II. Interannual variability and ENSO-related anomalies

Interannual variations are anomalies of the annual or seasonal cycles which can reach extremes some years or may disappear. Such anomalies can last the whole year through or more. In order to fully understand how such anomalies can happen, it is essential to study the air-sea interactions and their influence on the atmospheric and oceanic mean circulations. We have seen in the first part of this document how winds can induce the upper layer of the ocean, particularly in the Indian Ocean where the monsoon regime is strong and well defined. For the study of interannual variations, meteorological parameters such as air-ocean heat-budget, precipitation, evaporation, sea-level pressure (SLP) and even tropospheric circulation, have to be taken into account in the mechanisms directly generating such variations.

Some regions seem to be more inclined to experience interannual variability, probably because they are subject to strong air-sea interactions. These regions have been studied in a previous paper (Le Blanc, 1996) through the analysis of SLP, Ekman pumping and SST variations. Since the early ninetees climatologists have tried to understand the coupling of the Indian Monsson with the Southern Oscillation in order to predict the monsoon rainfall over Asia. Webster and Yang (1992) review the historical background of Monsoon and ENSO coupling. They explain that " early studies identified the mosoon as a regional physical entity and naturally, attempt to undersatnd its structure and variability focussed on local effects. [...] In recent decades, with the advent of more homogeneous data set from satellites and a more conventional data base, there has been little to counteract this global view of the monsoon. A number of questions remain, however, such as : what is the role of the monsoon in the global climate and does the variability in the monsoon lag, lead or occur simultaneously with the interannual variability of other global scale phenomena ? "

II.1. The Southern Oscillation and the Quasi-Biennial Oscillation

(i) The Southern Oscillation (SO)

The earliest work on the Southern Oscillation (SO) was detailed by Sir Gilbert Walker (1923) who defined the SO as follows : " when the pressure is high in the Pacific Ocean, it tends to be low in the Indian Ocean from Africa to Antartica. ". In a following paper he also described the SO as "  ...a swaying of pressure on a big scale backwards and forwards between the Pacific Ocean and the Indian Ocean...  ". Walker was interested in the development of a seasonal forecast scheme for the prediction of the strength of the Indian monsoon (Mock, 1981). Yet, the ability to predict the future course of the Southern Oscillation was severely hampered by a vagueness in the observational relationships and by the failure to identify underlying physical processes which might have allowed the relationships between the Indian monsoon and the SO to become unravelled (Webster and Yang, 1992).

The SO is well-known as a standing oscillation of the surface pressure between the Indian Ocean through Indonesia and the eastern south Pacific with a changing periodicity of 40 months.

Cadet (1984) studied the SO signal at the surface level over the Indian Ocean during the period 1954-1976. Time series of wind, pressure, SST, air temperature and cloud cover over different areas show the evidence of the SO over the Indian Ocean. Power spectral analysis of the time series demonstrate the existence of a major periodicity at 40 months.

The results also suggest an eastward propagation of zonal wind anomalies over the equatorial Indian Ocean and are in agreement with the results of Barnett (1983) who used a completely different technique to study the coupling between the wind systems in the Pacific and Indian Oceans. Since by definition, standing waves are not propagating, this global-scale east-west circulation is associated with the Quasi-Biennial Oscillation.

(ii) the Quasi-Biennial Oscillation (QBO)

Eastward propagation was also noticed by Yasunari (1985) who used zonal wind and velocity potential fields and found the circulation to be on the time-scale of the SO (40-60 months period). The eastward propagation is most prominent over the Indian Ocean toward the eastern Pacific.

Many studies show a predominant quasi-biennial periodicity in the spectral analysis of atmospheric field parameters over the Indian Ocean, such as pressure or winds (Barnett, 1983, 1984a, 1984b, 1985 ; Krishnamurty et al., 1985), and in oceanic parameters of the upper-layer (Yasunari, 1991 ; Tourre and White, 1995). Tourre and White (1995) found " that in the Pacific and Indian oceans, the two dominant EOF modes in SST and heat-content are associated with the El Niño/Southern Oscillation (ENSO) signal. [...] The ENSO signal in the Indian Ocean explains as much of the interannual variability in that ocean as the ENSO signal does in the Pacific. [...] For each variable in each ocean, the dominant mode is in phase with the peak phase in El Niño off the west coast of South America, while the second mode or the precursor mode leads the first by at least 6 months. [...] The appearance of the ENSO signal in two EOF modes in the Indian Ocean, one leading the other, also suggests an eastward propagation along the equator (and a south eastward propagation in the eastern Indian Ocean) similar to its better known counterpart in the Pacific Ocean.

