2.
GROWTH OF THE TOGA PROGRAM

In the 1960s, Jacob Bjerknes was the first to link El Niño and the Southern Oscillation. He also suggested that the combined ENSO phenomenon resulted from dynamics that coupled the atmosphere and the upper ocean in the region of the equatorial Pacific. Coherent research efforts on ENSO developed during the 1970s, motivated in part by recognition of disruptive effects in the Americas from climate variations thought to be associated with ENSO. The intense ENSO warm event of 1982–1983 served as an impetus for the organized international program that became TOGA.

The TOGA Program grew out of studies of interactions between the atmosphere and ocean, especially the work of Jacob Bjerknes. Early observational, theoretical and modeling efforts to understand the dynamics of ENSO brought particular attention to the tropical Pacific region. Because of the large interannual variations in climate that seemed to result from ENSO, many scientists were willing to commit themselves to the planning process and the organized effort that became TOGA.

The scientific developments that ultimately led to the establishment of the TOGA Program and the highlights of early program plans and documents for TOGA have been presented previously (NRC 1990). Here we highlight the development of the scientific concepts and the events that shaped the plan for U.S. participation in the TOGA program, as it began in 1985.

ENSO: A COUPLED OCEAN-ATMOSPHERE PHENOMENON

The origin of the empirical and prediction studies of the TOGA program can be traced to the work of Sir Gilbert T. Walker, who assumed the post of Director General of the Observatory in India in 1904. In a region where monsoon failure and famine were disastrously coupled, Walker set about to understand and improve forecasts of the interannual variations of the Indian monsoon. Over the next three decades, Walker began to calculate correlations between the time series available to him—sea-level pressure, rain amounts, air temperature, sunspot activity, and others. In the process of performing these empirical studies, Walker established the existence of the Southern Oscillation as a global



The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 12
Learning to Predict Climate Variations Associated with El Niño and the Southern Oscillation: Accomplishments and Legacies of the TOGA Program 2. GROWTH OF THE TOGA PROGRAM In the 1960s, Jacob Bjerknes was the first to link El Niño and the Southern Oscillation. He also suggested that the combined ENSO phenomenon resulted from dynamics that coupled the atmosphere and the upper ocean in the region of the equatorial Pacific. Coherent research efforts on ENSO developed during the 1970s, motivated in part by recognition of disruptive effects in the Americas from climate variations thought to be associated with ENSO. The intense ENSO warm event of 1982–1983 served as an impetus for the organized international program that became TOGA. The TOGA Program grew out of studies of interactions between the atmosphere and ocean, especially the work of Jacob Bjerknes. Early observational, theoretical and modeling efforts to understand the dynamics of ENSO brought particular attention to the tropical Pacific region. Because of the large interannual variations in climate that seemed to result from ENSO, many scientists were willing to commit themselves to the planning process and the organized effort that became TOGA. The scientific developments that ultimately led to the establishment of the TOGA Program and the highlights of early program plans and documents for TOGA have been presented previously (NRC 1990). Here we highlight the development of the scientific concepts and the events that shaped the plan for U.S. participation in the TOGA program, as it began in 1985. ENSO: A COUPLED OCEAN-ATMOSPHERE PHENOMENON The origin of the empirical and prediction studies of the TOGA program can be traced to the work of Sir Gilbert T. Walker, who assumed the post of Director General of the Observatory in India in 1904. In a region where monsoon failure and famine were disastrously coupled, Walker set about to understand and improve forecasts of the interannual variations of the Indian monsoon. Over the next three decades, Walker began to calculate correlations between the time series available to him—sea-level pressure, rain amounts, air temperature, sunspot activity, and others. In the process of performing these empirical studies, Walker established the existence of the Southern Oscillation as a global

