Abstract

The equatorial wave dynamics of sea level variations during negative Indian Ocean dipole (nIOD) events are investigated using the LICOM ocean general circulation model forced with the European Centre for Medium-Range Weather Forecast reanalysis wind stress and heat flux from 1990 to 2001. The work is a continuation of the study by Yuan and Liu, in which the equatorial wave dynamics during positive IOD events are investigated. The model has reproduced the sea level anomalies of satellite altimeter data well. Long equatorial waves extracted from the model output suggest two kinds of negative feedback during nIOD events: the western boundary reflection and the easterly wind bursts. During the strong 1998–99 nIOD event, the downwelling anomalies in the eastern Indian Ocean are terminated by persistent and strong upwelling Kelvin waves from the western boundary, which are reflected from the wind-forced equatorial Rossby waves over the southern central Indian Ocean. During the 1996–97 nIOD, however, the reflection of upwelling anomalies at the western boundary is terminated by the arrival of downwelling equatorial Rossby waves from the eastern boundary reflection in early 1997. Therefore, the negative feedback of this nIOD event is not provided by the western boundary reflection. The downwelling anomalies in the eastern basin during the 1996–97 nIOD event are terminated by easterly wind anomalies over the equatorial Indian Ocean in early 1997. The disclosed equatorial wave dynamics are important to the simulation and prediction of IOD evolution.

1. Introduction

The Indian Ocean dipole (IOD) events are strong interannual anomalous events over the tropical Indian Ocean (Saji et al. 1999; Webster et al. 1999). The positive IOD (pIOD) event is characterized with cooler sea surface temperature anomalies (SSTA), lower sea level, and shallower thermocline in the eastern equatorial Indian Ocean and warmer SSTA, higher sea level, and deeper thermocline in the western basin than those in normal years (Annamalai and Murtugudde 2004; Vinayachandran et al. 2009). The negative IOD (nIOD) event is the opposite, characterized by enhanced downwelling in the east, upwelling in the west, and stronger westerly winds driving warm surface water along the equator toward Indonesia (Vinayachandran et al. 2002). The SSTA patterns associated with IOD events have a significant impact on precipitation over the African, Asian, and Australian continents and on the prediction of tropical cyclones in the Indian Ocean (Mason et al. 1999; Li and Mu 2001; Behera et al. 2005; Iizuka and Matsuura 2012; Cai et al. 2012; Weller and Cai 2013; Saha and Wasimi 2013). Therefore, it is important to understand the dynamics of these nIOD events.

Oceanic equatorial waves are excited during IOD events, which are responsible for the evolution of sea level and temperature anomalies (Chambers et al. 1999; Le Blanc and Boulanger 2001). Several existing studies used to suggest that equatorial Kelvin waves forced by atmospheric intraseasonal oscillations (ISOs) terminate the 1994 pIOD event (Rao and Yamagata 2004; Rao et al. 2007). However, some other studies (Han et al. 2006, 2007; Thompson et al. 2009) suggest that ISOs are unlikely to play an important role during the termination of the IOD event because the atmospheric ISOs are either weak during the IOD events (e.g., the 1997 pIOD) or lag in the evolution of the SSTA (e.g., the 1994 pIOD). Yuan and Liu (2009, hereinafter YL09) have shown that atmospheric ISOs force upwelling Kelvin waves rectified through oceanic nonlinearities, which cannot be responsible for the termination of pIOD events. Instead, they suggest that the pIOD events be terminated by two Kelvin wave pulses from the western boundary reflected from equatorial and off-equatorial Rossby waves, respectively. Figures 11 and 12 show the schematics of the IOD negative feedback. The roles of the western boundary reflections are also indicated implicitly by Masumoto and Meyers (1998), Rao et al. (2002), Huang and Kinter (2002), Feng and Meyers (2003), Li et al. (2003), and Jury and Huang (2004). However, the equatorial wave dynamics responsible for the termination of downwelling anomalies in the eastern Indian Ocean during nIOD events have not been studied yet.

The objective of this paper is to study the long equatorial wave dynamics and the roles of western and eastern boundary reflections in the termination of the downwelling anomalies in the eastern Indian Ocean during nIOD events. We will use a numerical hindcast from an ocean general circulation model (OGCM) simulation to conduct the study. The model and data used in this study are described in section 2. The results associated with equatorial waves are analyzed in section 3. The equatorial wave dynamics of the sea level anomalies during a few nIOD events are discussed based on the propagation and reflection of the equatorial waves in section 4. The role of the boundary-reflected and wind-forced waves in the nIOD sea level evolution is investigated in section 5. Discussion and conclusions are given in sections 6 and 7, respectively.

2. Models and data

The OGCM used here is LICOM 1.0, which is developed by the Institute of Atmospheric Physics, Chinese Academy of Sciences. The model employs the η coordinates that represent a free sea level and approximate z coordinates in the deep ocean. The domain of the model covers the quasi-global ocean from 78.5°S to 90°N with a horizontal resolution of 0.5° longitude × 0.5° latitude. There are 30 stacks of boxes in the vertical with their middle depths at approximately −12.5, −37.5, −62.5, −87.5, −112.5, −137.5, −162.5, −187.5, −212.5, −237.5, −262.5, −287.5, −313.5, −344.6, −387.8, −450.7, −540.3, −662.8, −823.6, −1026.6, −1274.6, −1569.0, −1909.9, −2295.8, −2724.0, −3190.3, −3689.6, −4215.6, −4761.3, and −5319.1 m, respectively. The mixing scheme used in the model represents enhanced mixing in the surface layer (Pacanowski and Philander 1981). The model has been spun up for over 900 years from the initial motionless ocean with the temperature of Levitus and Boyer (1994) and salinity of Levitus et al. (1994). A hindcast run is forced with the wind stress and heat flux of the European Centre for Medium-Range Weather Forecasts (ECMWF) ERA-40 reanalysis from 1990 to 2001. The surface salinity is relaxed to the climatological field (Levitus et al. 1998). The hindcast used here is the same as that in YL09. The simulated average vertical profiles of density and buoyancy frequency square over the equatorial Indian Ocean between 5°S and 5°N agree with the World Ocean Atlas 2001 data well, as shown in YL09.

The wave decomposition method used here follows Yuan et al. (2004). The method makes use of the orthogonal relations of the long equatorial Kelvin and Rossby mode functions and projects the simulated fields onto the mode functions of these waves. The projection treats the nonlinear terms of the OGCM momentum equations as forcing terms and thus includes the nonlinear effects of the OGCM simulation in the extracted coefficients. The three-dimensional dynamic height and zonal velocity in reference to 2500-m depth are first projected onto the eigenfunctions to extract the coefficients associated with each baroclinic mode. These coefficients are then decomposed into equatorial Kelvin and Rossby waves. Details of the extraction procedure have been described in Yuan et al. (2004).