However, it is an important but difficult problem to determine whether the SO and the QBO are dynamically similar phenomena in the troposphere. Yasunari (1985) notes that some studies suggested a different spatial structure of the QBO to that of the SO, while other studies treated this mode as part of the SO in a braod sense. This last author finally treats these two modes as independant of each other, in reference to the fact that the SO is directly coupled to El Niño events in the equatorial Pacific Ocean but the QBO is not.

Thus, the coupling or decoupling of the QBO and the SO is not yet fully understood. Mason and Tyson (1992) suggests that the phase of the QBO might act as a ''pacemaker'' of interannual variability ; passively transmitting global signals such as the El Niño.

II.2. El Niño in the Indian Ocean

links to El Niño
Before studying the possible role of the Indian Ocean in triggering global atmospheric anomalies in the SO cycle, leading to the El Niño phenomenom, the ENSO situation within the Indian Ocean will be described. Empirical Orthogonal Function (EOF) studies of SLP, zonal wind stress, SST and Ekman pumping, show that during ENSO years, the SLP is high in the Indian Ocean when it is low in the Pacific. Zonal winds are directed to the west because of the high pressure over the Indonesian continent and the Indo-Pacific passage. The wind signal slightly leads the SST which is higher than average. This pattern is particularly true for the equatorial and south-tropical basin. It is less defined in the northern basins of the Arabian Sea and the Bay of Bengal where the seasonal cycle is much stronger due to the monsoon regime.

II.2.1. Interannual variability of the Indonesian throughflow

(i) ENSO-related variations

There are two theories concerning the interannual variations of the throughflow. The first one admitting that there are variations of sea level at both ends of the throughflow, but that the difference between these levels remain almost unchanged (Wyrtki, 1987). The sea level were measured at Davao (Philippines) and Darwin (Australia). Low sea level are observed at both stations during El Niño. Wyrtki explains that the difference between the sea level of both stations - which is supposed to be the driving force of the thransport in the throughflow - is not related to El Niño because of the behavior of the wind field over the Indonesian region. According to Barnett (1983), the convergence of surface winds over Indonesia is subject to strong interannual variations in its intensity and location because of the coupling of the trade winds over the Pacific and the monsoons over the Indian Ocean. Strong convergence of the wind field, is associated with a high state of the Southern Oscillation and sea level rises both in the eastern Indian Ocean and in the western Pacific. During El Niño, winds are divergent over Indonesia, and sea level drops on both sides of the Indonesian archipelago. These characteristics of the wind field imply that the difference of sea level between the western Pacific and the eastern Indian Ocean is only weakly affected by the principal wind patterns associated with ENSO. Wyrtki attributes the slow interannual variations of the sea level difference between Davao and Darwin to flluctuations in the wind field which are not associated with the Southern Oscillation, or may be due to uncertainties in the sea level records themselves.

However, according to the theory of Clarke and Liu (1994), the volume transport of throughflow is expected to vary during ENSO, with larger than normal transport during la Niña, when strong easterlies along the equatorial Pacific build up high sea level in the western Pacific. In the theory, the Pacific sea level is transmitted to the northwestern coast of Australia and influences throughflow by geostrophy. Meyers (1996) show similar results and concludes that Wyrtki did not use a sea level difference that is representative of the relevant pressure gradient due to his choice of Darwinto representing the Indian Ocean. Clarke and Liu (1994) show that Darwin is, in fact, representative of the Pacific Ocean during ENSO cycle. The difference between Davao and the southern coast of Java is a more appropriate gradient to measure the Pacific to Indian Ocean pressure gradient.

Moreover, Meyers (1996) showed that correlation to zonal wind anomalies are associated with thermocline variations along the Java coast : easterly wind anomalies over the equatorial Indian Ocean are associated with shallow thermocline and colder waters.

(ii) the role of the throughflow in interannual variations associated with ENSO

In their study, Tourre and White (1995) conclude that " wether the El Niño/Southern Oscillation (ENSO) in the Indian Ocean has an influence upon the ENSO signal in the Pacific Ocean remains to be tested [...]. The global EOF mode displays both SST and heat content loadings in the eastern Indian Ocean to be in phase with those in the western Pacific Ocean. This suggests that the two domains may be dependent upon each other. Further research is needed before this can be confirmed, not the least is to establish whether interannual Kelvin waves can negociate the multitude of islands that separate the two oceans. "

Winds along the east coast of Africa and along the west coast of Indonesia may generate significant upwelling and downwelling extremes that can propagate into the interior via Rossby waves and Kelvin waves, exerting a significant influence upon the ocean-atmosphere interaction there.

Observation seem to show that the 1982-83 El Niño might have been generated by such an anomaly. EOF analysis of various parameters such as SLP, zonal winds and SST, show that the signal in the Indian Ocean slightly precedes the one in the SOI (Le Blanc, 1996). This suggests that this ENSO might have been triggered by an anomaly in the wind field during the spring-transition period - April/May, when the coupled ocean-atmosphere system is the least stable and robust (Webster and Yang, 1992) and thus more susceptible to outside influences - which propagated through an equatorial Kelvin wave and reached the Indonesian coast much quicker than it would do through advection.

