Show Summary Details

Page of

Printed from Oxford Research Encyclopedias, Climate Science. Under the terms of the licence agreement, an individual user may print out a single article for personal use (for details see Privacy Policy and Legal Notice).

date: 27 June 2022

The Indian Ocean Dipolefree

The Indian Ocean Dipolefree

  • Saji N. HameedSaji N. HameedThe University of Aizu


Discovered at the very end of the 20th century, the Indian Ocean Dipole (IOD) is a mode of natural climate variability that arises out of coupled ocean–atmosphere interaction in the Indian Ocean. It is associated with some of the largest changes of ocean–atmosphere state over the equatorial Indian Ocean on interannual time scales. IOD variability is prominent during the boreal summer and fall seasons, with its maximum intensity developing at the end of the boreal-fall season. Between the peaks of its negative and positive phases, IOD manifests a markedly zonal see-saw in anomalous sea surface temperature (SST) and rainfall—leading, in its positive phase, to a pronounced cooling of the eastern equatorial Indian Ocean, and a moderate warming of the western and central equatorial Indian Ocean; this is accompanied by deficit rainfall over the eastern Indian Ocean and surplus rainfall over the western Indian Ocean. Changes in midtropospheric heating accompanying the rainfall anomalies drive wind anomalies that anomalously lift the thermocline in the equatorial eastern Indian Ocean and anomalously deepen them in the central Indian Ocean. The thermocline anomalies further modulate coastal and open-ocean upwelling, thereby influencing biological productivity and fish catches across the Indian Ocean. The hydrometeorological anomalies that accompany IOD exacerbate forest fires in Indonesia and Australia and bring floods and infectious diseases to equatorial East Africa. The coupled ocean–atmosphere instability that is responsible for generating and sustaining IOD develops on a mean state that is strongly modulated by the seasonal cycle of the Austral-Asian monsoon; this setting gives the IOD its unique character and dynamics, including a strong phase-lock to the seasonal cycle. While IOD operates independently of the El Niño and Southern Oscillation (ENSO), the proximity between the Indian and Pacific Oceans, and the existence of oceanic and atmospheric pathways, facilitate mutual interactions between these tropical climate modes.


  • Climate Systems and Climate Dynamics
  • History of Climate Science
  • Climate Impact: Extreme Events


The study of the ocean and the atmosphere as mutually interacting systems has provided deep insights into the workings of the climate system (Bjerknes, 1969; Gill, 1982; McCreary & Anderson, 1985; Yamagata, 1985). Some of the spectacular fluctuations of climate state from one year to another owe their origin to instabilities of the coupled ocean–atmosphere system; these instabilities are understood to be a fundamental driver, among other things, of the occasional failures of the Indian monsoon, extensive droughts in Indonesia and Australia, and floods over East Africa and Peru.

The tropics, in particular, provide a source region for the most significant climate perturbations generated by ocean–atmosphere interactions: here, conditions favor the maintenance of coupled instabilities over several seasons, leading to a distinct modulation of the seasonal cycle. In the tropics, these coupled instabilities manifest as distinct, recurrent patterns of sea surface temperature (SST), wind, and rainfall anomalies. Furthermore, they influence a significant portion of the globe, through planetary waves excited by their anomalous tropical heating patterns, raising their importance even more.

The first of the tropical coupled phenomena to be studied, the El Niño and Southern Oscillation (ENSO), arises out of ocean–atmosphere interactions in the tropical Pacific. The tropical Atlantic also exhibits an ENSO-like mode, sometimes referred to as the Atlantic Niño. Toward the end of the 20th century, a new mode of coupled ocean–atmosphere interaction was discovered over the tropical Indian Ocean; known as the Indian Ocean Dipole Mode (IOD), this important phenomenon is discussed in this article.

The discovery of IOD was an important milestone in the study of ocean–atmosphere interactions. Prior to the discovery of IOD, climate scientists thought that the tropical Indian Ocean did not generate an inherent mode of coupled variability; in fact, the conventional view maintained that the tropical Indian Ocean passively responded to, and amplified, ENSO-associated variations (Wallace et al., 1998). The concept that IOD arises as an inherent climate mode of the Indian Ocean took a while to get established as various debates vigorously examined aspects of IOD (Saji & Yamagata, 2003b; Yamagata et al., 2004), including its validity and independence from ENSO. Some of these important issues and their resolution are considered, among other things, in this article. However, the article first reviews the developments that led to the discovery of IOD and its exposition as an inherent coupled mode of the tropical Indian Ocean. Then the article discusses how many unique aspects of IOD are due to the peculiar mean state upon which the coupled instability driving IOD develops. Finally, some of the important impacts associated with the phenomenon are reviewed.

The Discovery of IOD

Inherent modes of climate variability resulting from coupled ocean–atmosphere interaction were discovered in the 1960s in the tropical Pacific Ocean (Bjerknes, 1969) and in the 1980s in the Atlantic Ocean (Hisard, 1980; Zebiak, 1993). In contrast, until nearly the end of the 20th century, it was widely believed that such a mode did not exist in the tropical Indian Ocean (Wallace et al., 1998). Discussions regarding year-to-year variations of SST in the Indian Ocean focused on a basinwide anomaly pattern that was remotely driven by ENSO (Lau & Nath, 1996; Weare, 1979). Furthermore, a strong relation between Indian summer monsoon rainfall and ENSO was recognized for a long time (Goswami et al., 1999; Pant & Parthasarathy, 1981).

Thus, in the beginning of 1999, the prevailing view on Indian Ocean climate variations was that the Indian Ocean was incapable of generating its own coupled mode, and that the year-to-year variations in SST and rainfall over the Indian sector were mostly forced by ENSO (Wallace et al., 1998). However, by the middle of 1999, this paradigm was strongly challenged by discoveries that revealed new, unexpected facets of climate variations within the Indian Ocean region.

In early 1999, researchers at the newly established Frontier Research Center for Global Change (FRCGC) in Japan documented an interesting pattern of climate variability that occurred during 1994 in the tropical Indian Ocean (Behera et al., 1999; Vinayachandran et al., 1999). This singular event, which took the climate community by surprise, was characterized by strong westward wind anomalies over the central equatorial Indian Ocean that persisted for a long time, from April to October; accompanying this was an unusual pattern of SST variation characterized by cool SST anomalies in the eastern tropical Indian Ocean (Vinayachandran et al., 1999) and warm SST anomalies in the central and western Indian Ocean. Further, newly available sea surface height (SSH) measurements, from satellite remote sensing, showed that these were accompanied by abnormally low SSH at the eastern Indian Ocean (Behera et al., 1999; Meyers, 1996; Potemra & Lukas, 1999), while hydrographic measurements revealed that the normally eastward-flowing Wyrtki jet had considerably weakened during the boreal spring of 1994 (Reppin et al., 1999). This singular Indian Ocean event, in 1994, set the stage for the discovery of IOD at FRCGC; there, further research revealed that such patterns were not unique for 1994, but also occurred in other years (Saji et al., 1999).

Among these events, the 1961 event is of particular importance—with remarkable similarity to 1994, the 1961 event was also characterized by positive SST anomalies in the west and central Indian Ocean and negative anomalies further east (Flohn, 1987; Reverdin et al., 1986). Furthermore, the SST anomalies co-occurred with large anomalies of cloud cover and surface winds over the tropical Indian Ocean. This event had caught the attention of the famous German climatologist Hermann Flohn (Craig, 2005), whose research eventually demonstrated that atmospheric circulation anomalies accompanying this unusual Indian Ocean climate signal were behind the catastrophic East African rains of 1961 (Flohn, 1987; Grove, 1996; Kite, 1981; Lamb, 1966; Odingo, 1962).

The excessively rainy season of 1961/62 over East Africa, which peaked in November 1961, was a remarkable event that caused worldwide attention because of its impact on the White Nile. The excessive rains of 1961 had caused a discontinuous rise of Lake Victoria; the subsequent discharge of the White Nile continued for at least 20 years after the event. Furthermore, there were large-scale floods all over equatorial East Africa, resulting in the loss of homes and lives and damage to crops, while emergency food had to be flown into marooned villages (Conway, 2002). The event caused huge economic damages: an estimate by Odingo (1962) suggested that the total flood-damage costs, at the time for Kenya, were around five million British pounds. This mega event over East Africa was accompanied by severe climate anomalies over other countries surrounding the Indian Ocean: India experienced the highest monsoon rainfall ever since records were set up (Grove, 1996), and, at the same time, large parts of the Indonesian archipelago suffered a devastating drought (Flohn, 1987). Most interestingly, there was no El Niño in the Pacific. Noting this remarkable event and its occurrence during a non-ENSO year, Reverdin et al. (1986) suggested that the 1961 event must have involved coupled ocean–atmosphere interactions within the Indian Ocean. However, during this time, neither the general characteristics of IOD events nor the nature of ocean–atmosphere interaction leading to IOD events had been established.

While the 1994 event provided the first evidence of the role of the subsurface ocean in IOD events, it was not widely noted, except by a few researchers. In contrast, the IOD event that occurred 3 years later received worldwide attention. This was, perhaps, because the 1997 IOD event occurred during the strongest El Niño ever observed in the 20th century (Birkett et al., 1999; Murtugudde et al., 2000; Webster et al., 1999). In this year, too, East Africa experienced widespread heavy rainfall, and Indonesia a devastating drought; however, unlike during the 1994 and 1961 IOD events, Indian rainfall was close to normal (Conway, 2002). Although it occurred during a strong El Niño event, the 1997 IOD event exhibited strong coevolution of oceanic and atmospheric variables over the Indian Ocean. This led some researchers to argue that the event arose out of ocean–atmosphere interactions within the Indian Ocean (Murtugudde et al., 2000; Webster et al., 1999). Others, although they examined the same data, argued that Indian Ocean variability was driven by El Niño (Chambers et al., 1999; Yu & Rienecker, 1999).

While the above studies all focused on individual events, Saji et al. (1999) took a different approach. They sought to understand the common features associated with these unusual events in order to explore their underlying dynamics, and to understand their relation to ENSO. To this end, they devised a simple method to identify similar events from observations. This method involved the construction of an SST-based index, the so-called Dipole Mode Index, or DMI. The DMI, defined as the difference in SST anomaly between the western equatorial Indian Ocean (50E–70E, 10S–10N) and the southeastern equatorial Indian Ocean (90E–110E, 10S–Equator), was designed with two considerations in mind—the first exploited the fact that the SST anomaly exhibited a dipolelike structure when the events were at their peak strength. Second, it provided a simple filter to remove the ENSO-induced basinwide SST anomaly pattern. Note that although SST anomaly in the eastern tropical Indian Ocean is a defining aspect of IOD events, the presence of the ENSO-induced basinwide SST anomaly renders it impossible to separate the two SST modes using only SST from the eastern tropical Indian Ocean.