A simple linear characteristic wave model is used to evaluate the role of the eastern and western boundary reflections, which is the same as that used in Yuan and Han (2006). The zonal wind stress is projected onto the equatorial Kelvin and Rossby wave modes and is integrated along the characteristic lines of the equatorial waves to obtain the wave solutions. The linear characteristic wave model has the eigenvalues and eigenfunctions calculated from the zonally averaged vertical profiles of density over the equatorial Indian Ocean within 5°S and 5°N in LICOM. In addition, the wind stress curl along 12°S is also integrated along the Rossby wave characteristic line to obtain the sea level anomalies (SLA) forced by the propagation of wind-forced Rossby waves in the southern Indian Ocean. The parameters of the Rossby wave characteristic model follow those used in Kessler (2006).

The dipole mode index (DMI) time series are calculated using the monthly NCEP reanalysis sea surface temperature from 1948 to 2001. Years of pIOD and nIOD events during the above period are shown in Table 1. Three nIOD events (1992, 1996–97, and 1998–99) are selected for case studies in this paper. The altimeter data used in this study are the French Archiving, Validation, and Interpretation of Satellite Oceanographic Data (AVISO; Ducet et al. 2000), which are gridded at the resolution of ⅓° longitude × ⅓° latitude and over the global ocean from 82°S to 82°N. The interannual SLA are relative to the climatological seasonal average, which are calculated using the monthly data from 1993 to 2001.

Table 1.

Positive and negative IOD events in the twentieth century.

Positive and negative IOD events in the twentieth century.
Positive and negative IOD events in the twentieth century.

The analyses of this paper are based on monthly averages of the LICOM 1.0 hindcast. The interannual anomalies of the hindcast are relative to the climatological monthly averages from 1993 to 2001.

3. Results

In this section, the model results are validated by the altimeter SLA, based on which of the equatorial waves are extracted. The wave reflections at the western and eastern boundaries are then analyzed based on these extracted waves.

a. Model validation

The simulated sea level anomalies of the hindcast run are compared with the satellite altimeter data of the AVISO project. Figure 1 shows the evolution of the monthly anomalies of the altimeter SLA, model SLA, and zonal wind stress anomalies averaged between 5°S and 5°N. The upwelling anomalies of the pIOD events during 1994 and 1997–98 and the downwelling anomalies of the nIOD events during 1996–97 and 1998–99 in the eastern Indian Ocean are simulated well by the model. The simulated SLA are weaker than the observed in the 1990s, which may be because of the inaccuracies of the drag coefficient in the wind stress calculation. Despite the inaccuracies, the phases of the SLA associated with the IOD events are reproduced well, suggesting that the model-simulated results can be used to study the equatorial wave dynamics of the SLA evolution.

Fig. 1.

Hovmöller plots of monthly SLA (a) from AVISO altimeter data and (b) simulated by the OGCM, and (c) zonal wind stress anomalies, all averaged between 5°S and 5°N along the equator. The anomalies are based on the monthly climatology of 1993–2001. Contour interval is 5 cm for SLA and 0.02 Pa for wind stress anomalies. Shading indicates negative values.

Fig. 1.

Hovmöller plots of monthly SLA (a) from AVISO altimeter data and (b) simulated by the OGCM, and (c) zonal wind stress anomalies, all averaged between 5°S and 5°N along the equator. The anomalies are based on the monthly climatology of 1993–2001. Contour interval is 5 cm for SLA and 0.02 Pa for wind stress anomalies. Shading indicates negative values.

During the 1996–97 and 1998–99 nIOD events, downwelling anomalies are generated in the eastern equatorial Indian Ocean, which are evidenced by the simulated and observed positive SLA in the eastern basin (Figs. 1a,b). Similar phenomena are found in 1992 in the model although no altimeter data are available. Negative SLA are generated in the central basin at the mature phase of the nIOD events, which propagate westward and reach the western boundary at the beginning of the next year. The propagation of the SLA suggests that the equatorial waves play an important role in the evolution of the nIOD events. Similar propagation of SLA is also evident during the pIOD years, the dynamics of which have been investigated by YL09.

The zonal wind stress anomalies over the equatorial Indian Ocean suggest that positive wind stress anomalies force downwelling anomalies in the eastern equatorial Indian Ocean whereas negative wind stress anomalies force upwelling anomalies (Fig. 1c). The timing of the wind stress anomalies is not always in precise agreement with the SLA, suggesting wave propagation and reflection during the response of the equatorial ocean to the anomalous wind forcing, the dynamics of which will be disclosed in the text to follow.

b. Extracted equatorial waves

Coefficients of long equatorial Rossby and Kelvin waves are extracted from the model simulation to study the dynamics of the SLA evolution during the nIOD events. Figure 2 shows the coefficients of the decomposed Kelvin and the first meridional-mode Rossby waves of the first baroclinic mode during 1990–2001 using the wave decomposition method of Yuan et al. (2004).

Fig. 2.

Hovmöller plots of (a) decomposed equatorial Kelvin wave and (b) the first meridional-mode Rossby wave coefficients of the first baroclinic mode monthly anomalies from the model output during 1993–2001. The contour units correspond to 0.33 m for Kelvin wave sea level and 0.14 m for the first meridional-mode Rossby wave sea level on the equator. Contour interval is 2 units. Shading indicates negative values.

Fig. 2.

Hovmöller plots of (a) decomposed equatorial Kelvin wave and (b) the first meridional-mode Rossby wave coefficients of the first baroclinic mode monthly anomalies from the model output during 1993–2001. The contour units correspond to 0.33 m for Kelvin wave sea level and 0.14 m for the first meridional-mode Rossby wave sea level on the equator. Contour interval is 2 units. Shading indicates negative values.

The structure of the equatorial wave coefficients is reminiscent of the SLA in Fig. 1, which suggests eastward propagation of downwelling Kelvin waves from the western boundary in 1992 and 1998. In 1996, the downwelling Kelvin waves are generated in the central equatorial Indian Ocean at the end of the year, which propagate to the eastern boundary. Upwelling Kelvin waves are generated at the western boundary at the end of 1998, which follow the downwelling Kelvin waves to the eastern boundary of the Indian Ocean. The coincidence of the arrival of these upwelling Kelvin waves with the phase reversal of the SLA at the eastern boundary suggests that the downwelling anomalies in the eastern Indian Ocean during the 1998–99 nIOD are terminated by these Kelvin waves from the western boundary. However, no such upwelling Kelvin waves are found at the western boundary in early 1997. Instead, strong upwelling Kelvin waves are generated in the central-eastern equatorial Indian Ocean in 1997, which suggests different processes of negative feedback during the 1996–97 nIOD from those during the1998–99 nIOD.