OCR for page 12
Learning to Predict Climate Variations Associated with El Niño and the Southern Oscillation: Accomplishments and Legacies of the TOGA Program spatial pattern of interannual climate variations with identifiable centers of action. He was able to eliminate solar variability as a major contributor to the oscillation. Walker strongly suspected that oceanic processes were responsible for the oscillation, but was unable explore his ideas with the available data and thus missed the connection to El Niño (Walker 1924, Walker and Bliss 1932). The first major breakthrough in understanding the mechanism of Walker's Southern Oscillation was to come forty years later. In the 1960s, Jacob Bjerknes began an examination of the meteorological conditions associated with El Niño, an interannually varying intrusion of warm equatorial waters along the western coast of South America. The warm intrusions had major economic impacts on the tuna fishing industry in the affected region. Bjerknes (1966) quickly recognized that El Niño was connected with large-scale fluctuations in trade-wind circulations in both the northern and southern hemispheres of the Pacific sector. He soon connected these trade-wind fluctuations to the Southern Oscillation (Bjerknes 1969). Bjerknes's most important contribution was the reasoning he used in explaining the coupling between the oceanic and atmospheric circulations. On the basis of empirical evidence, Bjerknes hypothesized that El Niño and the Southern Oscillation are the result of the coupling between the east-west atmospheric circulation in the Pacific sector and also a coupling between the current and thermal structure of the upper ocean in the eastern equatorial Pacific Ocean. He observed that when the trade winds are strong (the “normal” condition), relatively cool equatorial water extends from the South American coast to the central Pacific. Bjerknes attributed the cool waters to equatorial upwelling caused by easterly wind stress acting on the ocean surface. He reasoned that the resulting pattern of sea surface temperature reinforces the strength of the trade winds by favoring large-scale atmospheric cooling, descent, and cloud-free conditions over the equatorial eastern Pacific, accompanied by large-scale ascent with relatively large amounts of precipitation, convective clouds, and atmospheric heating over the central and western equatorial Pacific. In the equatorial region, the east-west atmospheric heating differences would be expected to drive an east-west atmospheric overturning, which Bjerknes named the Walker circulation. Bjerknes was able to link major decreases in the strength of the east-west gradient in equatorial sea surface temperature to decreases in the strength of the Walker circulation. He also linked changes in the gradient of sea surface temperature to disturbances in the planetary-scale atmospheric wave pattern over the sector covering the North Pacific Ocean and North America. This new information justified the cautious use of equatorial oceanic and atmospheric conditions for experimental climate forecasts (Bjerknes 1966, 1969). Bjerknes concluded that the variations in atmospheric heat input from the equatorial

OCR for page 12
Learning to Predict Climate Variations Associated with El Niño and the Southern Oscillation: Accomplishments and Legacies of the TOGA Program ocean were responsible for the observed interannual atmospheric fluctuations. The formulation of this hypothesis marked the beginning of a new era of climate studies in which the tropical ocean and global atmosphere were analyzed and modeled as coupled components of the global climate system. EMERGENCE OF A COHERENT EFFORT (1970–1984) The studies of Walker and Bjerknes established the coupled nature of El Niño in the ocean and the Southern Oscillation in the atmosphere as a single phenomenon now called ENSO. However, there was a need for an observing system and quantitative models that were capable of describing their interaction. The Bjerknes hypothesis provided a conceptual model of the feedbacks between the ocean and atmosphere that maintained the extremes of ENSO. The hypothesis emphasized the importance of sea surface temperature in producing the observed atmospheric anomalies during ENSO. However, many gaps in understanding remained, such as the mechanisms that control the initiation, development, termination, and irregular occurrence of warm events. The testable physical hypothesis of Bjerknes, however, provided a framework for the development of long-term observational strategies, process studies, and modeling research that would all join empirical studies to become major elements of the TOGA program (see Figure 2 in NRC 1990). Nature provided scientists and policy makers with ample justification for continued scientific efforts. The 1972 and 1976 warm anomalies of El Niño along the South American coast were accompanied by alarming declines in the anchovy population. The economic repercussions of the anchovy decline were felt in the global economy (Barber 1988). The 1976–77 northern winter, which coincided with the latter warm episode, brought drought to California and record cold, accompanied by fuel shortages, to much of the central and eastern United States (Canby 1984). This coincidence of events raised public consciousness about possible connections, although some of these events were probably not related to El Niño. A 30-day-hindcast simulation (Miyakoda et al. 1983) of a blocking event during that winter helped make the case for the possibility of long-range dynamical forecasts. A major breakthrough in the understanding of ENSO was the development of a more realistic view of the role of ocean dynamics in the evolution of upper-ocean thermal structure. Observational programs played a key role. During the 1970s, scientists involved in the North Pacific Experiment (NORPAX) Program (1971–1980), funded by the National Science Foundation and the U.S. Navy's Office of Naval Research, established the first oceanic monitoring system based on expendable bathythermograph (XBT) instruments deployed by ships of opportunity and island-based sea-level monitoring. Six major oceanic field