Numerical results obtained with a shallow-water model forced by winds show that a Kelvin wave reaching the Indonesian coastline would be coastaly trapped and most of its energy would propagate southward towards the interior of the throughflow. Such a downwelling Kelvin wave, reached the Indonesian coast in April 1982 (personal results). Wether such a wave may disbalance the net throughflow transport by rising the sea level on the Indian side of it is a question that remains. Since pressure gradient is the driving force of throughflow transport, it might be useful to study the effects of a sea level variation on the western Indonesian coasts on the throughflow currents.

(iii) the throughflow in General Circulation Models

The role of the Indonesian throughflow in interannual variations has already been investigated by Hirst and Godfrey (1993, 1994) who studied the response to a sudden change in Indonesian throughflow in a global ocean GCM and focused their work on the mechanisms underlying the subsurface temperature reponse to changes in the Indonesian throughflow. Their work attempts to provide a more complete picture of mechanisms of remote surface response, since they previously showed that the surface response is crucially dependent on the pattern of the subsurface temperature response.

(iv) SST variations due to the Indonesian throughflow

Hirst and Godfrey (1994) showed that regions of convective mixing or strong upwelling may be identified as " sensitive regions " that are liable to show a relatively large surface response to any remote change in surface forcing or ocean model geometry such as changes in the throughflow, capable of generating subsurface baroclinic adjustments that are able to propagate to the particular sensitive region. Thus their GCM model's baroclinic wave properties, and the spectrum of baroclinic modes excited by the throughflow changes, appear very important to the pattern and timing of the subsurface (and hence surface) temperature response. The baroclinic component affects surface heat flux strongly in the Leeuwin Current region but relatively weakly in the Agulhas Current and Tasman Sea.

Similar results have been observed by Morrow and Birol (1997) through the analysis of altimetry data. They find a clear interannual signal during their 3-year period 1995-1997 and note that if this signal of reduced dynamic height is related to fewer warm-core baroclinic eddies, this would have consequences in heat content in the upper ocean in the interior, with subsequent changes in SST and the surface wind field.

Thus, while the correlation between the signals of various parameters in the Pacific and Indian Oceans, analysed through different methods, is unanimously admitted, the mechanisms underlying the coupling of both oceans are still unknown wether atmospheric or oceanic. Possible global processes will be presented in the next chapter.

II.3. Monsoon and ENSO

links to Asian Monsoon
Normand (1953) noted that the " Indian Monsoon stands out as an active, not a passive, feature in world weather, more efficient as a broadcasting tool than an event to be forecast.... ". This idea has been suggested by several other authors but it has not yet been proved.

The climatological oupling between the Pacific and the Indian Oceans can be regarded as a coupling between the seasonal monsoon regime and the interannual Southern Oscillation. Therefore, conceptually, any anomaly in either cycle may influence the other and lead to strong variations in the annual cycle such as EL Niño in the eastern Pacific or similar events in the Indian Ocean. This concept allows the Indian Ocean to be not only passive affected by ENSO but also to actively influence the Southern Oscillation.

II.3.1. The forces in action

(i) Barnett's theory

" The SO, El Niño, and climatic variations in the Monsoon system are all part of one global scale phenomenon. " stipulates Barnett (1984b). " The new information is that this phenomenon appears in the SLP field to have a strong propagating component that appears first in the northern Indian Ocean and moves eastward into the eastern Pacific. Similar propagation of information was found in the surface wind field and equatorial precipitation regimes. These same conditions were amply demonstrated during the 1982-83 event and so it may be concluded that the evolution of that event bears many similarities to those in the historical record studies... In the surface wind field of the equatorial wave guide, the large-scale signal appears to take the form of a forced Kelvin wave. The mechanism that drives this wave appears to be altent heat release associated with precipitation anomalies that are phase-locked to and propagate with the surface wind anomalies. The long time scales associated with the atmospheric anomalies may be associated either with the slow interaction between the Walker and Hadley cells or with ocean-atmosphere coupling. "

Barnett et al. (1989, 1991) also showed that snow cover on the Himalayas was associated with variations in the regional and global climate. Increased snow cover over Eurasia leads to a subsequent reduction in precipitation over Southeast Asia in the spring and early summer, reduction of wind stress over the Indian Ocean, and other elements associated with a " poor " monsoon. It appears that snow-induced monsoon perturbations such as those of " poor " monsoon, may be one of the (multiple) triggers that can initiate an ENSO cycle.