The downwelling Kelvin waves of the nIOD events are reflected into westward-propagating downwelling equatorial Rossby waves at the eastern boundary, as shown in Fig. 2b, some of which reach the western boundary (1996–97) while others do not (1998–99). This difference suggests that the eastern boundary reflections may play different roles in the evolution of the 1996–97 nIOD and the 1998–99 nIOD events.

The reflection of the Rossby waves at the western boundary is more complicated than that at the eastern boundary. As elaborated by YL09, the downwelling Kelvin waves in pIOD events are composed of two pulses: one reflected from equatorial Rossby waves and one from the off-equatorial Rossby waves. In comparison, the generation of the upwelling Kelvin waves at the end of 1996 and 1998 is only involved in the reflection of equatorial Rossby waves. In the following sections, we will focus on the different negative feedback processes during these nIOD events.

c. Reflections at the western and eastern boundaries

Figure 3a shows the time series of the decomposed and reflected wave coefficients at the western boundary. The linearly reflected Kelvin wave is plotted in the dotted curve, which is calculated from the incoming equatorial Rossby waves of meridional modes 1 and 3 at the western boundary. The calculation of the reflected Kelvin wave coefficients from the incoming equatorial Rossby waves has been described in detail by Yuan et al. (2004), based on the theory of Cane and Gent (1984). The ratios of reflection at the Indian Ocean western boundary are 0.41 and 0.13 for the interannual Rossby waves of meridional modes 1 and 3, respectively. Higher meridional-mode Rossby waves make negligible contributions to the Kelvin wave amplitude. The contribution of the first meridional-mode Rossby wave to the reflection is also shown in the dashed curve in Fig. 3. The good agreement between the dotted and dashed curves suggests that the linearly reflected Kelvin wave comes primarily from the first meridional-mode Rossby wave. The difference between the decomposed and the linearly reflected Kelvin waves (thick solid curve in Fig. 3) is somewhat sizable, suggesting the nonlinear effects of the reflection and the off-equatorial reflections at the western boundary. However, during the period of 1996–97 and 1998–99 nIOD events, the difference from the linear reflection is small, suggesting that the weak nIOD anomalies, in comparison to those strong anomalies of the pIOD events, are essentially governed by the linear dynamics.

Fig. 3.

(a) Decomposed (thin solid) and linear reflected Kelvin wave (dotted) coefficients of the first baroclinic mode anomalies at the western boundary. The difference between the two is plotted with a thick solid curve. The time-shifted first meridional-mode Rossby wave coefficients have been multiplied by its reflection ratio (dashed). (b) Decomposed Kelvin wave coefficients (solid) and the first meridional-mode Rossby wave coefficients divided by the Kelvin wave reflection ratio (dotted) at the eastern boundary, the difference of which is plotted with a thick solid curve.

Fig. 3.

(a) Decomposed (thin solid) and linear reflected Kelvin wave (dotted) coefficients of the first baroclinic mode anomalies at the western boundary. The difference between the two is plotted with a thick solid curve. The time-shifted first meridional-mode Rossby wave coefficients have been multiplied by its reflection ratio (dashed). (b) Decomposed Kelvin wave coefficients (solid) and the first meridional-mode Rossby wave coefficients divided by the Kelvin wave reflection ratio (dotted) at the eastern boundary, the difference of which is plotted with a thick solid curve.

Since the western boundary of the Indian Ocean is not a meridional wall, it is necessary to take the reflections of the antisymmetric equatorial Rossby waves into consideration. However, the reflection ratios for the antisymmetric Rossby waves decrease with frequencies of the incoming equatorial Rossby waves according to the linear theory of Cane and Gent (1984). These ratios will decrease to negligible amplitudes at the interannual time scales [see Yuan and Han (2006) for the discussions]. Therefore, the reflections of the antisymmetric equatorial Rossby waves are neglected in the following dynamics analysis of the equatorial waves.

The solid thin line in Fig. 3a represents the coefficients of the Kelvin waves at the western boundary decomposed from the OGCM simulation, which include reflections from both equatorial and off-equatorial Rossby waves and from model nonlinearity. The upwelling Kelvin waves at the ends of 1996 and 1998 are reflected from the equatorial Rossby waves at the western boundary, suggested by the fact that the solid, dashed, and dotted curves are nearly identical to each other during those periods (Fig. 3a). In comparison with the double downwelling Kelvin waves during pIOD events in YL09, no double upwelling Kelvin wave pulses are identified during either nIOD event, however. The upwelling Kelvin waves during early 1999 are much more persistent than those during late 1996, suggesting that the upwelling Kelvin waves in 1996 and 1999 may play different roles in terminating the downwelling anomalies in the eastern Indian Ocean during the nIOD events. In addition, no upwelling Kelvin waves are generated at the western boundary in early 1997, suggesting that the downwelling anomalies in the eastern Indian Ocean during early 1997 are not terminated by the western boundary reflection. In the years after 2000, the western Indian Ocean is dominated by strong upwelling Kelvin waves, which are not explained well by the linear reflections of equatorial Rossby waves. The difference is actually involved in the reflections of the off-equatorial waves forced by the wind stress associated with the La Niña states in the Pacific Ocean through the atmospheric bridge, as suggested by the simple model experiments in YL09.

At the eastern boundary, the downwelling Rossby waves are generally in agreement with the linear reflection of downwelling Kelvin waves (Fig. 3b). The downwelling equatorial Rossby waves during late 1996 and 1998 are evidently reflected from the downwelling Kelvin waves at the eastern boundary. No abnormal reflections like those at the western boundary are observed during the years after 2000.

4. SLA and equatorial wave dynamics of the nIOD events

The above analyses suggest the role of equatorial waves and the western and eastern boundary reflections in the evolution of downwelling anomalies during the nIOD events. In this section, we use the decomposed waves and the SLA maps to study the detailed dynamics of the nIOD events in 1992, 1996–97, and 1998–99.

