Air-Sea Interactions in the Tropics:
Seasonal to Interannual Variability

Physical Oceanography - Numerical Modeling

Use the links below to quickly navigate this page.

Table of Contents: The TOGA/GOALS Experiments The TOGA-COARE Program The Pan-American Climate Study The Global Ocean Atmosphere Land Systems Experiment Indo-Pacific Ocean Circulation Modeling

Please see the bottom of this page for Preface of the June 29, 1998 Special Issue of the Journal of Geophysical Research, co-edited by L. Rothstein, summarizing the scientific community's efforts in understanding the El Nino phenomenon.

The research programs described on this page are all focused on the over-riding objective of contributing to our understanding, and prediction, of the El Niño. For a look at the significant effort that the scientific community is investing in El Nino research, we recommend you visit NOAA's El Nino Theme Page and FSU's COAPS El Nino Theme Page.

Forecasting of El Niño/La Niña has taken dramatic strides in the last few years. To look at the latest forecasts, please visit the Center for Ocean-Land-Atmosphere Studies and NOAA's Climate Prediction Center.


Table of Contents

The Tropical Ocean Global Atmosphere (TOGA)/GOALS Experiments

  • Coupled Ocean-Atmosphere Response Experiment (COARE)
  • Pan-American Climate Study (PACS)

  • (Rothstein, Dr. William Kessler [PMEL], Dr. Dake Chen [Columbia/Lamont])
  • Global Ocean Atmosphere Land Systems (GOALS)

  • (Rothstein, Dr. Antonio Busalacchi [NASA/GSFC], Dr. Dake Chen [Columbia/Lamont])
  • Indo-Pacific Ocean Circulation Modeling (CSIRO Division of Marine Research & the Australian LWRRDC)

  • (Rothstein, Zhang, Drs. Stuart Godfrey, Gary Meyers, Andreas Schiller and Peter McIntosh [CSIRO])

    Funding for these projects comes from the National Science Foundation (NSF) and the National Oceanic and Atmospheric Administration (NOAA).

    TOGA-COARE (TOGA Coupled Ocean Atmosphere Response Experiment)

    We are supported to continue our upper ocean modeling efforts for the Tropical Ocean Global Atmosphere-Coupled Ocean Atmosphere Response Experiment (TOGA-COARE) program, which has seen significant progress over the past few years in meeting our original objectives. Our work plan then was based upon a need for the proper representation of tropical ocean mixed layer physics within a three-dimensional ocean circulation model. Building directly upon this we list two fundamental goals for the present research: The first goal relates to our continued intention of working closely with COARE field scientists. The three-dimensional tropical upper ocean circulation model that we have developed with COARE funding over the past few years was specifically designed for the opportunity of working toward a COARE field data/model analyses program, and we continue to do so now that the field program results are being made available.

    The second goal directly addresses the highest priority of the entire COARE program. We focus this effort on regional air/sea interactions, specifically, on resolving the dynamics and thermodynamics on `westerly wind burst' scales, thought to be an important component of the regional coupled system, as well as an important component of the COARE region's physical influence on the rest of the tropical Pacific. We are working towards developing a model that both resolves the ocean domain analogous to the COARE field program and allows for explicit resolution of air/sea coupling on these wind burst scales. The numerical experiment strategy is designed for COARE objectives and our work plan calls for formal interactions with the field scientists of the program.

    We design the numerical experiments to test a number of hypotheses that are directly relevant to COARE goals. New hypotheses will undoubtedly evolve as the COARE observational database comes under scrutiny on its own and in light of the numerical ocean.

    Our initial set of general working hypotheses are:

    1. The large scale ocean circulation of the western Pacific is governed primarily by fluctuations in the local zonal wind stress and remotely forced equatorial wave dynamics.
    2. High frequency (`westerly wind burst') forcing of the ocean in the western Pacific/COARE domain has a rectified oceanic effect on the far-field variability.
    3. Understanding the regional coupled air/sea response on `westerly wind burst' scales is important to understanding the COARE region's connections to El Nino - Southern Oscillation (ENSO) development.
    We proceed to briefly detail these. Testing ALL of these hypotheses require close collaborations with COARE field scientists.