Using the DMI, Saji et al. (1999) identified six extreme IOD events—those of 1961, 1967, 1972, 1982, 1994, and 1997—during the 1958 to 1998 period. Of these, the events of 1972, 1982, and 1997 coincided with strong El Niño events. However, there were IOD events that occurred independent of which phase ENSO was in at the time of IOD development: thus, the events of 1961, 1967, and 1994 coincided with no ENSO, La Niña, and a weak El Niño, respectively. This led Saji et al. (1999) to conclude that IOD events were independent of ENSO; they also cited the weak simultaneous correlation between the time series of DMI and an ENSO index as further evidence in support of their conclusion.

To account for the existence of IOD events, they invoked the existence of an inherent mode of climate variability over the tropical Indian Ocean that was generated and maintained by ocean–atmosphere interactions. The original analysis of Saji et al. (1999) used a composite IOD event to argue for the existence of an inherent coupled mode in the tropical Indian Ocean. The next section of this article recreates their arguments for coupled air–sea interactions during IOD. However, instead of using composite analysis, the discussion examines the strong IOD event of 2006—an event that occurred some years after Saji et al. (1999) published their results.

Anatomy of an IOD Event

Figure 1 displays the evolution of SST and surface wind anomalies during the 2006 IOD event. Note that the evolution of the 2006 event is remarkably similar to the composite IOD event (Saji et al., 1999). The developing stage (top panel of Figure 1a) is characterized by cool SST anomalies that appear south of Sumatra and Java; these are accompanied by low SSH anomalies over the same region and strengthened southeasterly trades at the southeastern Indian Ocean. At the same time, the normally eastward-directed surface winds along the Equator have considerably weakened. Furthermore, at this stage, warm SST anomalies are only weakly developed over the central tropical Indian Ocean; in the ensuing months, cool SST anomalies intensify, while migrating toward the Equator along the Indonesian coastline. There is a clear lag in the development of warm SST anomalies in the central and west Indian Ocean, with the SST dipole structure clearly established only during the boreal fall (Figure 2). At this time, the dipole structure is also very clear in SSH and Outgoing Longwave Radiation (OLR) anomalies; the latter field is a proxy for rainfall.

Figure 1. Spatial structure of anomalies for the 2006 IOD event during its development (top), mature (middle), and dissipating (bottom) phases. The shaded contours in (a), (b), and (c) are for SST, SSH, and OLR anomalies, respectively. Surface wind anomalies are overlaid as vectors.

Data Sources: SST, TMI (Wentz et al., 2000); SSH, AVISO (Le Traon et al., 1998); OLR, NOAA (Liebmann & Smith, 1996); surface winds, NCEP Reanalysis 1 (Kalnay et al., 1996).

Figure 2. The coevolution of the time series of DMI (dark grey bars) and equatorial (80E–110E, 5S–5N) zonal wind anomalies (light grey bars) during the 2006 IOD event. Note the clear lag in the establishment of SST anomalies over the western equatorial Indian Ocean (60E–80E, 10S–10N, red line) relative to that over the eastern equatorial Indian Ocean (90E–110E, 10S–Equator, blue line). The time series shown here are normalized by their respective standard deviations.

The coevolution of the zonal SST gradient with the equatorial zonal wind anomalies, depicted in Figure 2, is a characteristic feature of IOD events. It provides strong empirical evidence for ocean–atmospheric interaction as the inherent mechanism driving IOD. The temporal evolution of IOD is characterized by an initiation phase in late spring to early summer, then there is a rapid development of coupled variability during summer and a peak in fall, and the dissipating phase, occurring in early winter, is characterized by the abrupt disappearance of cool SST anomalies in the eastern Indian Ocean.

It can be shown that the SST, SSH, and OLR fields displayed in Figure 1 are coupled to each other. The low SSH anomaly, in the eastern half, is a dynamic response to the equatorial easterly wind anomaly and is shaped by equatorial and coastal Kelvin waves (Feng et al., 2001; McCreary, 1976; Rao & Behera, 2005; Rao et al., 2002). Low SSH anomalies, in turn, lead to cool SST anomalies through oceanic processes, such as entrainment, that couple the subsurface and surface ocean temperatures (Behera et al., 1999; Cai et al., 2013; Murtugudde et al., 2000; Vinayachandran et al., 1999, 2007).

The feedback loop described above explains how surface winds affect SST through their impact on ocean dynamics. To close the loop, we need to consider how SST anomalies feedback to the surface wind anomalies. The immediate impact of cool SST anomalies is to reduce rainfall in the vicinity of the eastern equatorial Indian Ocean; this introduces diabatic heating anomalies in the middle troposphere that can force additional surface wind anomalies (Gill, 1980).

The nature of surface wind anomalies introduced by the SST anomaly is demonstrated through the numerical simulation shown in Figure 3. We forced a linear multi-level atmospheric model with diabatic heating (shaded contour) anomalies representative of IOD events (Hameed et al., 2018). Interested readers are referred to Watanabe and Kimoto (2000) for the details of the atmospheric model used in this simulation. Here, the model and the experiment are only briefly described before the results are discussed further. The atmospheric model (T42 resolution, 20 vertical sigma levels) uses the so-called primitive equations to model the dynamics of atmospheric flows. The model equations are linearized about a specified background state. For the experiment, we used the model linearized about the boreal summer climatological atmospheric state. Momentum damping in the model is through a linear drag that mimics Rayleigh friction. The model has no physics to generate diabatic heating, so we have prescribed an idealized heating that approximates the horizontal structure OLR anomaly shown in the top panel of Figure 1a. In the vertical, we prescribed the diabatic heating to have a single maximum at the midtroposphere. The surface wind anomalies simulated by the numerical model in response to negative diabatic heating anomalies at the eastern equatorial Indian Ocean have a structure very similar to that shown in Figure 1. This implies that the response to the SST anomaly acts in such a way as to enhance the SST anomaly further. In other words, there is a positive feedback loop between surface winds and SST involving thermocline and midtroposheric thermal perturbations as intermediary processes, and this feedback loop can explain the genesis and maintenance of IOD variability. The presence of such feedback has also been demonstrated in multiple coupled model experiments conducted by various groups all over the world (Behera et al., 2006; Crétat et al., 2017; Gualdi et al., 2003; Wang et al., 2016; Yang et al., 2015).

Figure 3. Numerical simulation of surface winds with a linearized primitive equation atmospheric model forced by a patch of diabatic heating anomalies (shaded contour) over the eastern equatorial Indian Ocean. Thick contours represent diabatic heating anomalies of −0.8 and −1.0 K/day; a rainfall rate of 10 mm/day corresponds to a diabatic heating rate of about 2.5 K/day. Over the eastern Indian Ocean, the standard deviation of rainfall during the early summer is about 4 mm/day.

The reader may notice that, although the preceding discussion provides an overall description of the coupled ocean–atmospheric processes leading to IOD variability, it fails to explain certain important features that stand out from Figures 1 and 2. For example, it is not clear why IOD is so strongly phase-locked to the annual cycle, and why it dissipates rather quickly after the peak. Another interesting feature is the relation between the SSH and SST anomalies. One can notice that, although the SSH anomaly is very symmetric about the equator at the eastern Indian Ocean, the SST anomaly is manifested only south of the equator. The explanation of these unique features of IOD variability can be sought in the mean thermal structure of the Indian Ocean, which is strongly affected by the monsoonal circulation and its strong annual cycle.

Understanding the Unique Features of IOD Variability

The IOD phenomenon is generated over a unique atmospheric ocean state over the equatorial Indian Ocean. Understanding this unique state is important for understanding the properties of IOD, including its spatial structure and the mechanisms that maintain and dissipate it; furthermore, such an understanding will also help advance projections of future IOD behavior.

Figure 4 depicts the depth-longitude section of ocean temperatures (Ridgway et al., 2002) averaged over 5S to 5N. Along the equator, there is a pronounced zonal tilt of the thermocline in all the three oceans; however, the direction of the zonal thermocline tilt over the equatorial Indian Ocean is unique: the thermocline is deepest at its eastern end—in sharp contrast to the other oceans, where it is shallowest at the eastern end. Over the Pacific and Atlantic Oceans, easterly trades drive equatorial upwelling, leading to a shallow thermocline at the east. In addition, the westward wind stress on the equatorial ocean, acting in the presence of the western boundary, makes the thermocline deeper at the west. On the other hand, surface wind stress over the equatorial Indian Ocean is toward the east throughout the year; this creates a deep thermocline in the equatorial eastern Indian Ocean.

Figure 4. Longitude-depth section of ocean temperatures averaged along the equator (5S to 5N). Data are monthly mean climatological values averaged between June and November.

Data Source: CARS 2009 (Ridgway et al., 2002).

A shallow thermocline is conducive for ocean–atmosphere interaction on interannual timescales, as the study of the ENSO phenomenon has amply demonstrated (McCreary, 1976; McCreary & Anderson, 1985; Philander, 1990). The shallower the thermocline, the more effectively the SST influenced is by temperature perturbations in the thermocline (Xie et al., 2002). By effectively coupling the thermocline to the surface ocean, a shallow thermocline allows feedback from the subsurface ocean to affect the tropical atmosphere, which is sensitive to SST perturbations.

On the other hand, the temperature perturbations in the thermocline are a result of oceanic processes directly related to wind-stress perturbations at the ocean–atmosphere interface; one such process occurs when surface winds force internal waves at the thermocline (McCreary, 1976). The subsequent propagation of internal waves, as westward-propagating Rossby waves and eastward-propagating Kelvin waves, also allows wind perturbations to affect thermoclines located far from the forcing region. The surface wind perturbation can also locally affect subsurface temperatures by changing the rate of oceanic upwelling. Where the thermocline is deep, any subsurface perturbation in response to wind forcing has weaker feedback to SST and thereby to winds. Therefore, a deep thermocline is not conducive for the development of ocean–atmosphere interaction.

In the presence of sharp horizontal gradients, SST can also be effectively modulated by wind-forced ocean currents through advective processes. The eastern edge of the warm pool is an area where such processes have been demonstrated to affect air–sea interaction during ENSO events (Clarke, 2008; Gill, 1983; Picaut et al., 1997). However, over the eastern equatorial Indian Ocean, the horizontal temperature gradient is not large enough for such processes to facilitate ocean–atmosphere interaction.

