Contents Introduction I. Annual and Semiannual Variations I.1. The Northern Tropical Indian Ocean I.1.1. The Somali Current I.1.2. The Arabian Sea I.1.3. The Bay of Bengal I.2 The Indonesian throughflow I.2.1. Geography of the throughflow I.2.2. The Driving Force I.2.3. Volume Transports I.2.4. The Zonal Jet I.3. The Equatorial Indian Ocean I.3.1. Equatorial Currents I.3.2. Equatorial Waves I.4 The South Indian Ocean I.4.1. The South Equatorial Current I.4.2. The East African Coastal Current I.4.3. The Mozambique Channel and the Agulhas current II. Interannual variability and ENSO-related anomalies II.1. The Southern Oscillation and the QBO II.2. El Niño in the Indian Ocean II.3. Monsoon and ENSO III. Applications References

Due to the disappearance of the Coriolios force at the equator as well as its variation with latitude, equatorial areas are distinctive in two ways. First, zonal momentum transferred from the atmosphere to the ocean concentrates on the equator to form a narrow accelerating jet. Second, the equator acts as a wave guide.

In the Indian Ocean, low-frequency waves and semi-annual formation of an equatorial jet are prominent features. These phenomena are related to the seasonnaly varying monsoon circulation. However, equatorial upwelling does not occur in the Indian Ocean as it does in the Pacific and Atlantic, because in those oceans, upwelling occurs as a result of the south-east trade winds blowing across the equator and casuing surface divergence. Since in the Indian Ocean, there is no such wind system in either season of the year, typical conditions for equatorial upwelling are missing.

Most of the measurements were collected during the Indian Ocean Experiment (INDEX - 1976-1979). A long term series of measurements was also taken near the Addu Atoll, at Gan (Knox, 1976). Results confrim the idea that the semiannual zonal mass redistribution that take place in the Indian Ocean are effected almost entirely by the near-equatorial currents. Currents of the equator have no measurable role in this process (Wyrtki, 1973 ; Reverdin and Cane, 1984).

I.3.1. Equatorial Currents

(i) the monsoonal surface currents

Currents in the Indian Ocean are stronger than in the Pacific or Atlantic Oceans during most of the year and and can reach 0.8 m.s-1 . The equatorial currents are composed of three currents :

• the eastward North Equatorial Current (NEC) between 5°N and 2°S, which reverses during the Southwest monsson to become the Southwest Monsoon Current ;
• the Equatorial Countercurrent (ECC) flowing between 2°S and 8°S , which weakens during the Southwest monsoon and combines with the Southwest Monsoon current mentioned above;
• the Southequatorial Current (SEC) which occupies the region south of 8°S with velocities rarely exceeding 0.3° m.s-1. Its flow is stronger during the Northeast monsoon than during the Southwest monsoon.

(ii) the Inter-monsoon Jets

Wyrtki (1973) describes the presence of a surface eastward jet during the transition periods (April-May and October-November) from the northeast to the southwest monsoon with velocities of 0.7 m.s-1 or more. The Equatorial jets are approximately 500 km wide, equatorially-trapped, to the east. The jet was also observed in drifting-buoys measurements, by Knox (1976), Gonella et al., (1982) and Reverdin and Luyten (1986).

O'Brien and Hurlburt (1974) produced similar features in a two-layer numerical simulation, spun-up from rest by a constant wind.

(iii) the Undercurrent

The first direct observations of the undercurrent were made during the International Indian Ocean Expedition (1959-1965). At that time it was seen to be an ocean-wide feature, but later observations showed that this may not have been typical. In general, in the Indian Ocean, the undercurrent seems to be stronger, more long-lived flow in the western part of the ocean than in the central or eastern parts.

Leetma and Stommel (1980) used vertical profiles of currents, temperature, salinity, taken during INDEX, to obtain an expanded time and space coverage of the equatorial undercurrent. The undercurrent is usually associated with the persistent trade winds from the east. This westward stress at the surface, and corresponding difference in elevation of the sea surface, is balanced by a zonal pressure gradient between the boundaries. In the Indian Ocean such a zonal presure gradient (similar to those in the Pacific or the Atlantic), can only be established during the northeast monsoon.