    The siting of the Intensive Observing Period (IOP) observational array provides a unique opportunity for making progress against hypothesis (i). The IOP site is an interesting area, close enough to the equator to be influenced by the Yoshida jet response to equatorial zonal wind changes yet taken far enough off the equator to be affected by other processes. The tools required to test (i) are the observational database and an upper ocean model that can properly account for the unique nature of the regional mixed layer. This means that the ocean model must have high resolution in and above the thermocline (for the time scales we highlight, the resolution of subthermocline variability is thought to be unimportant), properly represent regional mixing processes, and actively allow for hydrological forcing through a dynamical oceanic salinity budget. This was precisely the focus of our previous work for COARE (especially the mixing algorithms and activating the hydrology) and our present experiment strategy builds directly upon it.

    Pivotal in understanding the regional oceanic variability and the influence of this variability on the large scale air/sea system are the relatively high frequency `westerly wind bursts' associated with tropical depressions and storms, hypotheses (ii) and (iii). Wind bursts usually occur during the period from November - April (motivating the timing of the IOP) near the equator, typically lasting a few days but sometimes lasting a few weeks. They commonly reach storm amplitudes, i.e. as high as 15 m/sec causing intense local mixing of the upper ocean. There may also be heavy rainfall associated with these events, further complicating the upper ocean response. These wind systems are observed to propagate eastward into the western Pacific from the Indonesian region, suggestive of monsoon influence, but in-situ bursts also develop, undoubtedly governed by intense regional air/sea interactions. There are serious gaps in our knowledge of which atmospheric processes produce this relatively high frequency forcing. However, there is evidence that these wind bursts are an essential component of developing tropical depressions and storms. Westerly wind burst events have been reported to exceed 10 m/sec during the COARE IOP that have lasted one week and may have been influenced by nearby tropical cyclones. Furthermore, the dynamical connection between these strong low-level westerly wind bursts and the lower frequency Madden-Julian Oscillation (MJO) has also been made. Observations suggest that a distinctive feature of the MJO is the occurrence of westerly wind bursts in association with and often just following deep convective activity. Ample evidence now points to wind burst scale dynamics as plausibly at the dynamical heart of the COARE regional air/sea coupled system.

    There is much speculation about the far-field response generated by these wind bursts and their influence on ENSO, with the oceanic response the focus of our hypothesis (ii). A testable sub-hypotheses under (ii) is that an increased number of westerly wind bursts increases the eastward flow of fresher, warmer water along the equator, shifting atmospheric convection eastward with it. The longer term `memory' of the upper ocean of these forcings is also testable under our ocean modeling strategy. Additionally, the intensity and frequency of these tropical storms over the warm pool appear to be sensitive to interannual variations of the upper ocean structure, providing another clue that there is a strong interscale relationship at work in the warm pool. These issues are important in determining the influence of the warm pool dynamics/thermodynamics on the entire tropical ocean variability, and vice versa.

    COARE has set out to obtain an understanding of the energy budget in the warm pool and our third hypothesis purports that this goal cannot be accomplished unless one considers an active REGIONAL air/sea coupling. We concentrate our efforts first on westerly wind burst scales whose connections to larger scale atmospheric fluctuations and to far-field oceanic variability has been discussed in the above paragraphs. Wind bursts can greatly modify the regional energetics through their influence on surface fluxes of heat, moisture and momentum. Their structure, amplitude, and pathway is certainly determined by the underlying sea surface temperature (SST) field (the energy source), which in turn is evolving through dynamical and hydrological forcing of the upper ocean due to these same winds. Although much work has been focused on the ocean response to passive wind bursts, little work has been attempted in the fundamental description of this as a coupled air/sea system.

    There are both positive and negative feedback mechanisms at work in the coupled regional system. A detailed plausible scenario for positive feedback describes the wind bursts as an instability of the regional coupled air/sea system. High local SST induces a direct atmospheric thermal cell whose upward branch causes significant moisture flux aloft. Vortex stretching associated with this upward flow induces the spinup of paired cyclonic vortices at low levels centered near the equator. These are topped by anti-cyclonic vortices at high levels. The surface signature of this evolving system is a low level westerly jet, a `wind burst', on or near the equator. As the jet amplifies, increased air/sea transfer further enhances the upward thermal cell, resulting in a self-amplifying disturbance. The movement of this system is dependent upon the three-dimensional circulation of the upper tropical ocean which is dynamically linked with the source of the instability; changing SST in the warm pool due to local and non-local effects. Thus, progress with regional coupled models will require precise modeling of the physical processes of the coupled system which, in turn, requires emphasis on observing and modeling the surface fluxes in the warm pool.

  • Please see the COARE Program for the scientific details, and the data collected, during this program.