Thus, on the face of this unique situation, it may be thought that ocean–atmosphere interactions of the kind present during ENSO cannot develop over the equatorial Indian Ocean. How then does the ocean–atmosphere interaction needed to maintain IOD develop in the unique setting of the equatorial Indian Ocean?

The answer to this mystery lies in the existence of a unique shallow thermocline region that is maintained by coastal upwelling south of the equator, off Java in the eastern equatorial Indian Ocean (Wyrtki, 1962).

Figure 5a depicts ocean temperatures at 50-m depth in the near equatorial Indian Ocean. It is seen that subsurface temperatures at this depth increase dramatically to the east, with the west coast of Sumatra experiencing temperatures in excess of 29C. The climatological surface wind field overlaid on top of the temperature contours reveals that this is closely related to the easterly wind stress experienced by the equatorial Indian Ocean. On the other hand, subsurface temperatures cooler than 25C occur in three zones, two of them located south of the equator. The upwelling associated with all of these zones is evidenced by their high chlorophyll concentrations (Figure 5b).

Figure 5. Zones of near-equatorial upwelling are depicted with maps of ocean temperature at 50 m (left) and satellite observed surface chlorophyll concentration (right). Surface winds are plotted as vectors on the left panel. Data are monthly mean climatological values averaged between June and November.

Data sources: ocean temperature, CARS 2009 (Ridgway et al., 2002); surface wind, NCEP reanalysis (Kalnay et al., 1996); chlorophyll, SeaWifs (Hooker et al., 1992).

Here, the focus is on the coastal upwelling zone along the Java coast situated around 8S, as it is considered to be the most significant factor in the initiation and maintenance of IOD events (Delman et al., 2016). The upwelling is a seasonal phenomenon (Qu et al., 2005; Varela et al., 2016; Wyrtki, 1962) and is caused by along-shore trade winds, which are most effective in producing coastal upwelling between June and November, i.e., during the boreal summer monsoon season (Varela et al., 2016). It is to be noted, however, that while the seasonal cycle of these trade winds is important in modulating upwelling on the annual timescale, their perturbation on interannual timescales is not significant enough to produce interannual changes in coastal upwelling intensity. For these, the most important factor is the variability of surface winds over the equatorial Indian Ocean (Chen et al., 2016; Delman et al., 2016), which generate oceanic equatorial Kelvin waves that propagate along the equator and subsequently along the Sumatra-Java coasts as coastally trapped Kelvin waves.

Figure 6a depicts a vertical cross-section of ocean temperatures along 110E, from 15S to 8S. The climatological mean temperatures during August are shown in the figure. There is a sharp rise of isotherms from about 15S toward the Java coast. For example, the 25C isotherm rises from 100-m depth around 15S to about 20-m depth at the coast of Java. Interestingly, the warmest ocean temperatures in this cross-section do not occur at the surface, as one might expect, but at a depth of 60 m around 14S. This core of warm ocean temperatures reflects the transport of relatively warm and fresh water from the Pacific Ocean to the Indian Ocean through the Indonesian seas, the so-called Indonesia throughflow (ITF; Sprintall et al., 2009; Wijffels & Meyers, 2004).

Figure 6. Upwelling off Java is most clearly developed during the southeast summer monsoon season. Climatological ocean temperature section at 110E during August (left) and March (right); note that temperatures between 23C and 24C are shaded white. The annual cycle of mean ocean temperature, averaged from the surface to 50 m, is plotted along 110E (middle).

Source: CARS 2009 (Ridgway et al., 2002).

The coastal upwelling along Java has a strong seasonal cycle. This is shown as the depth-averaged ocean temperatures from the surface to 50 m as a function of the seasonal cycle in Figure 6a. Off Java, SST cooler than 26C is found between June and November, with the strongest cooling occurring in August; during this period, the cooling also has its maximum meridional extent, reaching as far south as 11S. It is clear that conditions favorable for upwelling, and by implication ocean–atmosphere interaction, exist only during the southeast monsoon. After November, ocean temperatures warm up considerably, reaching 27.5C in March at the Java coast; further, the signal of upwelling disappears as the thermocline becomes flat from 15S to the coast of Java, while migrating down to a depth of about 60 m. This is associated with the reversal of monsoon winds along the coast of Java during the northwest monsoon, which favors downwelling (Varela et al., 2016), rendering conditions unfavorable for feedback from the thermocline to the sea surface. This annual cycle of upwelling is very important in understanding the strong phase-locking of IOD to the annual cycle.

A second zone of oceanic upwelling exists in the central to west Indian Ocean, just to the south of the equator (Xie et al., 2002). Here, upwelling is produced by an entirely different property of the surface wind: it is the negative wind curl, between the southeasterly trades and equatorial westerlies, that is responsible for the unique open ocean upwelling zone. Figure 7 shows the correlation between ocean temperature anomalies at 100-m depth with that at the surface, for the four seasons. The data for this calculation are from the SODA2.4 reanalysis (Carton & Giese, 2008) and cover the period 1958 to 2010. In the south equatorial Indian ocean, strong correlation between the subsurface and surface temperatures is noted in the vicinity of the Java upwelling zone; to its west, positive correlations occur in the open-ocean upwelling zone in the west and central Indian Ocean. There is also a strong seasonality in the correlation, with maximum positive correlations observed during boreal summer and fall.

Figure 7. Correlation coefficients of ocean temperature at 100-m depth with that at the surface over the tropical Indian Ocean during boreal (a) spring, (b) summer, (c) fall, and (d) winter. Thick-line contours denote absolute correlations of 0.4 and 0.6.

Data Source: SODA reanalysis v2.4 (Carton & Giese, 2008).

The positive correlation between SST and subsurface temperatures in the eastern Indian Ocean reflects the seasonal cycle of upwelling in this region, with the positive relation disappearing once the northwest monsoon sets in, and the mean thermocline deepens off Java. This effectively cuts off the coupling between the subsurface and the surface, and abruptly terminates the positive feedback loop between the surface wind and SST anomalies. At the same time, the reduced cloudiness associated with IOD (Figure 1c) implies that the ocean surface receives abnormally high solar insolation, which tends to increase the SST anomaly. This is reflected in the negative correlation between SST and subsurface temperatures seen in Figure 7 during the boreal winter, with a positive SST anomaly developing at the end of the IOD cycle, while subsurface temperatures are still cooler than normal. Thus, it is the unique seasonal setting of the equatorial subsurface ocean, driven by the strong annual cycle of the Asian monsoon, that controls the spatiotemporal features of the IOD event—in particular, its strong phase-locking to the seasonal cycle and its abrupt dissipation in early boreal winter (Saji et al., 1999).

Positive and Negative IOD Events

So far, IOD has been introduced as a powerful modulator of Indian Ocean climate that renders the normally warm and wet climate of the eastern Indian Ocean cooler and dryer than normal. This is, however, only the positive phase of IOD; in its negative phase, IOD drives Indian Ocean climate in the opposite direction, rendering wetter than normal conditions at the equatorial eastern Indian Ocean. To depict this remarkable see-saw in ocean–atmosphere state during the positive and negative phases of IOD, Figure 8 shows the state of SST, rainfall, surface winds, and sea level during two strong IOD events. In this figure, the data are not plotted as departures from normal conditions, but the actual state of the variable itself is plotted. In each subfigure, the top panel depicts conditions during the strong negative IOD of 2005, while the bottom panel is for the 2006 (positive) IOD.

Figure 8. Absolute fields of (a) SST, (b) rainfall, (c) surface winds, and (d) SSH during opposite phases of IOD. The plots on the top row of (a), (b), (c), and (d) represent absolute fields during the mature phase of the 2005 negative IOD; bottom rows of (a), (b), (c), and (d) are similar, but for the 2006 positive IOD.

Data sources are the same as in Fig. 1, except for rainfall, which is from CMAP (Xie & Arkin, 1997).

What is remarkable about the figure is the striking east–west migration of climate state between the phases of IOD. In the negative phase, extremely warm SST covers the eastern Indian Ocean, nearly cutting off the seasonal upwelling off Java (Figure 8a, c); there is a pronounced zonal SST gradient, with temperatures clearly increasing from west to east along the equator. As a result of the warm eastern temperatures (Figure 8b), heavy rainfall occurs over Indonesia (Figure 8c), while conditions are markedly dry in the western Indian Ocean. These are accompanied by very strong westerlies over the central equatorial Indian Ocean, leading to a pronounced east–west slope in the sea level (Figure 8d).

These conditions nearly reversed during a strong positive IOD event, when the warmest temperatures around the equator are not at the eastern end, but over the central Indian Ocean. SST over the eastern Indian Ocean drops markedly below 27C, leading to a nearly complete disappearance of rainfall over the southeastern Indian Ocean (D’Arrigo & Wilson, 2008; Saji & Yamagata, 2003a); instead, the rain band migrates to the warmer oceans to the north and west, leading to unusually persistent and abundant rainfall over Sri Lanka (Zubair et al., 2003) and equatorial East Africa (Behera et al., 2005; Birkett et al., 1999; Conway, 2002; Flohn, 1987; Saji & Yamagata, 2003a; Ummenhofer et al., 2009b). There is a complete disappearance or reversal of the westerly surface wind jet over the equatorial central Indian Ocean, which nearly flattens or reverses the east–west slope of sea level.

So far, we have looked at the properties of IOD through the two events of 2005 and 2006. It is of interest to examine other IOD events that are recorded in the observational record, and to see how their features compare with the events during 2005 and 2006. In order to do this, we have identified positive and negative IOD events for the period 1958 to 2015, based on the methodology of Saji and Yamagata (2003b). This methodology not only uses the DMI to detect IOD events, but also checks if SST anomalies over the eastern Indian Ocean and surface wind anomalies over the equatorial Indian Ocean are in the right phase. Specifically, SST over the eastern Indian Ocean is required to be cooler than normal for the event to be classified as a positive IOD, and warmer than normal for it to be a negative IOD. Strong events are IODs that are required to have DMI and other criteria above 1 standard deviation for more than 3 months; however, moderate events need only hold these criteria above the 0.5 standard deviation threshold. Further, the basinwide anomaly associated with ENSO evolution was filtered out of the SST anomalies. Using this methodology, six strong positive IOD events (1961, 1963, 1967, 1994, 1997, and 2006) and six strong negative IOD events (1960, 1975, 1992, 1996, 1998, and 2005) were identified between 1958 and 2015. Figure 9 depicts all the IOD events detected during this period.