Thus, the presence of the undercurrent is related to the seasonal change of wind stress. Observations during the northeast monsoon indicated a weaker undercurrent than in the Pacific. With the onset of the southwest monsoon, the undercurrent finaly disappears.

I.3.2. Equatorial Waves in the Indian Ocean

The complexity of the surface circulation of the Indian Ocean - which contrasts with the relative simplicity of the gyral system of the Pacific and Atlantic Oceans - is a result of the frequency and rapidity with which the overlying wind system changes. Wind speed and directions change so fast that there is not always time for the upper ocean to adjust so that it is in equilibrium with the wind, because the ocean's response time, or its " memory ", is many times longer than that of the atmosphere. How is it then possible for the ocean to react as quickly as it does ? The answer lies in the generation of waves.

(i) Rossby and Kelvin Waves

Waves transmit the effects of changes in the overlying wind field, or perturbations due to coastal boundaries, from one region of the ocean to another, much faster than would be possible simply through transportation of water in wind-driven currents (3 to 10 times quicker for Kelvin waves). Luyten and Roemmich (1982) have analyzed current-meter observations that were taken in the western equatorial Indian Ocean from April 1979 to June 1980. The moorings were mainly on the equator between 50°E and 62°E, at nominal depths of 200, 500 and 750m.

The most energetic signal at all depths was a semiannual oscillation in the zonal velocity. A 180 day period, 15 cm.s-1 amplitude, upward phase propagation and westward phase propagation of zonal velocity were all measured directly. Measured vertical and zonal wave number and the ratio of kinetic to potential energy suggest a mixture of an equatorial Kelvin wave and a long equatorial Rossby wave of first meridional mode. These have zonal wavelengths of 24 000 and 8000 km respectively. The equatorial trapping scale is about 400 km. A downward energy flux is noticed (Luyten and Roemmich, 1982).

Luyten and Roemmich (1982) also found a strong signal in the 200m deep meridional velocity spectrum, with a peak at 26 days, which is not observed in the zonal velocity. This result corresponds to the mixed Rossby-gravity wave (or Yanai wave) and will be discussed later in the next chapter.

The authors raise a certain number of questions :

• Is the energy input by the semiannual wind equally divided between Kelvin and Rossby components ?
• What happens when the waves reach steeply angled coast-lines ? For example, at the western boundary, how much of the incident Rossby wave energy is reflected in the Kelvin mode, and how much of it is short Rossby waves which may be quickly dissipated in the boundary region ?

Gent et al. (1983) used a simple numerical model, with linear physics and a surface wind forcing at a single frequency, namely the semiannual component of the zonal wind stress of Hellerman and Rosenstein (1983) data set, and simulated quite well the observations of Luyten and Roemmich (1982). An important conclusion is that the form of the wind stress polewards of about 4° has very little influence on the solution at the equator. However, as is to be expected, the solution at the equator is very sensitive to the form of the zonal component of wind stress along the equator. This is in agreement with COADS (Comprehensive Ocean Atmosphere Data Set) Sea Surface Temperature (SST) observations at the equator, which variability is in phase with the zonal wind at the equator (with the wind signal slightly leading the SST signal ; Le Blanc, 1996). Since SST at the equator, seems to be mainly advected by wind-driven currents (Barnett, 1984 ; Le Blanc, 1996), it can be taken as a tracer of surface currents in this region. This correlation between zonal wind and SST is not as good in other regions below 5°S (Le Blanc, 1996). However, it is not sure wether the strong semiannual variations of zonal current along the equator influence SST either directly or by exciting Rossby waves at the eastern boundary (Godfrey et al. ; 1995).