    Pan-American Climate Study (PACS)

    Kessler, W.S., L.M. Rothstein and D. Chen (1997): The annual cycle of SST in the eastern tropical Pacific, as diagnosed in an ocean GCM. Journal of Climate, accepted.
    PACS Program Page
    See the Scientific Prospectus
    Implementation plan
    The purpose of this work is to study the ocean dynamics and thermodynamics of the annual cycle of sea surface temperature (SST) in the eastern tropical Pacific using a recently-developed tropical upper ocean model. A variety of studies have shown that, first, oceanic processes are critical to the annual variation of east Pacific SST; and second, that these SST changes are crucial to the annual march of the monsoon between equatorial South America and Central and North America. Issues to be addressed include the role of the meridional cross-equatorial wind in the May-July intensification of the east Pacific equatorial cold tongue; the vertical-meridional (equatorial upwelling) circulation and its effect on mixed layer depth; the role of annually-occurring high-frequency variability (the Madden-Julian Oscillation and the tropical instability waves); the processes which influence the annual cycle of SST in the Costa Rica dome; and the co-variability of coastal upwelling, SST and the stratus cloud decks off the coast of South America. The model to be used is a sigma-coordinate primitive equation formulation, originally developed by Drs. Peter Gent and Mark Cane in 1989, which allows for time- and spatially-varying layer thicknesses; this feature has been shown to be particularly important in simulating the variability of the eastern tropical Pacific where the thermocline and undercurrent come close to the surface. Our group has developed a new mixed layer model as the top layer of this Gent and Cane model. This mixed layer model incorporates the physics of a so-called `bulk' mixing configuration (allowing the mixed layer to change thickness by entrainment and detrainment) and a popular dynamical instability model. The heat fluxes for model input requires only cloud cover and wind speed as atmospheric input (thus avoiding specification of air temperature which tends to predetermine the SST variations). The model has an active hydrology driven by specification of evaporation minus precipitation.

    The research falls into two categories. First we are carrying out a series of process studies to address specific hypotheses regarding the east Pacific SST; second we are using the results of the process studies to improve the model simulation with an eye towards future coupling with an atmospheric model. At each stage, the model results are quantitatively assessed in comparison with data sources available, including the TAO moored temperature and velocity time series, satellite-based SST products, and compilations of hydrographic data which have been produced by the PI's and others. We expect that this study will address and refine the definition of problems which might profitably be studied in future PACS field programs.

    Global Ocean Atmosphere Land Systems (GOALS)

    CLIVAR Program
    GOALS Program
    As the Tropical Ocean Global Atmosphere (TOGA) program concludes and leads into a coordinated study of the Global Ocean Atmosphere Land System (GOALS) it is anticipated that there will be an increasing emphasis on the ocean-atmosphere coupling beyond the tropical Pacific. Research that strives to improve the predictability of the coupled system on regional to global scales will need to consider the degree to which simulating sea surface temperature (SST) variability in the other tropical oceans, as well as at higher latitudes, may improve short-range (seasonal-to-interannual) climate forecasts. The purpose of this research is to study the oceanic exchanges of mass and heat, both zonally and meridionally, in the Indian-Pacific Ocean region and how these transports influence interannual changes in SST there.

    To accomplish this we apply the GSO upper ocean general circulation model first to study the interactions between the tropical/subtropical upper Indian Ocean and the upper tropical Pacific Ocean and subsequently the entire upper subtropical Pacific Ocean. This program of study is designed to test a number of hypotheses formulated upon the role of the components of this pan-oceanic system in determining anomalous seasonal-to-interannual circulation and SST variability in the tropics and subtropics. To meet this overall objective we list two necessary modeling goals which we need to accomplish: To apply the GSO tropical Pacific upper ocean circulation model:

    The Indo-Pacific maritime warm pool will provide the focus for study as the temporal/spatial maintenance and variability of this pan-oceanic feature is thought to be central to understanding climate variability on seasonal-to-interannual time scales through the pool's interactions with the overlying atmosphere.

    We design the numerical experiments to test a number of hypotheses that are directly relevant to TOGA goals. New hypotheses will undoubtedly evolve as the TOGA observational database continues to come under scrutiny on its own and in light of our proposed and other numerical oceans.

    Our initial set of general working hypotheses are:

    1. The temporal/spatial maintenance and variability of the maritime warm pool on seasonal-to-interannual time scales is governed by a number of competing processes which need proper accounting in numerical upper ocean circulation models:
      1. the large-scale wind driven circulation of both Indian and Pacific Oceans;
      2. the Indo-Pacific throughflow;
      3. the local changes in surface/upper ocean turbulent heat and momentum fluxes; and
      4. the interactions of all of the above.
    2. Large scale changes in the tropical upper ocean circulation on interanual time scales are important in understanding mid-latitude thermal variations (SST, heat content, heat transport) on these same time scales.