Throughout this paper, we focus on the termination of the downwelling anomalies in the eastern Indian Ocean during the nIOD events. For the 1996–97 and 1998–99 nIOD events, the downwelling anomalies in the eastern Indian Ocean take place in summer, peak in fall, and disappear in the next spring and summer, respectively.

a. The 1998 nIOD event

Figure 4 shows the continuous evolution of sea level and wind stress anomalies from September 1998 to June 1999. The nIOD event in 1998–99 follows the pIOD event in 1997–98. By September 1998, downwelling Kelvin waves forced by the westerly winds anomalies arrive at the eastern boundary and kick off the evolution of the downwelling anomalies during the 1998–99 nIOD event. In the meantime, upwelling Rossby waves dominate the central equatorial and the off-equatorial southeastern Indian Ocean and propagate westward. These Rossby waves are the equatorial Rossby waves forced by the westerly wind anomalies and the off-equatorial Rossby waves forced by the cyclonic wind stress curl. The westward propagation of the off-equatorial upwelling Rossby waves is destructed by an anticyclonic gyre with positive SLA in the southwestern off-equatorial Indian Ocean during the nIOD event.

Fig. 4.

Sea level (contour, cm) and wind stress (vectors, Pa) anomalies during September 1998 to June 1999. Shading indicates negative values.

Fig. 4.

Sea level (contour, cm) and wind stress (vectors, Pa) anomalies during September 1998 to June 1999. Shading indicates negative values.

Because of the destruction by the anticyclonic gyre east of Madagascar, only the equatorial upwelling Rossby waves are able to arrive at the western boundary in late 1998 and are reflected into upwelling equatorial Kelvin wave. This upwelling Kelvin wave pulse does not terminate the downwelling anomalies in the eastern equatorial Indian Ocean immediately, although it weakens the downwelling anomalies substantially. Because of the strong and persistent upwelling equatorial Rossby waves from the central southern Indian Ocean, upwelling Kelvin waves are continuously produced, which propagate eastward to terminate the anomalous downwelling in June 1999.

It is evident that the termination of the 1998–99 nIOD is closely related to the reflections of the equatorial waves at the western boundary. Figure 5 shows the decomposed Kelvin and Rossby wave coefficients during the pIOD and nIOD events of 1996–99. The theoretical first baroclinic Kelvin wave coefficient is plotted at the bottom of Fig. 5a in a solid line for comparison. Double downwelling Kelvin waves from the western boundary can be identified during early 1998, but no double upwelling Kelvin waves are identified during early 1999 (Fig. 5). The agreement of the western boundary reflection with the linear theory during early 1999 suggests that only the reflections of the equatorial Rossby waves are involved in the termination of the downwelling anomalies in the eastern basin during the 1998–99 nIOD event (Fig. 3).

Fig. 5.

Hovmöller plots of (a) decomposed Kelvin wave and (b) the first meridional-mode Rossby wave coefficients of the first baroclinic mode anomalies of the model simulation during 1996–99. The solid line at the bottom of (a) indicates the theoretical first baroclinic Kelvin wave coefficient.

Fig. 5.

Hovmöller plots of (a) decomposed Kelvin wave and (b) the first meridional-mode Rossby wave coefficients of the first baroclinic mode anomalies of the model simulation during 1996–99. The solid line at the bottom of (a) indicates the theoretical first baroclinic Kelvin wave coefficient.

The zonal wind stress on the equator during late 1998 through the middle of 1999 is dominated by westerly winds anomalies, which cannot provide the negative feedback to terminate the downwelling anomalies during the 1998–99 nIOD event (Fig. 1c). It is the reflection of the upwelling equatorial Rossby waves at the western boundary that provides the negative feedback to the downwelling anomalies in the eastern Indian Ocean during the 1998–99 nIOD event. The schematic of the negative feedback is shown in Fig. 12.

Downwelling Rossby waves reflected at the eastern boundary in 1998 and early 1999 do not make their way to the western boundary. They are overwhelmed by the upwelling Rossby waves forced by westerly wind anomalies in the central equatorial Indian Ocean (Fig. 5b).

b. The 1996 nIOD event

The equatorial wave dynamics of the SLA during the 1996–97 nIOD are similar to those in the 1998–99 nIOD, except that the downwelling equatorial Rossby waves reflected from the eastern boundary have reached the western boundary in February 1997. Figure 6 shows the evolution of Indian Ocean SLA and wind stress anomalies in 1996–97. The equatorial eastern Indian Ocean is dominated by positive SLA since the summer of 1996, with negative SLA in the western and central equatorial Indian Ocean. Upwelling equatorial Rossby waves are seen to propagate westward, as suggested in Fig. 5, and to dominate the SLA near the western boundary in July–November 1996, resulting in a peak east–west SLA gradient at the end of 1996. The upwelling Rossby waves are reflected into upwelling Kelvin waves, most of which do not make their way to the eastern boundary until early 1997 (cf. Fig. 5). The reflected Kelvin waves are evidently not strong enough to terminate the downwelling anomalies in the eastern basin, because the eastern equatorial Indian Ocean is still dominated by positive SLA at this time.

Fig. 6.

Sea level (contour, cm) and wind stress (vectors, Pa) anomalies during September 1996 to April 1997. Shading indicates negative values.

Fig. 6.

Sea level (contour, cm) and wind stress (vectors, Pa) anomalies during September 1996 to April 1997. Shading indicates negative values.

Downwelling equatorial Rossby waves reflected from downwelling Kelvin waves at the eastern boundary follow the upwelling equatorial Rossby waves to the western boundary in late 1996 through early 1997. This is evidenced by a belt of positive SLA along roughly 5°S propagating to the west and eventually overwhelming the negative SLA in the western equatorial basin in February 1997 (Fig. 6). Figure 5 shows that the Kelvin waves change from upwelling to downwelling anomalies at the beginning of 1997, which are associated with the reflections of the equatorial Rossby waves. These downwelling Rossby waves are shown to be reflected from the downwelling Kelvin waves at the eastern boundary (Fig. 3b). The above analyses suggest that the eastern boundary reflection plays an important role in the evolution of the SLA during the 1996–97 nIOD event.