Further tests by Gent et al. (1983) show that simplified solutions having just the second vertical mode, or non-reflecting lateral boundaries do not approximate the full solution, nor simulate the observations, very well. The simplified solution retaining just the Kelvin and first Rossby waves works reasonably well on the equator, and hence in simulating the observations, but cannot be good away from it. This is in agreement with Topex/Poseidon observations made by Boulanger (personal communication) which first show the existence of Kelvin and Rossby waves and second that these are reflecting on the western and eastern boundaries of the Indian Ocean.

Gent et al. 's (1983) final conclusion is that " in this realistic problem, where the lateral boundaries are important and the forcing has a fairly rich wave-number spectrum, a description in terms of a few vertically propagating waves is not possible. "

Rossby waves have also been observed through drifting buoys measurements. Six buoys were launched during the Global Atmosphere Research Program (GARP) in May and June 1979, just before the onset of the Southwest monsoon. All the buoys immediatly accelerated eastward and were located between 3°N and 3°S ; two buoys reached the coast of Sumatra and the others slowed down and reversed direction before reaching the coast. When the first one reversed, the others were still proceding eastward and the later they would reach the coast, the earlier they would reverse. It is therefore unlikely that the change in direction was caused by a relaxation of the eastward winds. The rapid deceleration of the buoys occured at increasing distances from the coast strongly suggesting that the trajectories were affected by a disturbance propagating westward at approximately 55cm.s-1. According to models (O'Brien and Hurlburt, 1974 ; Cane, 1979 ; Philander and Pacanowski, 1980), the equatorial jet at first accelerates after the sudden onset of the winds, but a Kelvin wave generated at the western extreme of the forced region arrests this acceleration. This wave can propagate accross the wind-forced basin in the time it takes the wind to attain full strength. Equatorially trapped Rossby waves excited at the eastern extreme of the forced region also decelerate the jet, but move at only one-third the speed of the kelvin waves. These results suggest that the deceleration of the drifter buoys could have been caused by the equatorially trapped Rossby wave (Gonella et al., 1982).

(ii) the Yanai wave

As demonstrated previously, equatorially trapped waves are of major importance in the problem of equatorial ocean adjustment to unsteady wind forcing. Among the various types of equatorially trapped waves (the Kelvin wave, the Rossby wave, the inertia-gravity wave), the Yanai wave - or mixed Rossby-gravity wave - is unusual because it has no counterpart at higher latitudes. The Yanai wave and the Kelvin wave arise because the Coriolis acceleration changes sign at the equator. In the Indian Ocean, the Yanai wave was first observed by Luyten and Roemmich (1982) as mentioned previously. The zonal wavelength of this oscillation is estimated to be approximately 1000 -1300 km, with meridional velocities of 0.15 m.s-1 to 0.3 m.s-1. In addition, this oscillation exhibits a westward propagation, with eastward and downward energy propagation.

The group velocity of the Yanai wave is always positive (wave energy propagates eastward), but the phase can travel east or west depending on the frequency. Unlike the Kelvin wave, the Yanai wave is antisymmetric about the equator.

Reverdin and Luyten (1986), examined the data obtained from drifting buoys that were released in the western Indian Ocean, at the equator, between 50°e and 60°E from 1979 to 1982 (during the SINODE : Surface Indian Ocean Dynamic Experiment). The authors review the meanders indicated by buoy trajectories at the end of the Southwest monsoon. Large scale meanders of the Surface Equatorial Current in the eastern Atlantic and Pacific Oceans are well-known features. They have periods of 25 days and wavelengths of the order of 1200 km as has been first suggested by infrared observations of the equatorial Pacific (Legeckis, 1977). It is generally believed that these waves derive their energy from the shear between the South Equatorial Current and the North Equatorial Countercurrent. A contribution from the shear between the Equatorial Undercurrent and the South Equatorial Current has also been mentioned by Cox (1980).

In the eastern Atlantic and Pacific Oceans, there are strong meridional SST gradients near the equator - due to equatorial upwelling - so that meridional displacements have a strong SST signature, easily monitored by remote temperature sensing (Legeckis et al., 1983). In the Indian Ocean, thermal fronts are less contrasted, so that meridional motions cannot easily be inferred from satellite-based SST imagery. In that area, other measurements are needed to investigate oscillations close to the surface.