Figure 9. Time series of standardized DMI, averaged from June to November, from 1958 to 2015; orange (blue) bars indicate strong positive (negative) IODs; light green (turquoise) bars are moderate positive (negative) IODs. If, for a strong IOD, June to November averaged Nino3 exceeded 1σ‎, that El Niño/La Niña is marked by a filled circle. If, for a moderate IOD, June to November averaged Nino3 exceeded 0.5σ‎, that El Niño/La Niña is marked by an open circle.

Data sources: COADS (Woodruff et al., 2011), from 1958 to 1981; OISST (Reynolds et al., 2002) thereafter.

The Fourier spectra of the DMI time series, shown in Figure 10a, brings out three distinct peaks at 2, 3, and 4.5 years. Interestingly, the spectra are unique in that there is little indication of an increase of power with period. The clear biennial signal reflects the tendency for IOD events of opposite phase to follow each other (Feng & Meyers, 2003; Saji et al., 1999). The timescale is determined by the width of the Indian Ocean and the time taken for equatorial waves to travel to the western boundary from the eastern Indian Ocean, and back to the eastern boundary after reflection (Feng & Meyers, 2003). The 3- and 4.5-year timescales appear to be return periods associated with IOD, with the periodicity shifting over time (Figure 10b; Ummenhofer et al., 2017). An example is the strong 3-year peak is associated with a prominent 3-year return period of IOD during the 1990s, while there are longer return periods observed before and after the 1990s.

To a first order, negative IOD events are the mirror image of positive IOD events; however, there are some important differences (Cai et al., 2012; Ng & Cai, 2016; Ummenhofer et al., 2009b). Figure 11 illustrates this using SSH and zonal wind anomalies during positive (a, c) and negative (b, d) IOD events. A prominent difference is in the amplitude—positive events are stronger than negative events. This can be understood as a consequence of the deep thermocline in the eastern Indian Ocean: a shallowing thermocline is more effective in generating surface cooling than a deepening one is in generating surface warming (Cai et al., 2013).

Figure 10. Spectral analysis of standardized DMI values averaged from June to November. The Fourier spectrum shown in (a) has three distinct peaks around 2, 3, and 4.5 years. Peaks significant at the 95% significance level against the background spectra (red line) are shaded. The wavelet spectra shown in (b) suggest that the multiple peaks are associated with shifts of IOD periodicity over time.

Figure 11. Composite anomalies of SSH (left) and zonal surface wind (right) during positive (a, c) and negative (b, d) IOD events.

Data Sources: SSH, AVISO (Le Traon et al., 1998); surface winds, NCEP Reanalysis-1 (Kalnay et al., 1996).

To examine the common structure of IOD events, we resort to a correlation analysis using the DMI time series. In Figures 12 and 13, the covariation of September to November (SON) averaged data of surface winds, SST, land rainfall, and marine cloudiness anomalies with DMI is depicted. All the data are correlated with SON values of DMI, the season when IOD attains its peak. Besides this, the June to August (JJA) averaged values of rainfall and marine cloudiness are also correlated with SON DMI. The correlation of DMI with surface wind and SSH anomalies for the boreal fall season is shown in Figure 12, while that with marine cloudiness and land rainfall is shown in Figure 13.

Figure 12. Correlation of (a) zonal winds and (b) SSH with the DMI index for the boreal fall season.

Data sources: SSH, SODA (Carton & Giese, 2008), for the period 1958 to 2010; surface winds, NCEP Reanalysis-1 (Kalnay et al., 1996), for the period 1958 to 2015.

Figure 13. Correlation of September-October-November DMI with land rainfall and marine cloudiness during (a) June–July–August and (b) September–October–November. Marine cloudiness and land rainfall were separately correlated with DMI and the correlation maps were then merged together for this plot.

Data sources: Cloudiness, COADS (Woodruff et al., 2011); land rainfall, GPCC (Ziese et al., 2011). Data are for the period 1958 to 2014.

The remarkable structural similarity between these maps and the case study shown in Figure 1 is very striking. In addition to this, the correlation maps also help to gauge the contribution of IOD to regional variability. As expected (Figure 2), there is a strong correlation between the SST variability associated with IOD and equatorial zonal winds, with the correlation coefficient exceeding 0.8 over the central equatorial Indian Ocean. Similarly, IOD can be seen to contribute to a significant part of the sea-level variability over the tropical Indian Ocean. The prominent dipole structure of SSH anomalies and SST anomalies associated with IOD during the peak phase is also reflected in rainfall anomalies (Figure 13). Over the ocean, there are no sustained direct observations of rainfall. The only data set that can be considered a proxy for rainfall variations prior to the satellite era is marine cloudiness, which is essentially observations of cloud cover from a ship’s deck. Correlating cloudiness anomalies with the DMI provides a qualitative look at the structure of rainfall patterns associated with IOD. There is a striking similarity with the pattern of rainfall anomalies observed during 2006. Further, the correlation maps nicely align with the pattern obtained from correlation analysis between DMI and land rainfall, completing the picture of IOD’s signature on regional rainfall.

IOD–ENSO Interaction

The mechanisms that generate and sustain IOD lie entirely within the equatorial Indian Ocean, just as ENSO’s genesis mechanisms lie entirely within the Pacific basin. However, the physical proximity between the Indian and Pacific basins and the existence of oceanic (Wijffels & Meyers, 2004) and atmospheric pathways (Barnett, 1984; Hameed et al., 2018) between them raise questions about the possibility and nature of interactions between the two phenomena—a question first raised by McCreary and Anderson (1985).

The problem of IOD–ENSO interaction is rendered interesting due to the significant correlation that exists between the time series of IOD and ENSO during the boreal fall season: SON values of DMI and Nino3 are correlated at 0.55. This correlation has been interpreted in multiple ways over the years (for a review, see Saji & Yamagata, 2003b; and Yamagata et al., 2004). Co-occurrences of El Niño with positive IOD, and La Niña with negative IOD, account partly (Figure 9) for the large correlation. However, external factors, such as the global warming trend, decadal variations, and the ENSO-induced basinwide SST anomaly, introduce a spurious zonal SST gradient in the Indian Ocean SST (Saji & Yamagata, 2003b), which also partly contributes to DMI’s high correlation with Nino3 during boreal fall.

Before we can interpret the correlation between DMI and Nino3 indices in terms of physical processes, it is necessary to remove spurious zonal gradients introduced by extraneous factors. The construction of DMI as a zonal difference of SST anomaly was aimed to filter out the ENSO-induced basinwide SST anomaly. Because of this differencing procedure, any zonally inhomogeneous extraneous signal will project on the DMI. Neither the global warming signal (Alory et al., 2007) nor the decadal signal (Han et al., 2014) in Indian Ocean SST is zonally homogeneous; this has been shown to bias the correlation between IOD and ENSO (Meyers et al., 2007; Saji & Yamagata, 2003b).

The magnitude of the spurious DMI introduced by the ENSO-induced basinwide SST anomaly is particularly large (Saji & Yamagata, 2003b). The basinwide SST anomaly is a consequence of equatorial atmospheric adjustment to ENSO’s midtropospheric diabatic heating perturbations (Yulaeva & Wallace, 1994): atmospheric Kelvin waves that emanate from the diabatic heating anomalies spread the ENSO signal over the equatorial belt. The ENSO signal is nonuniformly expressed on Indian Ocean SST anomalies, with the signal established much later over the eastern tropical Indian Ocean (Meyers et al., 2007; Saji & Yamagata, 2003b) than over the western and central tropical Indian Ocean (the eastern Indian SST anomaly maximally correlates with Nino3 at 6-months lag, while the western Indian SST anomaly lags it by only 3 months). The definition of DMI assumes that there is no zonal homogeneity in the ENSO footprint, but, as discussed, this is not correct; hence, a spurious zonal SST gradient not associated with IOD is introduced. The spurious effect may be reduced by eliminating the estimated basinwide anomaly: one way is using lagged regression with ENSO (Saji & Yamagata, 2003b), while another way is to use lagged Empirical Orthogonal Function (EOF) analysis (Meyers et al., 2007). Removing the spurious DMI values resulted in the correlation’s dropping to 0.43 from the original value of 0.55 (a 39% drop in the variance explained relative to the original).

If the correlation reported in the preceding paragraph is interpreted in terms of the rate of co-occurrence between the two modes, a correlation of 0.43 implies that about 19% of IOD and ENSO events co-occur. Figure 9 shows the co-occurrence of ENSO events during moderate and strong IOD events. ENSO is considered to have co-occurred if the Nino3 index exceeded a normalized amplitude of 0.5σ‎ for moderate IOD events and 1σ‎ for strong IOD events, when averaged over the period of IOD development (June to November). From this, it can be seen that the positive IOD events of 1963, 1972, 1976, 1982, and 1997 co-occured with El Niño events, while the negative IOD events of 1964, 1971, 1981, 1996, and 2013 co-occurred with La Niña events. However, this implies a higher co-occurrence rate, of about 30%. Thus, the interpretation of the correlation representing co-occurrence is perhaps not appropriate.

Then, why is the ENSO–IOD correlation substantially weaker in the light of a 30% rate of co-occurrence? A clear explanation emerges from the consideration of the high scatter between the intensity of IOD and ENSO events (Figure 14). Strong IOD events occur during strong (1963, 1997) and weak (1994, 2006) El Niño as well as La Niña (1961, 1967) events. Notice also the lack of IOD events during the strong El Niño of 2015 and the strong La Niña of 2007. Figure 14 shows that there is a propensity for La Niña and negative IOD to often happen together. This is also seen in the correlations, when, upon removing negative IOD events from the calculation, the correlation becomes insignificant (0.11). The regression line (red line) between Nino3 and DMI is also shown in Figure 14. One can see that it is not a very good fit for most of the data points and is strongly affected by outliers. In fact, upon removing the five extreme years of 1996 and 2010 (negative IOD with La Niña) and 1972, 1982, and 1997 (positive IOD with El Niño), the correlation drops to 0.28.

Figure 14. Scatter plot with Nino3 on the x-axis and DMI on the y-axis. Both indices were averaged for the September to November period and were normalized by their respective standard deviations. Years of occurrence are shown for events exceeding 1σ‎. IOD events occurring during El Niño are colored red, and those during La Niña are blue.

Data source: COADS (Woodruff et al., 2011) from 1958 to 1981 and OISST (Reynolds et al., 2002) thereafter.