The downwelling anomalies in the eastern equatorial Indian Ocean are not terminated until March 1997, when easterly wind anomalies dominate over the equatorial basin. The evolution of the anomalies seems to suggest that the downwelling anomalies in the eastern Indian Ocean during the 1996–97 nIOD event will be terminated by the atmosphere–ocean coupled process. It is likely that the interactions of the atmospheric Hadley cell and the Walker cell generate the easterly wind anomalies in the spring 1997. Although the off-equatorial central southern Indian Ocean is dominated by negative SLA in late 1996 through early 1997, these anomalies evidently do not play a role in terminating the downwelling anomalies over the equatorial eastern Indian Ocean because of the absence of the western boundary reflections of these anomalies in early 1997 before the onset of the easterly wind anomalies. Therefore, the negative feedback and the equatorial wave dynamics of the SLA in the 1996–97 nIOD event are evidently very different from those in the 1998–99 nIOD event. The former is terminated by the atmospheric easterly wind anomalies, whereas the latter is terminated by the western boundary reflection.

c. The 1992 nIOD event

Similar to the 1998–99 nIOD event, the 1992 nIOD event is preceded by a weak and irregular pIOD event in 1991 based on the DMI. The eastern equatorial Indian Ocean is dominated by anomalous downwelling currents since April 1992 (Fig. 1b), which are evidently associated with the forcing of the westerly wind anomalies (Fig. 1c). Prior to the westerly wind bursts, downwelling SLA dominate the western equatorial basin, which propagate westward and are reflected into downwelling Kelvin waves since the beginning of 1992 (Fig. 7). These downwelling waves evidently weaken the negative SLA in the eastern basin forced by the easterly wind anomalies in early 1992 (Fig. 1). Upwelling Rossby waves and downwelling Kelvin waves are generated by the westerly wind bursts since April 1992, which propagate to the west and east, respectively. The reflection of the upwelling Rossby waves in late 1992 results in an upwelling Kelvin wave to reach the eastern boundary in December 1992 through January 1993. The coincidence of the arrival of this Kelvin wave with the termination of the downwelling anomalies near the eastern boundary suggests that the latter is terminated by the former.

Fig. 7.

Hovmöller plots of (a) decomposed Kelvin wave and (b) the first meridional-mode Rossby wave coefficients of the first baroclinic mode anomalies of the model simulation during 1991–93. Shading indicates negative values.

Fig. 7.

Hovmöller plots of (a) decomposed Kelvin wave and (b) the first meridional-mode Rossby wave coefficients of the first baroclinic mode anomalies of the model simulation during 1991–93. Shading indicates negative values.

Downwelling Rossby waves reflected at the eastern boundary follow the upwelling Rossby waves to the western boundary and are reflected into downwelling Kelvin waves to arrive at the eastern boundary in February–March 1993 (Fig. 7). The upwelling and downwelling waves are seen to be bounced back and forth between the western and eastern boundaries and to generate repercussions in the equatorial Indian Ocean in many months to follow.

5. Equatorial waves forced by the winds and boundary reflections

To quantitatively assess the relative roles of the wind-forced and reflected waves in the termination of the downwelling anomalies in the eastern Indian Ocean during the nIOD events, the linear characteristic wave model is used to evaluate these waves explicitly.

a. Western boundary reflection and easterly wind forcing

The linear characteristic wave model is integrated twice, from the decomposed Kelvin wave coefficients at the western boundary with no wind forcing and from zero value at the western boundary forced with the wind stress projected onto the Kelvin mode, respectively. Figure 8 shows their comparison.

Fig. 8.

Hovmöller plots. (a) Wind-forced Kelvin wave coefficients with western boundary wave condition at 49°E set to zero. (b) Kelvin wave coefficients forced by the western boundary condition.

Fig. 8.

Hovmöller plots. (a) Wind-forced Kelvin wave coefficients with western boundary wave condition at 49°E set to zero. (b) Kelvin wave coefficients forced by the western boundary condition.

During the 1996–97 nIOD event, upwelling Kelvin waves forced by the western boundary reflection persist until December 1996, after which significant downwelling Kelvin waves generated by the western boundary reflection propagate to the east (Fig. 8b). These downwelling Kelvin waves evidently are not responsible for the termination of the downwelling anomalies in the east. In comparison, significant wind-forced upwelling Kelvin waves are generated in the central to eastern equatorial Indian Ocean, which arrive at the eastern boundary in February 1997 (Fig. 8a). Evidently, it is the easterly wind forcing, not the western boundary reflection, that terminates the downwelling anomalies in the eastern basin.

During the 1998–99 nIOD event, upwelling Kelvin waves at the western boundary are generated persistently since the end of 1998 (Fig. 8b). The wind-forced Kelvin waves are downwelling anomalies since mid-1998 until May 1999 in the eastern basin. Therefore, the wind-forced Kelvin waves are not responsible for the termination of the downwelling anomalies in the east before May 1999. Although the winds start to force upwelling Kelvin wave anomalies in the late spring of 1999, the contribution of the Kelvin waves forced by the western boundary over time is evidently larger than that of the wind-forced upwelling waves, suggesting the importance of the western boundary reflection in the termination of the downwelling anomalies in the eastern Indian Ocean during the 1998–99 nIOD event.

b. The role of the eastern boundary reflections

To study the role of the eastern boundary reflections, the same linear characteristic–line wave model is used to separate the wind-forced and boundary-forced waves. Two experiments are conducted with the model. In the first experiment, the wind stress is integrated westward along the Rossby wave characteristic lines from zero at the eastern boundary (Fig. 9a). The calculated Rossby wave coefficients represent the wind-forced equatorial Rossby waves over the equatorial Indian Ocean. In the second experiment, the integration is repeated with the condition at the eastern boundary set to be the decomposed Rossby wave coefficients at 96°E (Fig. 9b). The calculated coefficients represent the Rossby waves forced by both the winds and the eastern boundary conditions. The Rossby waves forced by the eastern boundary conditions can be obtained by integrating the wave model with the wind stress set at zero (Fig. 9c). The first meridional-mode Rossby waves are reflected at the western boundary into equatorial Kelvin waves with the reflection ratio of 0.41 in amplitude. Then the Kelvin wave coefficients are obtained by integrating the wind stress along the Kelvin wave characteristic lines eastward (Figs. 9d–f).

Fig. 9.

Hovmöller plots. (a) The wind-forced Rossby wave coefficients with eastern boundary wave condition at 96°E set to zero. (b) The wind forced Rossby wave coefficients forced with wind and the eastern boundary condition. (c) The difference between (b) and (a). (d)–(f) The Kelvin wave coefficients forced with the reflected Rossby wave at 49°E in (a)–(c), respectively.

Fig. 9.

Hovmöller plots. (a) The wind-forced Rossby wave coefficients with eastern boundary wave condition at 96°E set to zero. (b) The wind forced Rossby wave coefficients forced with wind and the eastern boundary condition. (c) The difference between (b) and (a). (d)–(f) The Kelvin wave coefficients forced with the reflected Rossby wave at 49°E in (a)–(c), respectively.