(ii) Meehl's theory

The mechanism proposed by Meehl is a coupled air-sea mechanism in the tropical Indian and Pacific regions. Of critical importance was the interaction of the annual cycle with the large-scale east-west atmospheric circulation. Meehl (1987) used outgoing longwave radiation (OLR), clouds precipitation and SLP, from satellite and station data in the tropical Indian and Pacific sectors to study the annual cycle and its interannual fluctuations. Indian monsoon rainfall is chosen as an indicator of precipitation and convection in the summer monsoon region. " Examination of SLP, precipitation, and SST shows the dynamically coupled ocean-atmosphere system in the Indian-Pacific region to be to be involved with producing SO and QBO-type signals in atmosphere and ocean, with extremes in the system being manifested as Warm and Cold Events. "

In a later paper, Meehl (1993) studies the possible role of upper-ocean heat content in the biennial mechanism. Analyses of composite vertical temperature profiles from hydrographic station data for various near-equatorial areas in the Indian and Pacific oceans show that variations in the ocean heat content depend on the depth of the thermocline in the warm-pool region (both eastern Indian and western Pacific) and temperatures in the upper-ocean mixed layer away from the warm-pool (western Indian and eastern Pacific). The variations in thermocline depth in the Pacific are similar to those for ENSO events and are present in the biennial mode as well. Apparently, similar sets of mechanisms operate on both ENSO and biennial time scales. These results suggest that changes in the upper-ocean heat content contribute to the persistence of SST anomalies important to the biennial mechanism, that both the Indian and the Pacific are both involved in ENSO, and that ENSO could be an amplification of the biennial cycle - a Warm Event and La Niña a Cold Event.

Meehl joins Barnett in the sense that he shows persistent SST anomalies, associated with wind anomalies on the time scale of one seasonal cycle, influence the air-sea interaction, namely evaporation and thus latent heat release. For instance, strong winds and warm SST will lead to greater evaporation, strong convection and low SLP - i.e. strong monsoon characteristics - followed the next year by weak winds, less evaporation and high SLP - i.e. " poor " monsoon characteristics (Meehl, 1994). Of crucial importance for the monsoon biennial mechanism are the combination of convective heating anomalies over Africa and the western Pacific that has the consequence of involving alterations of the midlatitude circulation over Asia such that the subsequent monsoon development is affected. This feature has also been observed by Webster and Yang (1992) : " ...at the height of the monsoon, the largest radiative heating gradient is between Asia and north Africa. "

Further, it is noted that earlier studies of the tropospheric biennial mechanism (TBO) involving atmosphere-ocean coupled processes, showed a connection with the south Asian or Indian monsoon, but it was unclear wether the monsoon was an active or passive participant. A biennial mechanism involving coupled atmosphere-land-ocean interactions is postulated that involves the monsoon as an active participant in the TBO. (Meehl, 1996). The TBO is not perfeclty biennial and only appears to operate intermittently or in subset of years.

It is worth noting that the northern spring season is important for a number of reasons. This is the beginning of the time of transition for SST anomalies in the tropical Pacific. This transition appears to be necessary for the establishment of convective heating anomalies that could help maintain the midlatitude circulation pattern over Asia and thus contribute to subsequent monsoon development. Perhaps this is why northern spring season is problematic for forecasting many features in the tropical Indian-Pacific region and is sometimes referred to as the " spring predictability barrier ". The mechanisms of this transition involve ocean dynamics and coupled air-sea interaction in the tropical Pacific as well as features of the large-scale east-west circulation in the atmosphere involving the TBO (Meehl, 1996).

(iii) Yasunari's theory

As mentioned previously, Yasunari (1985) also observed east-west circulation in the tropics associated with the SO. He concludes that the dominant role of the oceanic Kelvin waves to the west of the maximum convection described by Philander et al. (1984), may at least in part explain this eastward propagation. In a later paper (1990) Yasunari concludes that " the variation of heat content anomaly in the western tropical Pacific during winter and spring, which seems to be modulated by the preceding Asian monsoon activity via anomalous surface wind field, may provide a necessary or at least a very favorable dynamical condition for triggering ENSO events in the eastern Pacific. "

II.3.2. Discussion and Conclusion

Although the work and results of several authors has been summarysed above, there theories are not in contradiction, neither very much different. Some other work has been done and could have been mentioned, but they show strong similarities with those above and do not concern the Indian Ocean just as much. Some authors tried to build up the Monsoon/ENSO " puzzle " by reviewing each piece (of work) of it in order to find out its mechanism (Kiladis, 1988, 1989 ; Webster and Yang, 1992). Although many pieces of the puzzle are now available, a comprehensive theory which explain the entire SO phenomenon still remains elusive (Kiladis, 1988) and a number of questions require answers and some are adressed by Webster and Yang (1992). These are for example :

In their paper the authors bring answers to those questions which might be summarised as follows :