One way to account for the consistent but nonlinear relation between IOD and ENSO is that ENSO can on occasion trigger IOD (Annamalai et al., 2003; Feng & Meyers, 2003), with the strength of IOD controlled by the character of coupled instability and external noise (such as those associated with intraseasonal oscillations) within the tropical Indian Ocean basin, while on other occasions IOD may be triggered from within the Indian Ocean. For this argument to be valid, ENSO dynamics should be capable of introducing significant anomalies over the eastern Indian Ocean, which is the key region for the genesis of IOD events. Figure 15 shows the correlation of Nino3 index with SST and its regression with surface winds over the period 1958 to 2015. Due to the correlation with IOD, such an analysis would also contain the impacts of IOD as well. To deduce the impact of ENSO alone, there are various possible ways. For example, Saji and Yamagata (2003a) used partial correlation analysis. Here, as an alternative method, we removed co-occurring events from the analysis: specifically, we removed the 1972, 1982, 1998, 1997, and 2010 events before performing the correlation/regression analysis. In early summer, ENSO is associated with weak westerly wind anomalies over the equatorial Indian Ocean (Figure 15, also see Figure 14 of Saji & Yamagata, 2003b) that act to decrease upwelling over Java. However, despite this, there are weak cool SST anomalies between Java and Australia. Wajsowicz and Schneider (2001) have shown that reduction of the Indonesian throughflow (ITF) can result in cool SST anomalies between Java and Australia. During El Niño, the ITF is reduced, due to the lowering of sea level over the western Pacific. Therefore, despite the sign of ENSO-associated winds being unfavorable, cool SST can be introduced over the eastern Indian Ocean. However, the core of the ITF is clearly separated from the Java upwelling zone (Figure 6a) and the SST anomaly introduced is quite weak. Therefore, it is questionable whether ocean–atmosphere teleconnections associated with ENSO are strong enough to trigger an IOD event early on its development phase. The signal does get stronger in the fall season, as easterly wind anomalies associated with ENSO get stronger (Saji & Yamagata, 2003a, 2003b). However, by this time, it may be too late to initiate IOD development. On the other hand, when an ENSO event co-occurs with an IOD, these mechanisms may strengthen the cooling over the eastern Indian Ocean and allow the IOD event to persist longer.

Figure 15. Correlation of September-October-November Nino3 index with Indian Ocean sea surface temperature anomalies during (a) June–July–August and (b) September–October–November. Also shown as vectors are the regression of Nino3 index with surface winds. To reduce the impact of co-occurring IOD-ENSO events on this analysis, the five extreme years of 1998 and 2010 (negative IOD with La Niña) and 1972, 1982, and 1997 (positive IOD with El Niño) were removed before the correlation and regression analysis were carried out.

However, since correlation does not provide any information on the direction of causality, it is logical to consider the other alternative, that it is IOD that affects ENSO. Hameed et al. (2018) provided an analysis that demonstrates that IOD events can have a strong impact on ENSO evolution. They argue from observational analysis and modeling experiments that the 1994 and 2006 El Niño-like events were a result of IOD forcing. Further, they demonstrate that super El Niños—the likes of 1972, 1982, and 1997 events that are characterized by strong SST variability in the far-eastern Pacific—are a result of the interaction between IOD and ENSO dynamics.

Regional Impacts

The most striking feature of IOD is the marked dipole structure, signifying an opposing tendency of ocean–atmosphere perturbations oriented in a zonal direction across the Indian Ocean. This marked east–west migration of ocean–atmosphere state associated with IOD drives significant changes in the oceanic and atmospheric environment with severe implications for regional societies, economies, and ecosystems. This article reviews some of the impacts described in the literature. For simplicity, the impacts are interpreted with reference to the positive phase of IOD.

In terms of ocean circulation, one of the most significant features of the Indian Ocean is the ITF. As the only major low-latitude connection in the global oceans, the Indonesian seas permit the transfer of Pacific waters into the Indian Ocean. The importance of the ITF for the global thermohaline circulation (Hirst & Godfrey, 1994; Schiller et al., 1998) and global atmospheric circulation (Schneider, 1998) is well recorded in the literature. On the seasonal timescale, the strong annual cycle of the monsoons has a strong impact on the ITF transport (Masumoto & Yamagata, 1996), with the strongest transports during the southeast monsoon, when coastal upwelling is well developed and sea level decreases at the Java coast (Wyrtki, 1987). The strong wind perturbations at the equator during IOD events (Figures 1 and 8c) can therefore substantially modulate the ITF (Meyers, 1996; Sprintall et al., 2009) through the wind’s impact on sea level at the eastern Indian Ocean (Figure 8d). During a positive IOD event, sea level becomes abnormally low over the eastern Indian Ocean through the impact of equatorial winds. This leads to an increase of the pressure head between the Pacific and Indian Oceans (Masumoto & Yamagata, 1996; Meyers, 1996; Wijffels & Meyers, 2004; Wyrtki, 1987) that can increase ITF transport during positive IOD years. However, some of the positive IOD events have in the past occurred at the same time as an El Niño event. El Niño events are associated with reduced sea level in the western Pacific. This can lead to a decrease of the pressure head between the oceans and lead to a reduced ITF transport. Only in recent years are detailed observations of the ITF becoming to be available. These observations are beginning to provide insights into the control of ITF during IOD and ENSO events (Hu & Sprintall, 2016; Liu et al., 2015; Meyers, 1996; Sprintall & Révelard, 2014; Wijffels & Meyers, 2004). Of particular interest are the observations described by Sprintall et al. (2009) of the opposing anomalies of ITF transport during the negative and positive IOD events of 2005 and 2006.

Staying with the ocean, another important effect of the anomalies of sea level and upwelling intensity during IOD is on Indian Ocean’s biological productivity. The seasonal upwelling off Java controls biological productivity in that region (Figure 5b); thus, the strong SSH variability during IOD events has large implications for fisheries, both coastal and offshore (Amri, 2012; Lumban-Gaol et al., 2015). It is during the upwelling season that catches of pelagic species, such as sardine (Ghofar, 2005) and tuna (Lumban-Gaol et al., 2015), are the highest. Further, offshore transport of the upwelled water and nutrients, by submesoscale filaments and eddies, inject plankton-rich waters into key fish spawning areas (Matsuura et al., 1997), which lie south of Java. However, fish catch not only is related to fish abundance but also is constrained by practical matters, such as the depth to which fishing lines are cast. An interesting situation is seen with respect to the Bigeye tuna (Thunnus obesus), which are abundant in the eastern Indian Ocean. These fish prefer an ambient temperature of 10C to 15C, which is normally found at depths of 150 m to 400 m (Hanamoto, 1987; Lumban-Gaol et al., 2015). However, the tuna longline used in conventional fishing vessels is set to reach water depths of only 100 m to 280 m. While biological productivity off Java is controlled by IOD, the thermocline depth of the waters where Bigeye tuna are abundant is also strongly controlled by the ITF transport. Although during positive IOD years the thermocline can migrate upward, leading to a shallowing of the Bigeye tuna fishing layer by at least 50 m (Lumban-Gaol et al., 2015), the increased ITF transport can counter this. When a positive IOD co-occurs with El Niño, enhanced biological productivity can lead to fish abundance. At the same time, the shallower thermocline enables the tuna longline to penetrate deep into the fishing layer, thereby increasing the fish catch. These observations appear to be consistent with the substantial increases of Bigeye tuna catch during the co-occurring IOD/El Niño of 1997 (Amri, 2012; Lumban-Gaol et al., 2015; Syamsuddin et al., 2013), and with the observation that the catch was not substantially high during the 1994 IOD event. In contrast, at the western Indian Ocean, primary productivity is negatively correlated with IOD (Lan et al., 2013). Lan et al. (2013) found that during positive IOD events, catch distributions of tuna were restricted to the northern and western margins of the Indian Ocean. Furthermore, they found that during negative IOD events, tuna catches expand into the central regions of the western Indian Ocean. These observations are consistent with the sea-level distribution shown in Figure 8d associated with positive and negative IOD events.

On the other hand, the zonal variation of IOD precipitation across the Indian Ocean wreaks havoc on countries situated on both the eastern and western coasts of the Indian Ocean. During positive IOD events, Indonesia suffers from severe droughts (Saji & Yamagata, 2003a) that persist over two seasons (Figure 13) and often trigger forest fires, with severe implications for air quality and health across the whole of Southeast Asia (Kunii et al., 2002). The forest fires of Indonesia are unique in that the continuous burning of peat, for a season or more (Page et al., 2002), produces high aerosol concentrations, leading to significant reduction of visibility (Wang et al., 2004). The seeds for these fires are anthropogenic activities (related to forest degradation and clearance activities) whose impact is magnified by the droughts (Wooster et al., 2012). Since the 1980s, the rapid and dramatic change of land use and population density over Indonesia has further exacerbated this problem (Field et al., 2009). Indonesian drought and forest fires are traditionally ascribed to El Niño, but, after the discovery of IOD, it is becoming clear that the droughts over parts of Indonesia are more strongly controlled by IOD (Saji & Yamagata, 2003a). This explains the large droughts during IOD years of 1961, 1963, 1991, 1994, and 2006 (Field et al., 2009), which occurred when conditions were near normal in the Pacific.

Although IOD brings higher than normal rainfall over equatorial East Africa (Behera et al., 2005; Saji & Yamagata, 2003a), the impact is usually more disastrous than beneficial. The effect of IOD on rainfall is felt strongly during the so-called short rainfall season of October-November-December; the rainfall anomalies are strongest between 10S and 10N, and extend from the Indian Ocean coast until about 25E (see Figure 13 and Figure 16; also Conway et al., 2005). In strong IOD years, massive hydrological anomalies occur: over the 1961 to 1964 period, which experienced two strong IOD events, the cumulative river flow anomaly for the White Nile upstream of the Sudd, the Blue Nile, Atbara, Congo, Tana, and Zambezi rivers was 1,428 km3, totaling the annual flow of the Congo (Conway et al., 2005), the second largest river in the world, by discharge. After the 1997 IOD event, the levels of Lake Victoria, Lake Tanganyika, and Lake Malawi rose by approximately 1.7 m, 2.1 m, and 1.8 m, respectively (Birkett et al., 1999).

Figure 16. Composite rainfall anomalies of October–November–December averaged rainfall over equatorial Africa are shown for (top) positive IOD events and (bottom) El Niño events over the period 1958 to 2014. In Figure 16a, all events were considered. However, in Figure 16b, co-occurring events of 1963, 1972, 1976, 1982, and 1997 were removed from both the top and bottom plots. The IOD years considered in Fig. 16a are those during 1961, 1963, 1967, 1972, 1976, 1982, 1983, 1991, 1994, 1997, 2006, 2008, 2011, and 2012. The El Niño years considered are those during 1963, 1965, 1972, 1976, 1982, 1986, 1987, 1997, 2002, 2004, 2009, and 2014.