The Rossby and Kelvin wave coefficients forced by the winds and the eastern boundary conditions reproduce the decomposed Rossby and Kelvin wave coefficients well (cf. Figs. 9b,e and Fig. 5). The wind-forced Rossby waves are evidently reflected into upwelling Kelvin waves at the western boundary during late 1996 through early 1997, which would have provided the negative feedback to the downwelling anomalies in the east (Figs. 9a,d). However, the eastern boundary reflection forces downwelling Kelvin waves over the entire equatorial Indian Ocean from the fall of 1996 through the spring of 1997 (Figs. 9c,f). The total Kelvin waves forced by both the winds and the eastern boundary reflection are downwelling anomalies at the western boundary since December 1996 (Figs. 9b,e), which are unable to provide the negative feedback to the downwelling anomalies in the east. Therefore, the downwelling anomalies in the eastern Indian Ocean during the 1996–97 nIOD are not terminated by the western boundary reflection. Instead, the wind-forced upwelling Kelvin waves are strong since the spring of 1997, which provide the negative feedback to terminate the downwelling anomalies during the 1996–97 nIOD. In comparison, the wind-forced upwelling equatorial Rossby waves are strong and persistent during late 1998 through early 1999, which dwarf the downwelling Kelvin waves forced by the eastern boundary condition during the same period. Thus, the western boundary reflection is able to provide the negative feedback to the downwelling anomalies in the east during the 1998–99 nIOD event.

In summary, the role of the eastern boundary reflection is important for the evolution of downwelling anomalies during the nIOD events and should not be overlooked in the simulation and prediction of the interannual variations of the Indian Ocean climate.

6. Discussion

To understand why the off-equatorial Rossby waves cannot reach the western boundary during the 1998–99 nIOD event, the linear Rossby characteristic wave model is used to investigate the dynamics of the anticyclonic gyre east of Madagascar. Figure 10 shows the Hovmöller plot of the SLA and the wind stress curl anomalies along 12°S. The eastern boundary condition is set at zero during the integration. The wind stress curl anomalies along 12°S in the south Indian Ocean are positive from middle 1997 through early 1998, which force downwelling Rossby waves to propagate and reach the east coast of Madagascar during late 1997 through 1999. Thus, the positive SLA east of Madagascar are opposite in sign to the negative wind stress curl anomalies along this latitude during 1998–99 nIOD. The results of the Rossby characteristic wave model show clearly that the gyre of the positive SLA is forced by the remote wind stress curl anomalies from 1997 through 1998 in the south Indian Ocean through the propagation of the Rossby waves. This area of positive SLA prevents the off-equatorial upwelling Rossby waves forced by the negative wind stress curl anomalies during the 1998–99 nIOD from reaching the western boundary.

Fig. 10.

Hovmöller plots of wind stress curl anomalies (10−7 N m−3) and SLA (cm) along 12°S from the wind-forced baroclinic Rossby characteristic wave model.

Fig. 10.

Hovmöller plots of wind stress curl anomalies (10−7 N m−3) and SLA (cm) along 12°S from the wind-forced baroclinic Rossby characteristic wave model.

During the 1996–97 nIOD, the negative sea level anomalies in the western equatorial Indian Ocean are terminated by the arrival of the downwelling equatorial Rossby waves from the eastern boundary reflection, resulting in no negative feedback from the western boundary since January 1997. Analyses in section 5a have demonstrated that the Kelvin waves forced by the easterly wind anomalies in the central to eastern equatorial Indian Ocean since February 1997 terminate the downwelling anomalies in the eastern Indian Ocean (Fig. 8a). The Walker circulation anomalies over the equatorial Indian Ocean begin to reverse since as early as January 1997 (Fig. 6), suggesting the proactive role of the easterly wind anomalies in the termination of the downwelling anomalies in the eastern Indian Ocean during the 1996–97 nIOD. We speculate that air–sea coupled processes are involved in generating the easterly wind anomalies during the 1996–97 nIOD. Further analyses are needed to disclose the coupled negative feedback process of this nIOD event.

The off-equatorial Rossby waves during the 1998–99 nIOD are generated mainly in the southern tropical Indian Ocean where the thermocline is shallow (Xie et al. 2002). The strong thermocline–SSTA interactions in the region are suggested by Sayantani and Gnanaseelan (2014) to play an important role in producing the persistent subsurface anomalies.

7. Conclusions

Long equatorial wave dynamics of sea level variations during nIOD events in the 1990s are studied using a numerical simulation of the LICOM model forced by the wind stress and heat flux of the ERA-40 reanalysis from 1990 to 2001. The simulated SLA over the equatorial Indian Ocean compare well with the AVISO altimeter data, suggesting that the model can be used to investigate the dynamics of the SLA. The model results are decomposed into equatorial waves to study the role of the propagation and reflections of the equatorial waves in the SLA evolution during the nIOD events. Significant asymmetry of equatorial wave dynamic is identified of the nIOD events from those of the pIOD events in YL09.

Figure 11 shows the schematic of the negative feedback process of pIOD events. Dashed and solid contours stand for upwelling and downwelling anomalies, respectively; block arrows stand for wind stress anomalies; and lines and curved arrows stand for propagation and reflections at the eastern boundary. The circled numbers 1–4 stand for the sequence of the negative feedback:

  1. Easterly wind anomalies force upwelling equatorial Kelvin waves (dashed bell shape) and downwelling Rossby waves (solid horseshoe shape) propagating eastward and westward, respectively.

  2. The equatorial Rossby waves are reflected into downwelling Kelvin waves at the western boundary, forming the first pulse of the negative feedback. Meanwhile, off-equatorial Rossby waves propagate westward at a slower speed.

  3. The off-equatorial Rossby waves reach the western boundary and are reflected into the equatorial downwelling Kelvin wave, forming the second pulse of the negative feedback.

  4. The second Kelvin wave pulse terminates the upwelling anomalies in the east (see the detailed process in YL09).

Fig. 11.

Schematic of negative feedback during pIOD events. The circled numbers 1–4 stand for the sequence of the negative feedback. EQ stands for equator; WB and EB stand for western and eastern boundary, respectively; block arrows represent wind stress anomalies; and lines and curved arrows represent propagation and reflection at the eastern boundary. Dashed contours (circled minus) stand for upwelling anomalies and solid contours (circled plus) downwelling anomalies. The bell shape stands for a Kelvin wave and the horseshoe shape stands for a Rossby wave. The Gaussian curve near the EB stands for a Kelvin wave form.

Fig. 11.

Schematic of negative feedback during pIOD events. The circled numbers 1–4 stand for the sequence of the negative feedback. EQ stands for equator; WB and EB stand for western and eastern boundary, respectively; block arrows represent wind stress anomalies; and lines and curved arrows represent propagation and reflection at the eastern boundary. Dashed contours (circled minus) stand for upwelling anomalies and solid contours (circled plus) downwelling anomalies. The bell shape stands for a Kelvin wave and the horseshoe shape stands for a Rossby wave. The Gaussian curve near the EB stands for a Kelvin wave form.