OCR for page 12
Learning to Predict Climate Variations Associated with El Niño and the Southern Oscillation: Accomplishments and Legacies of the TOGA Program programs begun during the pre-TOGA period—INDEX (Indian Ocean Experiment, 1976–1979), Hawaii to Tahiti Shuttle Experiment (1979–1980), EPOCS (Equatorial Pacific Ocean Climate Studies, 1979–1994), PEQUOD (Pacific Equatorial Ocean Dynamics Experiment, 1981–1983), SEQUAL (Seasonal Response of the Equatorial Atlantic, 1983–1984), and Tropic Heat (1984–1987)—provided an observational basis for a better understanding of the annual cycle and interannual variability of the three tropical oceans. These programs focused attention on the importance of equatorial ocean dynamics for ENSO. Wyrtki (1975) supplied a key piece of the ENSO puzzle. He recognized the importance of the dynamical response of the upper equatorial Pacific Ocean to the large-scale weakening of the trade winds in the central Pacific during the onset of a warm event. On the basis of empirical evidence, Wyrtki hypothesized that the warm equatorial surface waters that developed at the western side of the basin during the period of abnormally strong trade winds would “surge” eastward, probably in the form of an equatorial Kelvin wave*, depressing the thermocline as it passed. This would cause an abrupt rise in sea level and an increase in sea surface temperature through a reduction in the effects of upwelling on the eastern side of the basin. The Kelvin wave generated by this surge would cross the Pacific in approximately 50 days, presenting a potentially predictable dynamical feature of ENSO with a seasonal time scale. On the basis of this hypothesis linking temperature and wind anomalies, Wyrtki et al. (1976) made the first ENSO forecast. A series of cruises (the “El Niño Watch”) was conducted in 1975 off the coast of Ecuador and Peru, but a warm event did not occur. The reality of equatorially trapped Kelvin waves in the ocean was the subject of intense debate among theoreticians and observers, but was firmly established by analysis of sea-level and subsurface data from NORPAX, EPOCS, and PEQUOD. Evidence was found that connected the waves to forcing by wind stress (Knox and Halpern 1982, Eriksen 1982, Eriksen et al. 1983, Lukas et al. 1984, Mangum and Hayes 1984). This finding provided justification for interpreting the low-frequency variability in terms of long equatorial waves with low-order baroclinic structures having maximum amplitude in the upper ocean (see, e.g., Eriksen 1985). In contrast to features in the middle-and high-latitude oceans, large-scale features of the equatorial ocean circulation appeared to be linked to changes in the surface wind stress on monthly-to-seasonal time scales. *    Equatorial Kelvin waves are disturbances of thermal structure and currents, usually large-scale along the equator, trapped within a few degrees latitude of the equator. They propagate eastward at roughly 1–3 m/s. A different class of tropical disturbances, known as equatorial Rossby waves, propagates westward even more slowly. The region where these types of Kelvin and Rossby waves can exist, roughly within ten degrees latitude of the equator, is known as the “equatorial waveguide”. See Gill 1982 for a more detailed explanation.