    Indo-Pacific Ocean Circulation Modeling (CSIRO & LWRRDC)

    The CSIRO Division of Marine Research and the Australian Land & Water Resources R&D Corporation (LWRRDC) helped organize the one year sabbatical visit during 1997 of Lew Rothstein to the CSIRO Hobart lab to work with climate research scientists there and at the Division of Marine Research (DMR) towards a better understanding of long-term climate prediction for Australia. More specifically, two ocean general circulation models were similarly configured. Both models used available sets of observed winds and shortwave radiation data from 1985 through 1990; a similar parameterisation of other heat flux components was also used, as well as a similar (crude) treatment of rainfall effects. Thus, differences between the SSTs obtained in the two models should be due to differences in model physics and numerics. The results are now being compared to determine the strengths and weaknesses of both models in understanding ENSO.

    The original objectives proposed to LWRRDC were:

    1. Set up an intercomparison program between the GSO ocean circulation model, the LWRRDC model, and observations, to begin to highlight model deficiencies for helping to improve model ocean physics.
    2. Set up a data assimilation program for use in the GSO ocean circulation model (the method will be different from that in the LWRRDC model)
    3. In collaboration with CSIRO's Division of Atmospheric Research (DAR) and Australia's Bureau of Meteorology Research Centre (BMRC), couple an active AGCM to the Rothstein ocean model to begin to understand the role of the coupled air/sea system in predicting seasonal-to-interannual variability beyond the Pacific tropics.
    4. Progress on the first objective has been significant and is described below in the Sections entitled `The Annual Mean and Seasonal Cycle', `Interannual Variability', and `ENSO Process Studies'. Progress with the second objective proceeds in parallel with another (NSF) funded project and is summarized in the Section entitled `Data Assimilation'. The third objective proved too ambitious to attempt within the present time frame, though discussions are now under way to couple the GSO ocean circulation model to both the DAR and BMRC atmospheric models over the next few years.

      The Annual Mean and Seasonal Cycle

      The CSIRO ocean model (referred to below as ACOM-2) was run for 20 years with the seasonal mean of the fluxes over 1985-1990. A description of the Australian Community Ocean Model (ACOM) modeling efforts can be found at
    5.  ACOM Model.
    6. Flux corections needed to be applied to obtain mean seasonal temperatures close to observations, but in the tropics these corrections were mostly less than 25 W/m2 -- i.e. similar to uncertainties in observational estimates, especially the shortwave radiation. Introduction of a crude parameterisation of tidal mixing in Indonesia significantly reduced the flux corrections needed here. The fairly complex mean seasonal cycle of mixed layer depth was well reproduced in the model, as well as the seasonal cycle of currents. A paper describing seasonal Indian Ocean behaviour in detail in the model has been submitted for publication (Schiller et al, 1997.)

      The GSO upper ocean general circulation model was also run over a 20 year spinup with the same sesonal mean fluxes and tidal mixing parameterisation as for ACOM-2. As noted above the intercomparison of this model with ACOM-2 is proceeding with the initial results showing strong agreement on the large scale. Important differences do exist between the heat budgets of the two models within the Indonesian Archipelago which should enable identification of the responsible physical processes. The monthly mean figures of SST for both the GSO ocean circulation model simulation and the observations are found on

    7.  Monthly Mean SST Figures.
    8. Contours of the various terms of the model heat budget are found on

    9.  Annual Mean and Regional Heat Budget.
    10. And an animation of the model's representation of the seasonal cycle are found at

    11.  Seasonal Cycle Animation.
    12. Interannual Variability

      Both models were initialized with the seasonal cycle (see above) and then forced with surface fluxes as observed between the period 1985-1990. The ACOM-2 model accounted for about 60\% of the interannual variance of SST over the period 1985-1990, in the tropical Pacific and Indian Ocean. Nevertheless, some serious problems have been identified: the model did not get the weak warm SST event in the central equatorial Pacific in 1990, getting weak cold SST's instead. Similar problems have been reported in other models; potential causes are that the winds may have been wrong, the lack of interannual freshwater forcing may have contributed, or it may relate to persistent problems with ocean physics in the equatorial Pacific. The GSO ocean circulation model intercomparison should help decide which of these problems is responsible for these problems, however the initial interannual experiment with this model became unstable after 3 years of integration. We have identified that problem and are presently re-running the model. Results will be reported here when they become available.