The widespread flooding and inundation of low-lying areas during positive IOD years also have major socioeconomic impacts across East Africa, due to damage to life and infrastructure (Conway, 2002; Odingo, 1962), massive displacement of people from flooded areas (Conway et al., 2005), and infectious diseases (Hashizume et al., 2012; Munyua et al., 2010; Nguku et al., 2010). A particularly severe health hazard is Rift Valley fever (RVF), a mosquito-borne viral disease primarily affecting domesticated animals, that also causes mild to life-threatening illness in humans (Baba et al., 2016). In 1997, RVF was widespread in the Horn of Africa, involved five countries, and caused the loss of ~100,000 domestic animals and ~90,000 human infections (Woods et al., 2002). The economic impact was further amplified due to bans on livestock export from the region (Little et al., 2001). In 2006, RVF returned to East Africa, causing 75,000 human infections and 350 deaths, while spreading to larger areas than in 1997. The wide-ranging effects of the disease on the livestock and other sectors induced an economic loss of over Ksh 2.1 billion ($32 million) in the Kenyan economy alone (Rich & Wanyoike, 2010), and caused overall economic losses in East Africa exceeding $60 million (Little, 2009).

Note that rainfall anomalies over equatorial East Africa are often incorrectly ascribed to ENSO (Saji & Yamagata, 2003b). The reason that ENSO appears to be associated with rainfall anomalies over equatorial East Africa is due to the co-occurrence of IOD and ENSO events. To illustrate this, and to demonstrate the importance of IOD for East African short rains, four situations are presented in Figure 16. The top and bottom panels on the left depict composite rainfall anomalies over East Africa during positive IOD and El Niño years, respectively; for the plots shown on the right, however, co-occurring years were removed before the composite was calculated, in order to reveal the true impacts of IOD and ENSO. It is clear that ENSO does not strongly affect East African rainfall variations, and this conclusion is in agreement with many modeling studies (Behera et al., 2005; Ummenhofer et al., 2009b).

Regional impacts of IOD are also moderately felt over Australia (Ashok et al., 2003; Saji & Yamagata, 2003a), where the spatial signature of IOD varies a bit with the season (Figure 13) and is mixed with that due to ENSO (Meyers et al., 2007; Risbey et al., 2009). Risbey et al. (2009) found that years with co-occurring negative IOD and La Niña increased rainfall over southeastern Australia, while the opposite occurred during years with an El Niño and a positive IOD event. Ummenhofer et al. (2009a) suggested that IOD more than ENSO was the key driver of major droughts in the region of southeastern Australia; they demonstrated that the IOD impact explained nearly all of Australia’s iconic droughts of the 20th century. Cai et al. (2009b) suggested that the relative preponderance of positive IOD events since the 1950s potentially accounted for a large proportion of the long-term decline in austral autumn southeastern Australian rainfall. Cai et al. (2009a) showed that, out of the 21 significant bushfire seasons since 1950, 11 were preceded by a positive IOD event. Yuan and Yamagata (2015) have found that wheat yield is reduced (increased) by 28.4% (12.8%) during positive (negative) IOD events. Surprisingly, the IOD influence over India, as estimated by statistical methods, appears to be weak (Saji et al., 1999). However, this is likely a consequence of IOD–ENSO interaction (Ashok et al., 2001, 2004): when a positive IOD co-occurs with an El Niño, the IOD-induced enhancement of monsoon rainfall is countered by the suppression of the same rainfall by El Niño’s atmospheric teleconnections. The same authors further showed that, during decades of weak ENSO variability, IOD is strongly correlated with Indian summer monsoon rainfall.

Future Directions

The discussion in this article demonstrates a fair understanding of the essential nature of the IOD phenomenon, its structure, and the mechanisms of its genesis. There has also been a great deal of progress made in exploring and understanding the impacts of IOD. There are a number of areas where further research can yield a richer and deeper understanding of this important phenomenon and its impacts. First, the atmospheric and oceanic dynamics controlling IOD variability need to be systematically explored at a basic level, especially with simplified models. Most of the understanding of the structure and mechanisms of IOD are arrived at through analysis of data obtained from observations or reanalysis products and from complex numerical simulations. However, these need to be connected to the laws of fluid motion and the structure of IOD needs to be explained on the basis of these laws, so that a rich theory of IOD variability can be built upon the framework of geophysical fluid dynamics. Although this can to some extent be done by adapting the rich framework of ENSO theory, IOD variability has many unique characteristics that set it apart from ENSO. A large part of this is due to the strong annual cycle of the background state upon which IOD develops, which in turn is controlled by the monsoonal annual cycle. There are several specific areas where a dedicated study of IOD dynamics can enrich the theory of coupled ocean–atmosphere interaction. The interaction of planetary and Kelvin waves with a time-varying mean state not only can deepen the understanding of the temporal aspects of IOD variability, but also can be a key to understanding the seasonal structure of teleconnections excited by IOD. The latter will also help further the understanding of the interaction between IOD and ENSO.