In comparison with the pIOD events, during which the upwelling anomalies in the eastern Indian Ocean are terminated by double Kelvin waves reflected at the western boundary from equatorial and off-equatorial Rossby waves, the downwelling anomalies in the eastern Indian Ocean during the nIOD events are terminated by two kinds of processes: western boundary reflection and easterly wind anomalies. The 1998–99 nIOD event is terminated by the former and the 1996–97 nIOD event is terminated by the latter, as illustrated by the analyses of this paper.

Figure 12 shows the schematic of the negative feedback process during the nIOD events. Dashed and solid contours stand for the same anomalies as Fig. 11a. The numbers 1 and 2 are of opposite sign to pIOD negative feedback 1 and 2. Two kinds of negative feedback ensue: western boundary reflection (left; the 1998–99 nIOD event) and easterly wind burst (right; the 1996–97 nIOD event). During the 1998–99 nIOD event, the negative feedback is essentially the same as pIOD 3 and 4 with opposite signs (left 3 and 4 of Fig. 12), except that no reflections of the off-equatorial Rossby waves are involved in the generation of the Kelvin waves, because of the destruction of the off-equatorial Rossby wave propagation by the positive SLA east of the island of Madagascar. The upwelling Kelvin waves from the western boundary are reflected from upwelling equatorial Rossby waves forced by the wind anomalies in the central and southern equatorial Indian Ocean.

Fig. 12.

As in Fig. 11, but for nIOD events. The curved block arrows in the middle stand for two kinds of negative feedback.

Fig. 12.

As in Fig. 11, but for nIOD events. The curved block arrows in the middle stand for two kinds of negative feedback.

During the nIOD event in 1996–97, the upwelling Kelvin waves reflected from upwelling equatorial Rossby waves at the western boundary weaken the downwelling anomalies in the eastern basin in the beginning of the negative feedback process from late 1996 to early 1997. Different from the nIOD event in 1998–99, the downwelling Rossby waves reflected at the eastern boundary reach the west boundary and reverse the anomalies of the reflection. No negative feedback is coming from the western boundary after February 1997 since the SLA in the western basin are overwhelmed by the downwelling Rossby waves (Fig. 12, right 3). In the meantime, easterly wind anomalies start in the western basin. At last, the upwelling Kelvin waves forced by the easterly wind anomalies terminate the downwelling anomalies in the east (right 4 of Fig. 12).

The analyses of this study suggest that the equatorial wave reflections at the western and eastern boundaries play an important role in the termination of the downwelling anomalies in the eastern Indian Ocean during the nIOD events. In addition, the role of the ocean–atmosphere coupling should also be taken into consideration to understand the negative feedback processes of the nIOD events. In comparison with the double-Kelvin-wave negative feedback to the upwelling anomalies in the eastern Indian Ocean during the pIOD events, the second negative feedback of the nIOD events does not always come from the western boundary reflection. It is found that the downwelling anomalies in the eastern Indian Ocean during the 1996–97 nIOD event are terminated when the thermocline over the entire equatorial Indian Ocean is anomalously deep. Analyses of the equatorial waves forced by the winds and boundary reflections suggest that the downwelling anomalies during this nIOD event are terminated by the easterly wind anomalies in the spring of 1997. It is hypothesized that the persistent downwelling state in the equatorial Indian Ocean induces interactions of the Hadley cell and the Walker cell of the atmospheric circulation, which generate easterly wind anomalies to terminate the downwelling anomalies in the east during the 1996–97 nIOD event. Further studies are needed to understand the ocean–atmosphere coupling, which will bear considerable significance to the study and prediction of the tropical climate variations at the interannual time scales in the Indian Ocean.

Acknowledgments

J. Wang is supported by NSFC (Grant 41206018) and CAS (Grant XDA11010203). D. Yuan is supported by the National Basic Research Program of China project (Grant 2012CB956001), CMA (Grant GYHY201306018), CAS (Grant XDA11010301), NSFC (Grants 41176019, 41421005, U1406401), and SOA (Grant GASI-03-01-01-05).