OCR for page 12
Learning to Predict Climate Variations Associated with El Niño and the Southern Oscillation: Accomplishments and Legacies of the TOGA Program Simplified dynamical models of the equatorial upper ocean were found to reproduce the most important observed features of interannual sea-level and thermocline-depth variability, when the model ocean was driven by either idealized or observed wind-stress patterns (Hurlburt et al. 1976, McCreary 1976, Busalacchi and O'Brien 1981, Cane 1984). Not only did the simulated low-frequency oceanic waves generated by the imposed wind-stress anomalies affect the local region of the equatorial ocean where the waves were generated, but the waves could propagate systematically along the equatorial wave guide to remote coastlines. Ocean modeling studies also provided a better understanding of the large-scale adjustment of an equatorial ocean basin to time-varying wind forcing. Philander and Pacanowski (1980) demonstrated that the adjustment time scale of the equatorial upper ocean to uniform zonal wind forcing depends on the longitudinal width of the equatorial channel, and, furthermore, that low-order baroclinic Kelvin and Rossby waves may play a dominant role in the adjustment process. The dynamical processes by which equatorial waves are reflected at boundaries on the eastern and western sides of the equatorial oceans were investigated (Cane and Moore 1981, Anderson and Rowlands 1976, Cane and Sarachik 1977, Moore and Philander 1977, Clarke 1983, Cane and Gent 1984). A simulation of the response of a bounded equatorial ocean to a uniform meridional wind forcing demonstrated that cross-equatorial flow over the eastern equatorial Pacific Ocean might also play an important role in determining the structure and variability of the “cold tongue” of sea surface temperature extending off the South American coastline (Philander and Pacanowski 1981b). In parallel, atmospheric scientists were coming to a clearer understanding of the nature of thermally forced circulations in the tropics. Gill (1980) pointed out that thermally forced equatorial-wave solutions discussed by Matsuno (1966) and Webster (1972) bore a striking resemblance to the gross horizontal structure of climatological wind fields in the vicinity of large-scale tropical atmospheric heat sources. Zebiak (1982) demonstrated that when it is assumed that anomalous atmospheric heating is directly coupled to sea surface temperature anomalies, forced equatorial waves in the atmosphere can explain a significant fraction of the observed wind anomalies over the central and western Pacific Oceans during the different phases of ENSO. The coupled nature of ENSO was beginning to emerge. It was clear that observed anomalies of tropical sea surface temperature were able to generate plausible anomalies of tropical atmospheric circulation, and that anomalies of atmospheric surface wind stress were able to generate plausible fields of anomalies of both equatorial upper-ocean circulation and sea surface temperature. These provided the elements of a feedback loop between the ocean and atmosphere. A linear stability analysis by Philander et al. (1984) demonstrated

OCR for page 12
Learning to Predict Climate Variations Associated with El Niño and the Southern Oscillation: Accomplishments and Legacies of the TOGA Program that a simplified model of the tropical ocean and atmosphere system might contain large-scale propagating—possibly unstable—modes not present in either the ocean or the atmosphere alone. The working hypothesis for ENSO was thus extended to include not just oceanic Kelvin and Rossby waves, but large-scale Kelvin-like or Rossby-like modes of the tropical coupled ocean-atmosphere system. During the pre-TOGA period, atmospheric scientists also began intensive use of both global data sets and more complex models. They quantitatively examined the effects of regional equatorial circulation and heating anomalies on the atmospheric circulation in remote regions, “teleconnections” in the terminology of Bjerknes. Applying empirical techniques to the study of global atmospheric data sets, investigators succeeded in isolating patterns of the middle-and upper-tropospheric planetary waves associated with ENSO (Horel and Wallace 1981, Trenberth and Paolino 1981, van Loon and Madden 1981). Building on the early work of Rowntree (1972, 1976), modelers attempted to reproduce these patterns with the use of complex atmospheric general-circulation models. In these numerical experiments, “control” integrations were performed with a prescribed sea surface temperature to obtain the normal time-averaged state of the model. The general-circulation models were then perturbed by introducing anomalous patterns of sea surface temperature as lower boundary conditions. Statistically significant correlation patterns were obtained using several general-circulation models (Julian and Chervin 1978, Keshavamurty 1982, Blackmon et al. 1983, Shukla and Wallace 1983, Nihoul 1985—especially within the last, Cubash 1985, Boer 1985, von Storch and Kruse 1985, Palmer 1985, and Fennessy et al. 1985). Although a definitive explanation for the mechanism causing anomalous patterns of atmospheric planetary waves was lacking, theoretical studies conducted during this period indicated that the energy of planetary-scale waves generated by a regional process in the equatorial zone was able to propagate to remote regions of the global atmosphere within a few days under certain conditions. The striking resemblance between the observed and theoretically obtained wave patterns led to a working hypothesis that Rossby-wave propagation might be responsible for the “teleconnections” between the tropical Pacific and other locations (Hoskins and Karoly 1981; Webster 1981, 1982; Simmons 1982; Simmons et al. 1983). Results of atmospheric and oceanic modeling efforts, as well as statistical analyses of the lag-lead relationships between sea surface temperature and climate anomalies over North America (Barnett 1981), indicated that deterministic ocean models driven by observed wind stress might be used to produce useful forecasts a season in advance, even if it were not possible to predict the evolution of an entire ENSO cycle.

