Intraseasonal variability in the upper layer currents observed in the eastern equatorial Indian Ocean

1. Introduction

[2] In the tropical Indian Ocean, monsoon winds prevail over most of the region, leading to a dominant annual signal in surface current variability, with some regional currents reversing twice a year. Within a zonal band of a few degrees centered at the equator in the open ocean, there is a distinct semiannual signal of the swift eastward upper ocean currents, known as the Wyrtki jets or equatorial jets. Since the seminal work of Wyrtki [1973], who analyzed observed monthly surface currents to show the eastward equatorial jets along the equator in the Indian Ocean, it has been widely accepted that the jets occur twice a year during the monsoon transition periods in April/May and October/November, with a typical amplitude of the surface zonal current of about 80 cm/s or more on the equator [Han et al., 1999]. These Wyrtki jets have potential importance to the upper layer variability in the tropical Indian Ocean through zonal redistribution of the water mass, heat, and salt, which can modify the surface temperature of the warm water pool in the eastern tropical Indian Ocean in many ways. A change in the surface temperature there can result in variations of the air-sea interaction, thus affecting both the regional and global climate system.

[3] Detailed structures, generation mechanisms, and propagation characteristics of the Wyrtki jets have been investigated in several previous works using theoretical and numerical models [O'Brien and Hurlburt, 1974; Anderson and Carrington, 1993; Han et al., 1999]. The results indicate that the jet is a basic upper ocean response to the semiannual westerly winds over the equatorial region and can be interpreted as local Yoshida-jet dynamics [Yoshida, 1959; O'Brien and Hurlburt, 1974] and subsequent eastward propagation of the equatorial Kelvin wave [O'Brien and Hurlburt, 1974; Anderson and Carrington, 1993]. Cane [1980] pointed out the importance of nonlinearity in the dynamics of these jets, especially for differences in the ocean response to the westerly and easterly winds. Although observational evidence based on in situ data for such dynamical events in the equatorial Indian Ocean are limited due to scarcity of data, several direct current measurements have been conducted in the central and western equatorial Indian Ocean [Knox, 1976; McPhaden, 1982; Luyten and Roemmich, 1982; Reppin et al., 1999]. The data show complex variability in the zonal and meridional currents at several different timescales including the semiannual, annual, and interannual, as well as some evidence on the variability at intraseasonal timescales. In particular, in addition to the strong annual and semiannual signals, Reppin et al. [1999] shows a distinct spectral peak at 15.5 days in the surface currents, observed by current-meter moorings with acoustic Doppler current profilers (ADCP) at 80.5°E in the equatorial region.

[4] Studies of the surface winds over the equatorial Indian Ocean show that there is large variability in the zonal winds not only at the annual and semiannual timescales but also at intraseasonal timescales [e.g., Hendon and Glick,1997; Webster et al., 2002], associated with the Madden-Julian oscillation [Madden and Julian, 1971]. Given the basic dynamical response of the ocean to atmospheric forcing mentioned above, the eastward surface jets along the equator in the Indian Ocean may have strong intraseasonal variability. Indeed, results from the Joint Air-Sea Monsoon Interaction Experiment (JASMINE) cruises during April and May 1999 in the eastern equatorial Indian Ocean indicate abrupt change in the upper ocean velocity field within a few weeks [Webster et al., 2002]. Such cruise observations by research vessels, however, obtain only snapshots of the variability. Thus, time series observations of the upper ocean currents are required for detailed analysis of the intraseasonal variability in the equatorial Indian Ocean. Since the previous direct current observations were located in the central and western Indian Ocean and mainly focused on timescales at and longer than the semiannual period, the shorter timescale variability of the equatorial jets in the warm water region of the eastern Indian Ocean has remained an unsolved issue.

2. Current Observations

[5] To advance our understanding of the characteristics of the short-term variability in the equatorial eastern Indian Ocean, an upward-looking ADCP mooring was deployed at 90°E on the equator on November 14, 2000 (Figure 1). The 75-kHz self-contained Workhorse Long Ranger ADCP, RD Instruments, and a conductivity-temperature-depth (CTD) sensor, Sea-Bird Model SBE37, were deployed to be located at the depth of about 400 m. The ADCP measured the vertical profile of the velocity every one hour, averaging 27 pings at 6.7-s intervals to obtain 3-minute average. The vertical profile was binned using 8-m intervals from the surface to the depth of the instrument. However, because the surface layer data shallower than 40 m are contaminated with the signals reflected at the surface, we analyzed the data between the depths of 40 m and 400 m in the present study. The CTD measured the temperature, salinity, and pressure at the depth of the instrument every hour. The mooring was recovered on October 22, 2001, and the time series of the temperature, salinity, pressure, and the vertical profile of the velocity for about 11 months were obtained. The short gaps in the data were linearly interpolated in time. The pressure data obtained by the CTD were used to calibrate the depth of the instruments, and indicated no significant change in the depth due to the large swings of the mooring line. In addition to the above measurements, temperature and salinity profiles were observed in the upper 2000 m at the mooring site during the deployment and the recovery of the mooring, using the CTD aboard the R/V Mirai.