      ENSO Process Studies

      Picaut et al. (1997) suggested a mechanism for the ENSO phenomenon which is different from the more usual "Delayed Oscillator" mechanism, which operates (for example) in the Cane-Zebiak model. We have devised a simple experiment, using the GSO ocean circulation model and coupling to an idealised atmosphere, to test the Picaut et al. hypothesis in more detail. Runs of this model are in progress.

      We have also undertaken tests, both in the ACOM-2 and GSO model, to see whether the model reproduces the observed seasonal cycle of surface heat fluxes in the Indonesian region. Preliminary results with ACOM-2 suggest that the model reproduces fluxes fairly well, which implies that the crude parameterisation of vertical mixing used in the model is probably fairly satisfactory. However, this result is accompanied by a large cancellation between vertical and horizontal advective terms, which we do not understand at present.

      Data Assimilation

      A strategy has been laid out for the development of the generalized inverse of the nested-grid GSO ocean circulation model for the purpose of data assimilation. Two primary tasks are currently in progress. We are developing a linear iteration procedure, whereby the full non-linear model solution is approximated by a series of solutions to a linearized version of the equations. This is a necessary step for the implementation of the generalized inversion of the forward model. We have also begun to construct the adjoint of the forward model. With these two components in place, the construction of the full inversion using the representor method will be relatively straightforward. This project is funded mainly by an NSF grant with the final inverse model will be made available as a tool for the research under this LWRRDC project (See the TOGA COARE data assimilation project).

      The TOGA Decade: Reviewing the Progress of El Nino Research and Prediction

      Journal of Geophysical Research, Volume 103, #C7, June 29, 1998, 343 pages.

      Edited by: D.L.T. Anderson, E.S. Sarachik, P.J. Webster and L.M. Rothstein

      Physical Oceanography - Numerical Modeling

      The Tropical Ocean-Global Atmosphere (TOGA) program began as a program of the World Climate Research Program in 1985 with the following objectives: (1) 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; (2) 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 (3) 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.

      TOGA addressed these objectives by building the TOGA Observing System, by conducting a major process study in the tropical Pacific (the Coupled Ocean-Atmosphere Response Experiment, COARE), by developing a sequence of coupled ocean-atmosphere models of the tropical Pacific, by conducting a program of prediction studies through the TOGA Numerical Experimentation Group, by conducting analytic and diagnostic studies of the ENSO phenomenon, and by relating ENSO to seasonal-to-interannual variability in other tropical regions, especially in the monsoon region. These studies were coordinated by the TOGA Scientific Steering Group (SSG), were supported by the Intergovernmental TOGA Board, and were implemented by the International TOGA Project Office.

      At the last meeting of the TOGA SSG, a decision was made to solicit a series of review papers in order to report on and solidify the progress made during the TOGA decade in understanding seasonal-to-interannual variability and predictability. The papers in this volume constitute the final set of review papers for the International TOGA Program and correspond directly to the major areas of effort during the TOGA decade. The authorship is international in scope and represents the wide range of countries and scientists that actively participated in the TOGA program.

      The legacies of TOGA lie in the TOGA Observing System that is still maintained in the tropical Pacific Ocean; in the substantial progress that was achieved in both understanding the phenomenon of ENSO and in predicting its phases; in the establishment of an International Research Institute for Climate Prediction that grew out of a recommendation by the Intergovernmental TOGA Board; and in the follow-on program of the WCRP, the Global Ocean-Atmosphere-Land System (GOALS) component of the Climate Variability and Predictability (CLIVAR) Program, designed to pick up where TOGA left off by expanding the study of seasonal-to-interannual variability to the entire globe and by understanding its interactions with other crucial climatic time scales.

      The editors would like to recognize the lives and work of two extraordinary people whose untimely deaths meant that they were not alive to see the fruits of a program that they were so instrumental in building: Adrian Gill and Stanley Hayes. Adrian Gill was the first Chairman of the TOGA SSG and got the TOGA program off the ground during a difficult time when no precedents were available to guide him. Stanley Hayes, through his untiring efforts, started building the observing system that would ultimately become the TOGA Tropical Ocean-Atmosphere (TAO) Array, a very crucial part of the TOGA Observing System. They would be both amazed and proud to see what their efforts had achieved.

      Return to Curent Research