OCR for page 12
Learning to Predict Climate Variations Associated with El Niño and the Southern Oscillation: Accomplishments and Legacies of the TOGA Program DEVELOPMENT OF THE TOGA PROGRAM In 1982, the international climate research community established a study group under the leadership of the late Adrian Gill to examine short-term climate variability. The group considered the growing evidence that a significant part of climate variability on seasonal-to-interannual time scales could be understood in terms of interactions between the three major tropical oceans and the global atmosphere. ENSO was clearly the largest and most coherent signal in the seasonal-to-interannual range. Other examples of interest to the international community included the connection of anomalies of sea surface temperature in the tropical Atlantic with precipitation over northeast Brazil and the Sahel (Hastenrath and Heller 1977, Moura and Shukla 1981), and the connections of anomalies of sea surface temperature in the eastern Indian Ocean (Streten 1983) with rainfall anomalies over Australia. It was recognized that a longer-term investigation would be needed to verify the working hypotheses. The study group also recognized that such an investigation might be feasible because the seasonal-to-interannual oceanic variability appeared to be confined primarily to the upper ocean of the tropics, while the extratropical ocean circulation on these time scales appeared to be relatively insensitive to remote atmospheric influences. This recognition narrowed the scope of the investigation and provided a scientific basis for setting priorities on the observational and modeling activities. When the international TOGA Scientific Steering Group was formed in 1983 to define the TOGA program's scientific and technical scope, the NRC's Climate Research Committee was already engaged in drafting its own scientific plan for ENSO research (NRC 1983). By this time, ENSO research had emerged as a major theme of the U.S. National Climate Program. It is interesting to note that ENSO research had not been emphasized in the planning documents that supported the drafting and passage of the National Climate Program Act in 1978. ENSO research emerged from the grass-roots effort of a relatively small number of scientists, supported by traditional funding sources. These funding sources included grant sections of the National Science Foundation and mission-oriented agencies such as the National Oceanic and Atmospheric Administration, which has responsibility for monitoring and predicting daily-to-seasonal weather and climate. Given the interests of the U.S. scientific and policy-making communities, it was agreed from the outset that the primary U.S. contribution to the international TOGA field effort would be made in the Pacific and would focus on ENSO. The time scale of phenomena examined under the program would be seasonal to interannual, and the measure of progress in understanding would be the extent to which global coupled ocean-atmosphere circulation models, when initialized by state-of-the-art observations, would be able to model and predict oceanic and atmospheric phenomena.

OCR for page 12
Learning to Predict Climate Variations Associated with El Niño and the Southern Oscillation: Accomplishments and Legacies of the TOGA Program The 1982–83 El Niño exerted a very significant influence on the scientific planning for TOGA. In addition to focusing attention on the Pacific Ocean, this event surprised the scientific community with its magnitude and manner of evolution. However, the very existence of the large warming in the Pacific was obscured by the 1982 eruption of El Chichon. The aerosol veil resulting from the eruption obscured viewing by the AVHRR (Advanced Very-High Resolution Radiometer) satellite instrument and made satellite-derived estimates of sea surface temperature unreliable. A composite of six warm events after 1949 had shown a tendency for warm events to begin, peak, and end at preferred times of year, as well as to exhibit the hypothesized connection between warm El Niño events and the Southern Oscillation (Rasmusson and Carpenter 1982). The 1982–83 event didn't follow this pattern. It skipped the typical “onset phase” characterized by anomalously warm (El Niño) conditions along the South American coast during April, which typically spread westward along the equator toward the central Pacific (Rasmusson and Wallace 1983). Apparently, it was possible for individual ENSO cycles to differ significantly in detail from the composite picture. The concept of a rapid-response observational strategy (see NRC 1983) was therefore abandoned in favor of continuous, “real time” monitoring of the equatorial Pacific. SCIENTIFIC PLAN FOR TOGA (1985). The challenge for the organizers of the TOGA Program was to formulate a scientific plan that would reflect a new way of thinking about the climate system. It would not be possible to make fundamental progress by consideration of either the atmosphere or the oceans alone. Observational systems and models were needed to describe the system encompassing the tropical oceans and the global atmosphere, a system in which the processes determining sea surface temperature, atmospheric heating, surface wind stress, and ocean circulation were related to each other in feedback loops (see Figure 1). The scientific objectives adopted in the international plan (WCRP 1985) were: To gain a description of the tropical oceans and the global atmosphere as a time-dependent system, in order to determine the extent to which this system is predictable on time scales of months to years, and to understand the mechanisms and processes underlying that predictability; To study the feasibility of modeling the coupled ocean-atmosphere system for the purpose of predicting its variations on time scales of months to years; [and]