Reverdin and Luyten (1986) conclude that, " although the waves below the thermocline bear a close resemblance to the situation in the other oceans, the surface expression is very different. Here, most of the buoys are drifting towards the east between 0° and 5°N, whereas in the other oceans they are mainly drifting toward the west in this latitude band. The oceanographic pecularities of the upper western equatorial Indian Ocean are a deeper thermocline and a more intense seasonal cycle than in the eastern parts of the other oceans. Also, it is unlikely that the source of the meanders can be found without attention to the intense seasonally varying circulation near the Somali coast. [...]. It is not unlikely that the appearance of the equatorial meanders is associated with the northward displacements of the eddy system along the coast occuring at the end of July, early August. (see section I.1.1.).

Kindle and Thompson (1989), Moore and McCreary (1990), and Woodberry et al. (1989) also suggest that a possible source of these 26-day Yanai waves can be the western boundary. Kindle and Thompson (1989) used a nonlinear, reduced gravity model with realistic geometry of the Indian Ocean basin. This model is forced with climatological winds of Hellerman and Rosenstein (1983). Yanai waves with a period of 26 days were observed in their model simulations : the model shows that these waves were formed within 1400 km of the western boundary, with the initial group of wave packets generated during late July-August. They attributed the generation of the Yanai waves to "  instability associated with the circulation of the southern gyre during the latter stages of the southwest monsoon.  " This result is in agreement with Jensen's (1991) conclusion mentioned in section I.1.1.

Yanai waves were also found in the Indian Ocean modeling by Woodberry et al. (1989). The waves had a period of 28 days and could be seen emanating from the western boundary region. The strongest and most coherent signals of these waves were seen in the region between 50°E and 65°E, during July, August and September.

Those results motivated Tsai et al. (1989) to study satellite-derived SST measurements in the western equatorial Indian Ocean, in order to detect the signal of the equatorially trapped waves in the fluctuation of the SST data. For example, a downwelling Kelvin wave travelling along the equator will depress the thermocline ; consequently, the mixed layer becomes deeper and an anomalously warm surface water is formed. Hence changes in SST are indicative of vertical motions of the thermocline, which in turn can be related to the motion of propagating internal waves. Observation confirms this theory and show a clear correlation between COADS SST annual and inter-annual variability and Ekman pumping estimated from the Florida State University (FSU) pseudo-wind-stress field, in the Indian Ocean equatorial band but these data set are monthly means and do not allow the study of oscillations with periods shorter than two months (Le Blanc, 1996).

Tsai et al., (1989) show that there is a preferred oscillation in SST data with a period of about 25-27 days. This oscillation is antisymmetric about the equator. It is unlikely that this 25-27-day oscillation is a result of direct wind forcing, since the dominant period of atmospheric oscillation in the equatorial region is of 40 to 60 days. Further more, linear wave theory strongly suggest that the Yanai wave is responsible for the 25-27-day oscillation observed in the SST data.

From their data, it appears that the 26-day oscillation in SST data occurs only in the western Indian Ocean from 52°E to 60°E. Farther away from the western boundary, this signal is not seen. This is reasonable because they expect other physical effects such as dissipation and advection and interference with the land barrier of the Maldives Islands to weaken the signal, as the wave-energy propagates eastward. This can actually be seen in shallow-water modeling of the Indian Ocean, forced by a single wave (personal results).

(iii) Future work

A few questions still require further study in order to find answers namely :

• Do equatorially-trapped waves only adjust the upper-layers of the ocean to changes in winds or do they also play a role in generating climate anomalies in the Indian Ocean and the global climate system ?
• If so, is there a correlation between their forcing and climate anomalies, either regional or global ?
• How are waves affected after the passage of the Maldives Islands ? What signal reaches the eastern coast ?
• Does the region north of the throughflow form a separate dynamical regime ? That is, are Kelvin and Rossby waves " stopped " by the Indonesian passage (Godfrey et al., 1995) ?