  • Alory, G., Wijffels, S., & Meyers, G. M. (2007). Observed temperature trends in the Indian Ocean over 1960–1999 and associated mechanisms. Geophysical Research Letters, 34(2), L02606.
  • Amri, K. (2012). Study of primary productivity on Indian Ocean Dipole Mode event and its relationships to pelagic fish catch abundance in western part of Sumatra waters. PhD dissertation, Bogor Agricultural University, Indonesia, available online at
  • Annamalai, H., Murtugudde, R., Potemra, J., Xie, S. P., Liu, P., & Wang, B. (2003). Coupled dynamics over the Indian Ocean: Spring initiation of the zonal mode. Deep Sea Research Part II: Topical Studies in Oceanography, 50(12), 2305–2330.
  • Ashok, K., Guan, Z., Saji, N. H., & Yamagata, T. (2004). Individual and combined influences of ENSO and the Indian Ocean Dipole on the Indian summer monsoon. Journal of Climate, 17(16), 3141–3155.
  • Ashok, K., Guan, Z., & Yamagata, T. (2001). Impact of the Indian Ocean Dipole on the relationship between the Indian monsoon rainfall and ENSO. Geophysical Research Letters, 28(23), 4499–4502.
  • Ashok, K., Guan, Z., & Yamagata, T. (2003). Influence of the Indian Ocean Dipole on the Australian winter rainfall. Geophysical Research Letters, 30(15), 1821.
  • Baba, M., Masiga, D. K., Sang, R., & Villinger, J. (2016). Has Rift Valley fever virus evolved with increasing severity in human populations in East Africa? Emerging Microbes & Infections, 5(6), e58.
  • Barnett, T. P. (1984). Interaction of the monsoon and Pacific trade wind system at interannual time scales. Part III: A partial anatomy of the Southern Oscillation. Monthly Weather Review, 112(12), 2388–2400.
  • Behera, S. K., Krishnan, R., & Yamagata, T. (1999). Unusual ocean-atmosphere conditions in the tropical Indian Ocean during 1994. Geophysical Research Letters, 26(19), 3001–3004.
  • Behera, S. K., Luo, J.-J., Masson, S., Delecluse, P., Gualdi, S., Navarra, A., et al. (2005). Paramount impact of the Indian Ocean Dipole on the East African short rains: A CGCM study. Journal of Climate, 18(21), 4514–4530.
  • Behera, S. K., Luo, J. J., Masson, S., Rao, S.A., Sakuma, H., & Yamagata, T. (2006). A CGCM study on the interaction between IOD and ENSO. Journal of Climate, 19, 1688–1705.
  • Birkett, C., Murtugudde, R., & Allan, T. (1999). Indian Ocean climate event brings floods to East Africa’s lakes and the Sudd Marsh. Geophysical Research Letters, 26(8), 1031–1034.
  • Bjerknes, J. (1969). Atmospheric teleconnections from the equatorial Pacific. Monthly Weather Review, 97(3), 163–172.
  • Cai, W., Cowan, T., & Raupach, M. (2009a). Positive Indian Ocean Dipole events precondition southeast Australia bushfires. Geophysical Research Letters, 36(19), L19710.
  • Cai, W., Cowan, T., & Sullivan, A. (2009b). Recent unprecedented skewness towards positive Indian Ocean Dipole occurrences and its impact on Australian rainfall. Geophysical Research Letters, 36(11), L11705.
  • Cai, W., Van Rensch, P., Cowan, T., & Hendon, H. H. (2012). An asymmetry in the IOD and ENSO teleconnection pathway and its impact on Australian climate. Journal of Climate, 25(18), 6318–6329.
  • Cai, W., Zheng, X.-T., Weller, E., Collins, M., Cowan, T., Lengaigne, M., et al. (2013). Projected response of the Indian Ocean Dipole to greenhouse warming. Nature Geoscience, 6(12), 999–1007.
  • Carton, J. A., & Giese, B. S. (2008). A reanalysis of ocean climate using Simple Ocean Data Assimilation (SODA). Monthly Weather Review, 136(8), 2999–3017.
  • Chambers, D. P., Tapley, B. D., & Stewart, R. H. (1999). Anomalous warming in the Indian Ocean coincident with El Nino. Journal of Geophysical Research: Oceans, 104(C2), 3035–3047.
  • Chen, G., Han, W., Li, Y., & Wang, D. (2016). Interannual variability of equatorial eastern Indian Ocean upwelling: Local versus remote forcing. Journal of Physical Oceanography, 46(3), 789–807.
  • Clarke, A. J. (2008). An introduction to the dynamics of El Nino and the Southern Oscillation. London: Academic Press.
  • Conway, D. (2002). Extreme rainfall events and lake level changes in East Africa: recent events and historical precedents. In The East African great lakes: limnology, palaeolimnology and biodiversity (Vol. 12, pp. 63–92). Dordrecht: Springer.
  • Conway, D., Allison, E., Felstead, R., & Goulden, M. (2005). Rainfall variability in East Africa: implications for natural resources management and livelihoods. Philosophical Transactions of the Royal Society of London A: Mathematical, Physical and Engineering Sciences, 363(1826), 49–54.
  • Craig, C. D. (2005). Flohn, Hermann. In J. E. Oliver (Ed.), Encyclopedia of world climatology (p. 379). New York: Springer Science & Business Media.
  • Crétat, J., Terray, P., Masson, S., Sooraj, K. P., & Roxy, M. K. (2017). Indian Ocean and Indian summer monsoon: relationships without ENSO in ocean–atmosphere coupled simulations. Climate Dynamics, 49(4), 1429–1448.
  • D’Arrigo, R., & Wilson, R. (2008). El Nino and Indian Ocean influences on Indonesian drought: Implications for forecasting rainfall and crop productivity. International Journal of Climatology, 28(5), 611–616.
  • Delman, A. S., Sprintall, J., McClean, J. L., & Talley, L. D. (2016). Anomalous Java cooling at the initiation of positive Indian Ocean Dipole events. Journal of Geophysical Research: Oceans, 121(8), 5805–5824.
  • Feng, M., & Meyers, G. M. (2003). Interannual variability in the tropical Indian Ocean: A two-year time-scale of Indian Ocean Dipole. Deep Sea Research Part II: Topical Studies in Oceanography, 50(12), 2263–2284.
  • Feng, M., Meyers, G. M., & Wijffels, S. (2001). Interannual upper ocean variability in the tropical Indian Ocean. Geophysical Research Letters, 28(21), 4151–4154.
  • Field, R. D., van der Werf, G. R., & Shen, S. S. P. (2009). Human amplification of drought-induced biomass burning in Indonesia since 1960. Nature Geoscience, 2(3), 185–188.
  • Flohn, H. (1987). East African rains of 1961/62 and the abrupt change of the White Nile discharge. Paleoecology of Africa, 18, 3–18.
  • Ghofar, A. (2005). Co-existence in small-pelagic fish resources of the south coast of East Java, Straits of Bali, Alas and Sape-Indonesia. ILMU KELAUTAN: Indonesian Journal of Marine Sciences, 10(3), 149–157.
  • Gill, A. E. (1983). An estimation of sea-level and surface-current anomalies during the 1972 El Nino and consequent thermal effects. Journal of Physical Oceanography, 13, 586–606.
  • Gill, A. E. (1982). Atmosphereocean dynamics. New York: Academic Press.
  • Gill, A. E. (1980). Some simple solutions for heat-induced tropical circulation. Quarterly Journal of the Royal Meteorological Society, 106(449), 447–462.
  • Goswami, B. N., Krishnamurthy, V., & Annamalai, H. (1999). A broad-scale circulation index for the interannual variability of the Indian summer monsoon. Quarterly Journal of the Royal Meteorological Society, 125(554), 611–633.
  • Grove, A. T. (1996). African river discharges and lake levels in the twentieth century. The limnology, climatology and paleoclimatology of the East African lakes (pp. 95–102). The Netherlands: Gordon and Breach.
  • Gualdi, S., Guilyardi, E., Navarra, A., Masina, S., & Delecluse, P. (2003). The interannual variability in the tropical Indian Ocean as simulated by a CGCM. Climate Dynamics, 20(6), 567–582.
  • Hameed, S. N., Jin, D., & Thilakan, V. (2018). A model for super El Niños. Nature Communications, 9(1), 2528.
  • Han, W., Vialard, J., McPhaden, M.J., Lee, T., Masumoto, Y., Feng, M., et al. (2014). Indian Ocean decadal variability: A review. Bulletin of the American Meteorological Society, 95(11), 1679–1703.
  • Hanamoto, E. (1987). Effect of oceanographic environment on Bigeye tuna distribution. Bulletin of Japanese Society of Fisheries Oceanography, 51(3), 203–216.
  • Hashizume, M., Chaves, L. F., & Minakawa, N. (2012). Indian Ocean Dipole drives malaria resurgence in East African highlands. Scientific Reports, 2, 269.
  • Hirst, A. C., & Godfrey, J. S. (1994). The response to a sudden change in Indonesian throughflow in a global ocean GCM. Journal of Physical Oceanography, 24(9), 1895–1910.
  • Hisard, P. (1980). Observation de réponses de types El Niño dans l’Atlantique tropical oriental Golfe de Guinée. Oceanologica Acta, 3(1), 69–78.
  • Hooker, S. B., Firestone, E. R., Esaias, W. E., Feldman, G. C., Gregg, W. W., & Mcclain, C. R. (1992). SeaWiFS technical report series (Vol. 1). An overview of SeaWiFS and ocean color. Greenbelt, Maryland: Goddard Space Flight Center. Retrieved from
  • Hu, S., & Sprintall, J. (2016). Interannual variability of the Indonesian throughflow: The salinity effect. Journal of Geophysical Research: Oceans, 121(4), 2596–2615.
  • Kalnay, E., Kanamitsu, M., Kistler, R., Collins, W., Deaven, D., Gandin, L., et al. (1996). The NCEP/NCAR 40-Year Reanalysis Project. Bulletin of the American Meteorological Society, 77(3), 437–471.
  • Kite, G. W. (1981). Récents changements enregistrés dans le niveau du Lac Victoria. Hydrological Sciences Journal, 26(3), 233–243.
  • Kunii, O., Kanagawa, S., Yajima, I., Hisamatsu, Y., Yamamura, S., Amagai, T., et al. (2002). The 1997 haze disaster in Indonesia: its air quality and health effects. Archives of Environmental Health: An International Journal, 57(1), 16–22.
  • Lamb, H. H. (1966). Climate in the 1960’s changes in the world’s wind circulation reflected in prevailing temperatures, rainfall patterns and the levels of the African lakes. The Geographical Journal, 132(2), 183–212.
  • Lan, K.-W., Evans, K., & Lee, M.-A. (2013). Effects of climate variability on the distribution and fishing conditions of Yellowfin tuna (Thunnus albacares) in the western Indian Ocean. Climatic Change, 119(1), 63–77.
  • Lau, N.-C., & Nath, M. J. (1996). The role of the “atmospheric bridge” in linking tropical Pacific ENSO events to extratropical SST anomalies. Journal of Climate, 9(9), 2036–2057.
  • Le Traon, P. Y., Nadal, F., & Ducet, N. (1998). An improved mapping method of multisatellite altimeter data. Journal of Atmospheric and Oceanic Technology, 15(2), 522–534.
  • Liebmann, B., & Smith, C. A. (1996). Description of a complete (interpolated) outgoing longwave radiation dataset. Bulletin of the American Meteorological Society, 77, 1275–1277.
  • Little, P. D. (2009, February). Hidden value on the hoof: cross-border livestock trade in Eastern Africa. Policy Brief, 2. Available from
  • Little, P. D., Teka, T., & Azeze, A. (2001). Cross-border livestock trade and food security in the Horn of Africa: an overview. Washington, DC: USAID/REDSO.
  • Liu, Q.-Y., Feng, M., Wang, D., & Wijffels, S. (2015). Interannual variability of the Indonesian throughflow transport: A revisit based on 30 year expendable bathythermograph data. Journal of Geophysical Research: Oceans, 120(12), 8270–8282.
  • Lumban-Gaol, J., Leben, R. R., Vignudelli, S., Mahapatra, K., Okada, Y., Nababan, B., et al. (2015). Variability of satellite-derived sea surface height anomaly, and its relationship with Bigeye tuna (Thunnus obesus) catch in the Eastern Indian Ocean. European Journal of Remote Sensing, 48, 465–477.
  • Masumoto, Y., & Yamagata, T. (1996). Seasonal variations of the Indonesian throughflow in a general ocean circulation model. Journal of Geophysical Research: Oceans, 101(C5), 12287–12293.
  • Matsuura, H., Sugimoto, T., Nakai, M., & Tsuji, S. (1997). Oceanographic conditions near the spawning ground of southern bluefin tuna; northeastern Indian Ocean. Journal of Oceanography, 53, 421–434.
  • McCreary, J. P. (1976). Eastern tropical ocean response to changing wind systems: with application to El Niño. Journal of Physical Oceanography, 6(5), 632–645.