OCR for page 12
Learning to Predict Climate Variations Associated with El Niño and the Southern Oscillation: Accomplishments and Legacies of the TOGA Program To provide scientific background for designing an observing and data transmission system for operational prediction if this capability is demonstrated by the coupled ocean-atmosphere system. While the formal objectives of the international TOGA Program did not mention ENSO specifically, it had been known that the ENSO phenomenon dominated the interannual variability of sea surface temperatures in the tropical Pacific (Weare et al. 1976). Furthermore, it was believed, although not demonstrated until recently (e.g., Kawamura 1994, Mann and Park 1994), that ENSO drives the dominant seasonal-to-interannual signal in sea surface temperature globally. In fact, the 1983 NRC report that presented a science plan on short-term climate variations strongly emphasized ENSO, even in the report title. The early decision, therefore, was to concentrate on the big seasonal-to-interannual signal—ENSO in the tropical Pacific. It was also known that ENSO occurs irregularly with a spectral peak at about a 40-month period (Rasmusson and Carpenter 1982). The irregular nature of ENSO argued for an open-ended program that would continue until a few ENSO cycles could be observed in detail. Practical considerations of national and international program management dictated that the program have a predetermined completion date. It was decided that TOGA would have a ten-year duration, beginning on 1 January 1985. This was perhaps the shortest period that would permit significant progress by examining more than one ENSO cycle and the longest period likely to be approved by the sponsoring U.S. agencies and other participating governments, even if it was not sufficient to capture the full range of ENSO behavior. The scientific objectives clearly established the performance of coupled ocean-atmosphere models as the ultimate measure of success for the TOGA Program. Consistent with the international objectives, the scientific plan for U.S. participation in TOGA (NRC 1986) identified improvement of atmospheric and oceanic circulation models, and the successful coupling of these components, as central elements. It was also recognized that model development and predictability studies would be impossible without a major improvement in the observing system over the data-sparse areas of the tropical oceans. Prediction schemes would require a specification of the initial thermal and current structure of the upper ocean. The fragmentary oceanic and atmospheric observations during the pre-TOGA period had left unanswered many questions about the structure and dynamics of the seasonal-to-interannual variability of the coupled system. An observational system of unprecedented spatial and temporal coverage would be required to guide model development, validate model performance, and permit the empirical description of at least one ENSO cycle. The U.S. contribution to the TOGA observing system would focus on the Pacific Ocean, and on the ENSO phenomenon in particular.

OCR for page 12
Learning to Predict Climate Variations Associated with El Niño and the Southern Oscillation: Accomplishments and Legacies of the TOGA Program Finally, it was envisioned that one or more process studies would be required for significant improvement of model parameterizations of key physical processes, such as upper-ocean mixing, atmospheric turbulence and convection, and interactions between clouds and radiation. Process studies would also permit sorting out of the relative importance of the myriad atmospheric and oceanic phenomena with similar or shorter time scales that might play a role in altering the behavior of coupled modes of seasonal-to-interannual variation. The drafters of the original science plan (NRC 1983) could not determine the optimal location for the process experiment(s)—eastern, central, or western Pacific—and therefore recommended a separate experiment for each. As it turned out, an experiment was performed in each of these regions. The major process study during TOGA, COARE, was performed in the western Pacific, the region thought to be most important for the development of ENSO.