3. Results

[6] Figure 2a shows the zonal wind stress time series which was obtained from the SeaWinds instrument on the Quick-SCAT satellite, and was averaged between 80°E and 90°E along the equator. The record is dominated by strong intraseasonal variability superposed on the mean eastward winds. Figures 2b and 2c show the time series of the observed zonal and meridional currents, respectively, to which a low-pass filter (cut-off timescale of 5 days) was applied to remove higher frequency variability. We can identify the layered structure in the zonal current: the upper layer current shallower than about 120 m, the Equatorial Undercurrent in the layer between 120 m and about 300 m, and deeper currents below 300 m. Similar structures, on the other hand, cannot be identified in the meridional current, which is dominated by rather short-term variability over the whole range of observed depths.

[7] Strong vertical shear in the zonal current is observed at the depth of about 120 m (Figure 2b), the depth corresponds to the upper part of the pycnocline as derived from the density profiles taken at the observed site during the deployment and recovery of the mooring. This strong vertical shear region gradually rises from November 2000 to April 2001, and again from August to October 2001.

[8] Just below the pycnocline is a layer of significant semiannual signal corresponding to the Equatorial Undercurrent in the Indian Ocean. The flow is eastward from December 2000 to April 2001 and from August to October in 2001 and westward during the other periods. Since the complicated variability in the layer above the pycnocline, only limited portions of the eastward period show a typical vertical structure of the Equatorial Undercurrent; i.e., the westward surface current on top of the eastward undercurrent. In this layer there is clear upward phase shift of the zonal current cores over the depth range from 300 m to the pycnocline, suggesting the importance of equatorial wave propagation, as found in the previous observations [e.g., McPhaden, 1982; Reppin et al., 1999].

[9] The zonal current below about 300 m flows in the opposite direction to that of the Equatorial Undercurrent for the periods from November 2000 to January 2001, May to July and September/October in 2001, showing a three-layer structure in the zonal current field. During the other periods, upward phase shifts from the depth of the ADCP to the level of the Equatorial Undercurrent are clearly observed. The distinct signal of this phase shift appears from April to May 2001.

[10] In the layer shallower than about 120 m, shorter timescale variability is observed throughout the whole period of observation, showing the alternating eastward and westward currents with a period of about a month. Large events of the eastward current greater than 40 cm/s appear in November/December 2000, April/May, July/August, September, and October 2001, some of which occur during the monsoon transition periods of the classical Wyrtki jets. However, the duration of the strong eastward currents is confined at most to about 20 days, and even in that period, the eastward jets consist of several strong cores of only 10 days. This characteristic of the variability can also be seen in the weaker events of the eastward current as well as the events of the westward current. A spectrum analysis of the zonal current variability confirms that the dominant period is 30 to 50 days down to the depth of 390 m, although most of the energy is confined to the layer shallower than 120 m (Figure 3a).

[11] To compare interpretations based on monthly mean data (as by Wyrtki [1973]) with the present observations, we have calculated the simple monthly mean value of the zonal current of Figure 2b and plotted it in Figure 2d. There is a distinct semiannual signal at all depths from 40 m down to 400 m, although the phase of the signal changes with depth, indicating the strong baroclinic structure. In the surface layer, the relatively strong eastward flows occur during November/December 2000, April/May and September/October 2001. These flows are consistent with the classical view of the semiannual Wyrtki jets. However, the maximum zonal current associated with the semiannual Wyrtki jets (∼30 cm/s) is less than half of the maximum zonal current of about 70 cm/s associated with the intraseasonal variability, which is smoothed out by monthly averaging in Figure 2d. The westward flow appears only from January to March 2001, at the depth of 40 m, as has been observed during this season previously [Hacker et al., 1998].