  • McCreary, J. P., & Anderson, D. L. T. (1985). Simple models of El Nino and the Southern Oscillation. Elsevier Oceanography Series, 40, 345–370.
  • Meyers, G. M. (1996). Variation of Indonesian throughflow and the El Niño-Southern Oscillation. Journal of Geophysical Research: Oceans, 101(C5), 12255–12263.
  • Meyers, G. M., McIntosh, P., Pigot, L., & Pook, M. (2007). The years of El Niño, La Niña, and interactions with the tropical Indian Ocean. Journal of Climate, 20(13), 2872–2880.
  • Munyua, P., Murithi, R. M., Wainwright, S., Githinji, J., Hightower, A., Mutonga, D., et al. (2010). Rift Valley Fever outbreak in livestock in Kenya, 2006–2007. American Journal of Tropical Medicine and Hygiene, 83(2 Suppl.), 58–64.
  • Murtugudde, R., McCreary, J. P., & Busalacchi, A. J. (2000). Oceanic processes associated with anomalous events in the Indian Ocean with relevance to 1997–1998. Journal of Geophysical Research: Oceans, 105(C2), 3295–3306.
  • Ng, B., & Cai, W. (2016). Present-day zonal wind influences projected Indian Ocean Dipole skewness. Geophysical Research Letters, 43(21), 11,392–11,399.
  • Nguku, P. M., Sharif, S. K., Mutonga, D., Amwayi, S., Omolo, J., Mohammed, O., et al. (2010). An investigation of a major outbreak of Rift Valley fever in Kenya: 2006–2007. The American Journal of Tropical Medicine and Hygiene, 83(2 Suppl.), 5–13.
  • Odingo, R. S. (1962). Abnormal and unseasonal rains in East Africa. Geographical Review, 52(3), 440–442.
  • Page, S. E., Siegert, F., Rieley, J. O., Boehm, H.-D. V., Jaya, A., & Limin, S. (2002). The amount of carbon released from peat and forest fires in Indonesia during 1997. Nature, 420(6911), 61–65.
  • Pant, G. B., & Parthasarathy, S. B. (1981). Some aspects of an association between the Southern Oscillation and Indian summer monsoon. Archives for Meteorology, Geophysics, and Bioclimatology, Series B, 29(3), 245–252.
  • Philander, S. G. H. (1990). El Nino, La Nina, and the Southern Oscillation. San Diego, CA: Academic Press.
  • Picaut, J., Masia, F., & Du Penhoat, Y. (1997). An advective-reflective conceptual model for the oscillatory nature of the ENSO. Science, 277(5326), 663–666.
  • Potemra, J. T., & Lukas, R. (1999). Seasonal to interannual modes of sea level variability in the western Pacific and eastern Indian Oceans. Geophysical Research Letters, 26(3), 365–368.
  • Qu, T., Du, Y., Strachan, J., Meyers, G., & Slingo, J. (2005). Sea surface temperature and its variability. Oceanography, 18(4), 50.
  • Rao, S. A., & Behera, S. K. (2005). Subsurface influence on SST in the tropical Indian Ocean: Structure and interannual variability. Dynamics of Atmospheres and Oceans, 39(1), 103–135.
  • Rao, S. A., Behera, S. K., Masumoto, Y., & Yamagata, T. (2002). Interannual subsurface variability in the tropical Indian Ocean with a special emphasis on the Indian Ocean Dipole. Deep Sea Research Part II: Topical Studies in Oceanography, 49(7), 1549–1572.
  • Reppin, J., Schott, F. A., Fischer, J., & Quadfasel, D. (1999). Equatorial currents and transports in the upper central Indian Ocean: Annual cycle and interannual variability. Journal of Geophysical Research: Oceans, 104(C7), 15495–15514.
  • Reverdin, G., Cadet, D. L., & Gutzler, D. (1986). Interannual displacements of convection and surface circulation over the equatorial Indian Ocean. Quarterly Journal of the Royal Meteorological Society, 112, 43–67.
  • Reynolds, R. W., Rayner, N. A., Smith, T. M., Stokes, D. C., & Wang, W. (2002). An improved in situ and satellite SST analysis for climate. Journal of Climate, 15(13), 1609–1625.
  • Rich, K. M., & Wanyoike, F. (2010). An assessment of the regional and national socio-economic impacts of the 2007 Rift Valley fever outbreak in Kenya. The American Journal of Tropical Medicine and Hygiene, 83(2 Suppl.), 52–57.
  • Ridgway, K. R., Dunn, J. R., & Wilkin, J. L. (2002). Ocean interpolation by four-dimensional weighted least squares—Application to the waters around Australasia. Journal of Atmospheric and Oceanic Technology, 19(9), 1357–1375.
  • Risbey, J. S., Pook, M. J., McIntosh, P. C., Wheeler, M. C., & Hendon, H. H. (2009). On the remote drivers of rainfall variability in Australia. Monthly Weather Review, 137(10), 3233–3253.
  • Saji, N. H., Goswami, B. N., Vinayachandran, P. N., & Yamagata, T. (1999). A dipole mode in the tropical Indian Ocean. Nature, 401, 360–363.
  • Saji, N. H., & Yamagata, T. (2003a). Possible impacts of Indian Ocean Dipole Mode events on global climate. Climate Research, 25(2), 151–169.
  • Saji, N. H., & Yamagata, T. (2003b). Structure of SST and surface wind variability during Indian Ocean Dipole Mode events: COADS observations. Journal of Climate, 16(16), 2735–2751.
  • Schiller, A., Godfrey, J. S., McIntosh, P. C., Meyers, G. M., & Wijffels, S. E. (1998). Seasonal near-surface dynamics and thermodynamics of the Indian Ocean and Indonesian throughflow in a global ocean general circulation model. Journal of Physical Oceanography, 28(11), 2288–2312.
  • Schneider, N. (1998). The Indonesian throughflow and the global climate system. Journal of Climate, 11(4), 676–689.
  • Sprintall, J., & Révelard, A. (2014). The Indonesian throughflow response to Indo-Pacific climate variability. Journal of Geophysical Research: Oceans, 119(2), 1161–1175.
  • Sprintall, J., Wijffels, S. E., Molcard, R., & Jaya, I. (2009). Direct estimates of the Indonesian throughflow entering the Indian Ocean: 2004–2006. Journal of Geophysical Research: Oceans, 114(C7), C07001.
  • Syamsuddin, M. L., Saitoh, S.-I., Hirawake, T., Bachri, S., & Harto, A. B. (2013). Effects of El Niño–Southern Oscillation events on catches of Bigeye tuna (Thunnus obesus) in the eastern Indian Ocean off Java. Fishery Bulletin, 111(2), 175–188.
  • Ummenhofer, C. C., Biastoch, A., & Böning, C. W. (2017). Multidecadal Indian Ocean variability linked to the Pacific and implications for preconditioning Indian Ocean Dipole events. Journal of Climate, 30(5), 1739–1751.
  • Ummenhofer, C. C., England, M. H., McIntosh, P. C., Meyers, G. M., Pook, M. J., Risbey, J. S., et al. (2009a). What causes southeast Australia’s worst droughts? Geophysical Research Letters, 36(4), L04706.
  • Ummenhofer, C. C., Sen Gupta, A., England, M. H., & Reason, C. J. C. (2009b). Contributions of Indian Ocean sea surface temperatures to enhanced East African rainfall. Journal of Climate, 22, 993–1013.
  • Varela, R., Santos, F., Gómez-Gesteira, M., Álvarez, I., Costoya, X., & Días, J. M. (2016). Influence of coastal upwelling on SST trends along the south coast of Java. PloS One, 11(9), e0162122.
  • Vinayachandran, P. N., Kurian, J., & Neema, C. P. (2007). Indian Ocean response to anomalous conditions in 2006. Geophysical Research Letters, 34(15), L15602.
  • Vinayachandran, P. N., Saji, N. H., & Yamagata, T. (1999). Response of the equatorial Indian Ocean to an unusual wind event during 1994. Geophysical Research Letters, 26(11), 1613–1616.
  • Wajsowicz, R. C., & Schneider, E. K. (2001). The Indonesian throughflow’s effect on global climate determined from the COLA coupled climate system. Journal of Climate, 14(13), 3029–3042.
  • Wallace, J. M., Rasmusson, E. M., Mitchell, T. P., Kousky, V. E., Sarachik, E. S., & Storch, H. von. (1998). On the structure and evolution of ENSO-related climate variability in the tropical Pacific: Lessons from TOGA. Journal of Geophysical Research: Oceans, 103(C7), 14241–14259.
  • Wang, H., Murtugudde, R., & Kumar, A. (2016). Evolution of Indian Ocean dipole and its forcing mechanisms in the absence of ENSO. Climate Dynamics, 47(7–8), 2481–2500.
  • Wang, Y., Field, R. D., & Roswintiarti, O. (2004). Trends in atmospheric haze induced by peat fires in Sumatra Island, Indonesia and El Niño phenomenon from 1973 to 2003. Geophysical Research Letters, 31(4), L04103.
  • Watanabe, M., & Kimoto, M. (2000). Atmosphere-ocean thermal coupling in the North Atlantic: A positive feedback. Quarterly Journal of the Royal Meteorological Society, 126(570), 3343–3369.
  • Weare, B. C. (1979). A statistical study of the relationships between ocean surface temperatures and the Indian monsoon. Journal of the Atmospheric Sciences, 36(12), 2279–2291.
  • Webster, P. J., Moore, A. M., Loschnigg, J. P., & Leben, R. R. (1999). Coupled ocean–atmosphere dynamics in the Indian Ocean during 1997–98. Nature, 401(6751), 356–360.
  • Wentz, F. J., Gentemann, C., Smith, D., & Chelton, D. (2000). Satellite measurements of sea surface temperature through clouds. Science, 288(5467), 847–850.
  • Wijffels, S. E., & Meyers, G. M. (2004). An intersection of oceanic wave guides: Variability in the Indonesian throughflow region. Journal of Physical Oceanography, 34, 1232–1253.
  • Woodruff, S. D., Worley, S. J., Lubker, S. J., Ji, Z., Freeman, J. E., Berry, D. I., et al. (2011). ICOADS release 2.5: Extensions and enhancements to the surface marine meteorological archive. International Journal of Climatology, 31, 951–967.
  • Woods, C. W., Karpati, A. M., Grein, T., McCarthy, N., Gaturuku, P., Muchiri, E., et al. (2002). An outbreak of Rift Valley fever in northeastern Kenya, 1997–98. Emerging Infectious Diseases, 8(2), 138–144.
  • Wooster, M. J., Perry, G. L. W., & Zoumas, A. (2012). Fire, drought and El Niño relationships on Borneo (Southeast Asia) in the pre-MODIS era (1980–2000). Biogeosciences, 9(1), 317–340.
  • Wyrtki, K. (1962). The upwelling in the region between Java and Australia during the south-east monsoon. Marine and Freshwater Research, 13(3), 217–225.
  • Wyrtki, K. (1987). Indonesian through flow and the associated pressure gradient. Journal of Geophysical Research: Oceans, 92(C12), 12941–12946.
  • Xie, P., & Arkin, P. A. (1997). Global precipitation: A 17-year monthly analysis based on gauge observations, satellite estimates, and numerical model outputs. Bulletin of the American Meteorological Society, 78(11), 2539–2558.
  • Xie, S. P., Annamalai, H., Schott, F. A., & McCreary, J. P. (2002). Structure and mechanisms of South Indian Ocean climate variability. Journal of Climate, 15(8), 864–878.
  • Yamagata, T. (1985). Stability of a simple air-sea coupled model in the tropics. Elsevier Oceanography Series, 40, 637–657.
  • Yamagata, T., Behera, S. K., Luo, J.-J., Masson, S., Jury, M. R., & Rao, S.A. (2004). Coupled ocean–atmosphere variability in the tropical Indian Ocean. In Earth’s Climate: The Ocean-Atmosphere Interaction (pp. 189–211). Washington, DC: American Geophysical Union.
  • Yang, Y., Xie, S.P., Wu, L. X., Kosaka, Y., Lau, N.C., & Vecchi, G.A. (2015). Seasonality and predictability of the Indian Ocean Dipole Mode: ENSO forcing and internal variability. Journal of Climate, 28, 8021–8036.
  • Yu, L., & Rienecker, M. M. (1999). Mechanisms for the Indian Ocean warming during the 1997–98 El Nino. Geophysical Research Letters, 26(6), 735–738.
  • Yuan, C., & Yamagata, T. (2015). Impacts of IOD, ENSO and ENSO Modoki on the Australian winter wheat yields in recent decades. Scientific Reports, 5, 17252 EP.
  • Yulaeva, E., & Wallace, J. M. (1994). The signature of ENSO in global temperature and precipitation fields derived from the microwave sounding unit. Journal of Climate, 7(11), 1719–1736.
  • Zebiak, S. E. (1993). Air–sea interaction in the equatorial Atlantic region. Journal of Climate, 6(8), 1567–1586.
  • Ziese, M., Becker, A., Finger, P., Meyer-Christoffer, A., Rudolf, B., & Schneider, U. (2011). GPCC First Guess Product at 1.0: Near real-time first guess monthly land-surface precipitation from rain-gauges based on SYNOP data.
  • Zubair, L., Rao, S. A., & Yamagata, T. (2003). Modulation of Sri Lankan Maha rainfall by the Indian Ocean Dipole. Geophysical Research Letters, 30(2).