REFERENCES

REFERENCES
Annamalai
,
H.
, and
R.
Murtugudde
,
2004
: Role of the Indian Ocean in regional climate variability. Earth’s Climate: The Ocean-Atmosphere Interaction, Geophys. Monogr., Vol. 147, Amer. Geophys. Union, 213–246.
Behera
,
S. K.
,
J. J.
Luo
,
S.
Masson
,
P.
Delecluse
,
S.
Gualdi
,
A.
Navarra
, and
T.
Yamagata
,
2005
:
Paramount impact of the Indian Ocean dipole on the east African short rains: A CGCM study
.
J. Climate
,
18
,
4514
4530
, doi:.
Cai
,
W.
,
P.
Rensch
,
T.
Cowan
, and
H. H.
Hendon
,
2012
:
An asymmetry in the IOD and ENSO teleconnection pathway and its impact on Australian climate
.
J. Climate
,
25
,
6318
6329
, doi:.
Cane
,
M. A.
, and
P. R.
Gent
,
1984
:
Reflection of low-frequency equatorial waves at arbitrary western boundaries
.
J. Mar. Res.
,
42
,
487
502
, doi:.
Chambers
,
D. P.
,
B. D.
Tapley
, and
R. H.
Stewart
,
1999
:
Anomalous warming in the Indian Ocean coincident with El Niño
.
J. Geophys. Res.
,
104
,
3035
3047
, doi:.
Ducet
,
N.
,
P. Y.
Le Traon
, and
G.
Reverdin
,
2000
:
Global high-resolution mapping of ocean circulation from TOPEX/Poseidon and ERS-1 and -2
.
J. Geophys. Res.
,
105
,
19 477
19 498
, doi:.
Feng
,
M.
, and
G.
Meyers
,
2003
:
Interannual variability in the tropical Indian Ocean: A two-year time-scale of Indian Ocean dipole
.
Deep-Sea Res. II
,
50
,
2263
2284
, doi:.
Han
,
W.
,
T.
Shinoda
,
L.
Fu
, and
J. P.
McCreary
,
2006
:
Impact of atmospheric intraseasonal oscillations on the Indian Ocean dipole during the 1990s
.
J. Phys. Oceanogr.
,
36
,
670
690
, doi:.
Han
,
W.
,
D.
Yuan
,
W. T.
Liu
, and
D. J.
Halkides
,
2007
:
Intraseasonal variability of Indian Ocean sea surface temperature during boreal winter: Madden-Julian Oscillation versus submonthly forcing and processes
.
J. Geophys. Res.
,
112
, C04001, doi:.
Huang
,
B.
, and
J. L.
Kinter
III
,
2002
:
Interannual variability in the tropical Indian Ocean
.
J. Geophys. Res.
,
107
,
3199
, doi:.
Iizuka
,
S.
, and
T.
Matsuura
,
2012
: Cyclones: Formation, Triggers and Control. K. Oouchi and H. Fudeyasu, Eds., Nova Science Publishers, 37–60.
Jury
,
M. R.
, and
B.
Huang
,
2004
:
The Rossby wave as a key mechanism of Indian Ocean climate variability
.
Deep-Sea Res. I
,
51
,
2123
2136
, doi:.
Kessler
,
W. S.
,
2006
:
The circulation of the eastern tropical Pacific: A review
.
Prog. Oceanogr.
,
69
,
181
217
, doi:.
Le Blanc
,
J. L.
, and
J. P.
Boulanger
,
2001
:
Propagation and reflection of long equatorial waves in the Indian Ocean from TOPEX/POSEIDON data during the 1993–1998 period
.
Climate Dyn.
,
17
,
547
557
, doi:.
Levitus
,
S.
, and
T. P.
Boyer
,
1994
: Temperature. Vol. 4, World Ocean Atlas 1994, NOAA Atlas NESDIS 4, 117 pp.
Levitus
,
S.
,
R.
Burgett
, and
T. P.
Boyer
,
1994
: Salinity. Vol. 3, World Ocean Atlas 1994, NOAA Atlas NESDIS 3, 99 pp.
Levitus
,
S.
, and Coauthors
,
1998
: Introduction. Vol. 1, World Ocean Database 1998, NOAA Atlas NESDIS 18, 346 pp.
Li
,
C.
, and
M.
Mu
,
2001
:
The influence of the Indian Ocean dipole on atmospheric circulation and climate
.
Adv. Atmos. Sci.
,
18
(
5
),
831
843
.
Li
,
T.
,
B.
Wang
,
C.-P.
Chang
, and
Y.
Zhang
,
2003
:
A theory for the Indian Ocean dipole-zonal mode
.
J. Atmos. Sci.
,
60
,
2119
2135
, doi:.
Mason
,
S.
,
P. R.
Waylen
,
G. M.
Mimmack
,
B.
Rajaratnam
, and
J. M.
Harrison
,
1999
:
Changes in extreme rainfall events in South Africa
.
Climatic Change
,
41
,
249
257
, doi:.
Masumoto
,
Y.
, and
G.
Meyers
,
1998
:
Forced Rossby waves in the southern Indian Ocean
.
J. Geophys. Res.
,
103
,
27 589
27 602
, doi:.
Pacanowski
,
R. C.
, and
S. G. H.
Philander
,
1981
:
Parameterization of vertical mixing in numerical models of tropical oceans
.
J. Phys. Oceanogr.
,
11
,
1443
1451
, doi:.
Rao
,
S. A.
, and
T.
Yamagata
,
2004
:
Abrupt termination of Indian Ocean dipole events in response to intraseasonal disturbances
.
Geophys. Res. Lett.
,
31
,
L19306
, doi:.
Rao
,
S. A.
,
S. K.
Behera
,
Y.
Masumoto
, and
T.
Yamagata
,
2002
:
Interannual subsurface variability in the tropical Indian Ocean with a special emphasis on the Indian Ocean dipole
.
Deep-Sea Res. II
,
49
,
1549
1572
, doi:.
Rao
,
S. A.
,
S.
Masson
,
J. J.
Luo
,
S. K.
Behera
, and
T.
Yamagata
,
2007
:
Termination of Indian Ocean dipole events in a coupled general circulation model
.
J. Climate
,
20
,
3018
3035
, doi:.
Saji
,
N. H.
,
B. N.
Goswami
,
P. N.
Vinayachandran
, and
T.
Yamagata
,
1999
:
A dipole mode in the tropical Indian Ocean
.
Nature
,
401
(
6751
),
360
363
.
Saha
,
K.
, and
S.
Wasimi
,
2013
:
Interrelationship between Indian Ocean dipole (IOD) and Australian tropical cyclones
.
Int. J. Environ. Sci. Dev.
,
4
,
647
651
, doi:.
Sayantani
,
O.
, and
C.
Gnanaseelan
,
2014
:
Tropical Indian Ocean subsurface temperature variability and the forcing mechanisms
.
Climate Dyn.
,
44
,
2447
2462
, doi:.
Thompson
,
B.
,
C.
Gnanaseelan
,
A.
Parekh
, and
P. S.
Salvekar
,
2009
:
A model study on oceanic processes during the Indian Ocean Dipole termination
.
Meteor. Atmos. Phys.
,
105
,
17
27
, doi:.
Vinayachandran
,
P.
,
S.
Iizuka
, and
T.
Yamagata
,
2002
:
Indian Ocean dipole mode events in an ocean general circulation model
.
Deep-Sea Res. II
,
49
,
1573
1596
, doi:.
Vinayachandran
,
P.
,
P. A.
Francis
, and
S. A.
Rao
,
2009
: Indian Ocean dipole: Processes and impacts. Current Trends in Science, Indian Academy of Sciences, 569–589.
Webster
,
P. J.
,
A. M.
Moore
,
J. P.
Loschnigg
, and
R. R.
Leben
,
1999
:
Coupled ocean-atmosphere dynamics in the Indian Ocean during 1997–98
.
Nature
,
401
,
356
360
, doi:.
Weller
,
E.
, and
W.
Cai
,
2013
:
Asymmetry in the IOD and ENSO teleconnection in a CMIP5 model ensemble and its relevance to regional rainfall
.
J. Climate
,
26
,
5139
5149
, doi:.
Xie
,
S. P.
,
H.
Annamalai
,
F. A.
Schott
, and
J. P.
McCreary
,
2002
:
Structure and mechanisms of south Indian Ocean climate variability
.
J. Climate
,
15
,
864
878
, doi:.
Yuan
,
D.
, and
W.
Han
,
2006
:
Roles of equatorial waves and western boundary reflection in the seasonal circulation of the equatorial Indian Ocean
.
J. Phys. Oceanogr.
,
36
,
930
944
, doi:.
Yuan
,
D.
, and
H.
Liu
,
2009
:
Long-wave dynamics of sea level variations during Indian Ocean dipole events
.
J. Phys. Oceanogr.
,
39
,
1115
1132
, doi:.
Yuan
,
D.
,
M. M.
Rienecker
, and
P. S.
Schopf
,
2004
:
Long wave dynamics of the interannual variability in a numerical hindcast of the equatorial Pacific Ocean circulation during the 1990s
.
J. Geophys. Res.
,
109
,
C05019
, doi:.