[12] The intraseasonal variability in the zonal current observed in the surface layer is thought to result from the wind-forced variability associated with the intraseasonal disturbances in the atmosphere, as shown by the time series of the zonal wind stress averaged from 80°E to 90°E on the equator (Figure 2a). To quantify the association between the wind and current variability, coherence analysis was applied to the zonal current variability at the depth of 40 m and the surface zonal wind stress variability at each longitude along the equator. Large coherence (>0.8) is observed between the zonal current variability and the wind stress variability from 80°E to 90°E for the periods from 30 to 40 days. This suggests that the intraseasonal variability in the zonal current at 90°E is mostly driven by the zonal winds between 80°E and 90°E, including both the local and remote responses to the wind stresses. This is also consistent with the fact that, since only the first baroclinic Rossby wave is available at the period of 30 to 40 days, the higher order baroclinic response consists of Kelvin waves and the local response to the zonal winds [Han, 2005].

[13] The meridional current, on the other hand, shows higher frequency variability than that of the zonal current (Figure 2c). Note the complicated vertical structure with rather strong flow occurring sporadically in the layer deeper than 200 m, suggesting the existence of vertically coherent responses. It is difficult from this figure to detect the clear semiannual or annual signals. The distribution of the power spectrum density of the meridional current (Figure 3b) indicates that the large amplitude variation is located in the period between 10 and 20 days in the layer shallower than 100 m. The period of the maximum variability tends to shift to longer timescales in the deeper levels, with a period near 30 days at the depth of 200 m. Reppin et al. [1999] demonstrates the existence of the 15-day variability in the meridional current at 80.5°E at the equator observed by the ADCP mooring. They also report that the period of the intraseasonal variability is broadened to a range of 10–50 days in the deeper layer, which is a similar period-range found in our observations. Recently, it has been suggested by D. Sengupta et al. (A biweekly upwelling mode in the equatorial Indian Ocean, submitted to Geophysical Research Letters, 2004) that the biweekly fluctuation in the eastern equatorial Indian Ocean has the structure of mixed Rossby-gravity waves forced by the wind stress curl over the region.

4. Summary and Discussion

[14] New direct observations of the upper ocean currents in the eastern equatorial Indian Ocean indicate that the classical Wyrtki jets are the monthly averaged view of the highly variable zonal current field, whose dominant period is 30 to 50 days associated with the atmospheric wind variability. Although it is difficult to compare directly to the surface current data observed previously, the least-square fit of the semiannual harmonic to the zonal current variability at 40-m depth gives the amplitude of about 10 cm/s. On the other hand, the amplitude of the intraseasonal variability reaches more than 50 cm/s at the same depth. Not only the transition periods of the monsoons, but also other times of the year, have energetic intraseasonal activity of the upper layer zonal current at 90°E. It is this intraseasonal variability that obscures the semiannual and annual signals in the upper layer of the eastern equatorial Indian Ocean. However, due to the seasonal march of the large-scale wind fields associated with the monsoon over the equatorial region, the eastward jets tend to be stronger during the monsoon transition periods. The responses in the upper ocean in the central and western equatorial Indian Ocean, on the other hand, may be quite different as suggested by Han [2005].

[15] The above results on the variability in the eastern tropical Indian Ocean may have several implications for the air-sea interaction processes there, hence an impact on both the regional and global climate system. For example, the generation of the barrier layer and its maintenance depend strongly on the zonal intrusion of the eastward jets into the eastern equatorial Indian Ocean [Masson et al., 2002]. The atmospheric intraseasonal disturbances may be influenced by the sea surface temperature change due to the oceanic responses excited by the winds associated with the atmospheric disturbances. Further, the seasonal and interannual modulation of the intraseasonal variability may affect the degree of coupling between the atmosphere and ocean on longer timescales, such as for the Indian Ocean Dipole events [Saji et al., 1999].

[16] Recent deployment of two TRITON (TRIangle Trans-Ocean buoy Network) buoys at (1.5°S, 90°E) and (5°S, 95°E) now provides us with high quality temperature and salinity data in the upper 750 m on a real-time basis [Kuroda, 2002]. Together with the velocity data from the ADCP mooring, our understanding of the heat and salt budgets in the eastern Indian Ocean will be advanced. However, to examine the detailed structure of the equatorial waves and basin wide phenomena, such as the Indian Ocean Dipole and the local (Indian Ocean) effects of ENSO from the Pacific Ocean, and to investigate mechanisms through which the intraseasonal and interannual variations are linked, long-term observation of the upper ocean conditions by an extensive mooring array spanning the tropical Indian Ocean may be needed.