of this SSTA patch was located to the east of Sumatra (Figure 2) during September and had values exceeding 2.5°C. These anomalies were also found to extend westward between the equator and about 5°S. Warm SSTA of the order of about 0.5°C persisted in the western EIO during June through August, weakened in September and intensified to about 1°C during October. A well defined dipole pattern in SSTA was present in the EIO during October–December. The model is able to reproduce the evolution of the dipole pattern of SSTA remarkably well (Figures 2a, 2c and 2e). The difference between the observed and simulated SST was found to be generally within 0.5°C. Close to the Sumatra coast, however, during the peak phases of the IOD, this difference was as high as 1°C.
 The expression of the IOD was also seen in other physical variables and in sea surface chlorophyll concentration. The eastward flowing Wyrtki jet which influences the slope of the thermocline in the EIO was absent during the fall of 2006 (Figure 3a). The model currents show that the eastward spring jet was stronger but the flow was westward during August–December, the latter being opposite to the normal conditions. Consequently, the sea level (Figure 3b) remained lower than normal in the east and higher in the west, coinciding with regions of SSTA. The model simulation show that the thermocline shallowed in the eastern EIO (Figure 3c) in response to easterly wind anomalies. The thermocline anomalies in the west were marked by two regions of anomalously deep thermocline on either side of the equator, to the west of India, around Lakshadweep–Maldives island chain. Both shallow thermocline anomalies in the east and deep anomalies in the west propagated westwards, typical of Rossby wave propagations during IOD. Maximum deep thermocline anomalies were co-located with SSTA, during the peak of the IOD, suggesting the influence of ocean dynamics. The impact of the IOD is also seen in the large deficit in rainfall in the eastern EIO and excess in the west (Figures 1c and 1d). The anomalies of net freshwater flux caused positive salinity anomalies off Sumatra and negative anomalies in the central EIO (Figure 3e). The impact on marine biology is captured by SeaWiFS data which show unusually large concentration of chlorophyll to the west of Sumatra (Figure 3d).
 Considering that SST is the single parameter that affects atmosphere directly, we have examined mechanisms that lead to the large SSTA. The terms of the temperature equation [Vialard and Delecluse, 1998; Kurian and Vinayachandran, 2007] for the upper 50 m were averaged over the dipole boxes [Saji et al., 1999] marked in Figure 2e and over two boxes (marked in Figure 2f) where anomalies were high during 2006. In the east, the cooling tendency began towards the end of May (Figure 4a) and continued till the end of October. This negative tendency, however, was interrupted by two periods of warming, one during July and the other during September–October. The onset of the cooling in the east was caused by the loss of heat to the atmosphere as suggested by air-sea fluxes (Figures 4a and 4b, green curve) that closely follows the temperature tendency during June–August. Vertical process (upwelling and entrainment) lead to significant amount of cooling during August–December and influenced the variations of the tendency significantly. The contribution by oceanic processes were responsible for the negative temperature tendency during the peak phase of the IOD. In the west, the warming tendency began in the month of July (Figures 4c and 4d). As in the east, the temperature tendency in the west was marked by intraseasonal variations; there were two brief periods of cooling, one during August and the other during September. Major contribution to the warming came from air-sea fluxes. Horizontal advection also added to the warming but its contribution reduced significantly after August. The main difference between the two western boxes is the negative temperature tendency during May–June seen in Figure 4c compared to the warming tendency seen in Figure 4d.
5. Summary and conclusions
 Satellite observations of SST, winds, rainfall, sea level anomalies and chlorophyll show that anomalies characteristic of an IOD developed in the tropical Indian Ocean during 2006. The eastern Indian Ocean, particularly off Sumatra, was anomalously cold and the western Indian Ocean warm. The SSTA were accompanied by alongshore wind anomalies off Sumatra, easterly wind anomalies along the equator, suppressed convection in the east and enhanced convection in the west. The sea level was lowered in the east and elevated in the west. The evolution of the dipole event of 2006 is remarkably well captured by an ocean model which is forced by the state-of-the-art forcing functions available today. The model simulation was found to be useful for tracking surface as well as subsurface variations in the ocean associated with the dipole. The evaluation of the temperature equation show that the air-sea fluxes during the spring months kicked off the cooling process in the SSTA in the east, which was later on sustained by oceanic processes dominated by coastal upwelling and warm SSTA in the west were triggered by air-sea fluxes and advection. Variations of SSTA were significantly influenced by oceanic processes during the peak phase of the IOD.
 The success of the model, despite its lack of an interannually varying Indonesian Throughflow, suggests that the oceanic contribution from the Pacific Ocean is not crucial for the dipole. The spread of the SSTA in the east appears to be somewhat different from that in the previous dipole occurred during 1997. In 2006, the anomalies were confined primarily to the south of the equator whereas westward extension along the equator was also seen during 1997 [Murtugudde et al., 2000; Vinayachandran et al., 2002]. The limited period of model integration in our study prohibits examination of the cause of this difference and understanding such differences may provide further insight into the evolution of coupled air-sea processes during IOD.