A multi-model study of atmosphere predictability in coupled ocean–atmosphere systems

Of great importance for guiding numerical weather and climate predictions, understanding predictability of the atmosphere in the ocean − atmosphere coupled system is the first and critical step to understand predictability of the Earth system. However, previous predictability studies based on prefect model assumption usually depend on a certain model. Here we apply the predictability study with the Nonlinear Local Lyapunov Exponent and Attractor Radius to the products of multiple re-analyses and forecast models in several operational centers to realize general predictability of the atmosphere in the Earth system. We first investigated the predictability characteristics of the atmosphere in NCEP, ECMWF and UKMO coupled systems and some of their uncoupled counterparts and other uncoupled systems. Although the ECMWF Integrated Forecast System shows higher skills in geopotential height over the tropics, there is no certain model providing the most precise forecast for all variables on all levels and the multi-model ensemble not always outperforms a single model. Improved low-frequency signals from the air − sea and stratosphere − troposphere interactions that extend predictability of the atmosphere in coupled system suggests the significance of air − sea coupling and stratosphere simulation in practical forecast development, although uncertainties exist in the model representation for physical processes in air − sea interactions and upper troposphere. These inspire further exploration on predictability of ocean and stratosphere as well as sea − ice and land processes to advance our understanding of interactions of Earth system components, thus enhancing weather − climate prediction skills.

     

Predictability of the subtropical dipole modes in a coupled ocean–atmosphere model

Predictability of the subtropical dipole modes is assessed using the SINTEX-F coupled model. Despite the known difficulty in predicting subtropical climate due to large internal variability of the atmosphere and weak ocean–atmosphere coupling, it is shown for the first time that the coupled model can successfully predict the South Atlantic Subtropical Dipole (SASD) 1 season ahead, and the prediction skill is better than the persistence in all the 1–12 month lead hindcast experiments. There is a prediction barrier in austral winter due to the seasonal phase locking of the SASD to austral summer. The prediction skill is lower for the Indian Ocean Subtropical Dipole (IOSD) than for the SASD, and only slightly better than the persistence till 6-month lead because of the low predictability of the sea surface temperature anomaly in its southwestern pole. However, for some strong IOSD events in the last three decades, the model can predict them 1 season ahead. The co-occurrence of the negative SASD and IOSD in 1997/1998 austral summer can be predicted from July 1st of 1997. This is because the negative sea level pressure anomalies over the South Atlantic and the southern Indian Ocean in September–October (November–December) that trigger the occurrence of the negative SASD and IOSD are related to the well predicted tropical Indian Ocean Dipole (El Niño/Southern Oscillation). Owing to the overall good performances of the SINTEX-F model in predicting the SASD, some strong IOSD, and El Niño/Southern Oscillation, the prediction skill of the southern African summer precipitation is high in the SINTEX-F model.     

Uncertainty in the ocean–atmosphere feedbacks associated with ENSO in the reanalysis products

The evolution of El Ni?o-Southern Oscillation (ENSO) variability can be characterized by various ocean–atmosphere feedbacks, for example, the influence of ENSO related sea surface temperature (SST) variability on the low-level wind and surface heat fluxes in the equatorial tropical Pacific, which in turn affects the evolution of the SST. An analysis of these feedbacks requires physically consistent observational data sets. Availability of various reanalysis data sets produced during the last 15?years provides such an opportunity. A consolidated estimate of ocean surface fluxes based on multiple reanalyses also helps understand biases in ENSO predictions and simulations from climate models. In this paper, the intensity and the spatial structure of ocean–atmosphere feedback terms (precipitation, surface wind stress, and ocean surface heat flux) associated with ENSO are evaluated for six different reanalysis products. The analysis provides an estimate for the feedback terms that could be used for model validation studies. The analysis includes the robustness of the estimate across different reanalyses. Results show that one of the “coupled” reanalysis among the six investigated is closer to the ensemble mean of the results, suggesting that the coupled data assimilation may have the potential to better capture the overall atmosphere–ocean feedback processes associated with ENSO than the uncoupled ones.     

Impact of initialization procedures on the predictive skill of a coupled ocean–atmosphere model

The sensitivity of the predictive skill of a decadal climate prediction system is investigated with respect to details of the initialization procedure. For this purpose, the coupled ocean–atmosphere UCLA/MITgcm climate model is initialized using the following three different initialization approaches: full state initialization (FSI), anomaly initialization (AI) and FSI employing heat flux and freshwater flux corrections (FC). The ocean initial conditions are provided by the German contribution to Estimating the Circulation and Climate of the Ocean state estimate (GECCO project), from which ensembles of decadal hindcasts are initialized every 5 years from 1961 to 2001. The predictive skill for sea surface temperature (SST), sea surface height (SSH) and the Atlantic meridional overturning circulation (AMOC) is assessed against the GECCO synthesis. In regions with a deep mixed layer the predictive skill for SST anomalies remains significant for up to a decade in the FC experiment. By contrast, FSI shows less persistent skill in the North Atlantic and AI does not show high skill in the extratropical Southern Hemisphere, but appears to be more skillful in the tropics. In the extratropics, the improved skill is related to the ability of the FC initialization method to better represent the mixed layer depth, and the highest skill occurs during wintertime. The correlation skill for the spatially averaged North Atlantic SSH hindcasts remains significant up to a decade only for FC. The North Atlantic MOC initialized hindcasts show high correlation values in the first pentad while correlation remains significant in the following pentad too for FSI and FC. Overall, for the current setup, the FC approach appears to lead to the best results, followed by the FSI and AI procedures.     

Isolating mesoscale coupled ocean–atmosphere interactions in the Kuroshio Extension region

The Kuroshio Extension region is characterized by energetic oceanic mesoscale and frontal variability that alters the air–sea fluxes that can influence large-scale climate variability in the North Pacific. We investigate this mesoscale air-sea coupling using a regional eddy-resolving coupled ocean–atmosphere (OA) model that downscales the observed large-scale climate variability from 2001 to 2007. The model simulates many aspects of the observed seasonal cycle of OA coupling strength for both momentum and turbulent heat fluxes. We introduce a new modeling approach to study the scale-dependence of two well-known mechanisms for the surface wind response to mesoscale sea surface temperatures (SSTs), namely, the ‘vertical mixing mechanism’ (VMM) and the ‘pressure adjustment mechanism’ (PAM). We compare the fully coupled model to the same model with an online, 2-D spatial smoother applied to remove the mesoscale SST field felt by the atmosphere. Both VMM and PAM are found to be active during the strong wintertime peak seen in the coupling strength in both the model and observations. For VMM, large-scale SST gradients surprisingly generate coupling between downwind SST gradient and wind stress divergence that is often stronger than the coupling on the mesoscale, indicating their joint importance in OA interaction in this region. In contrast, VMM coupling between crosswind SST gradient and wind stress curl occurs only on the mesoscale, and not over large-scale SST gradients, indicating the essential role of the ocean mesocale. For PAM, the model results indicate that coupling between the Laplacian of sea level pressure and surface wind convergence occurs for both mesoscale and large-scale processes, but inclusion of the mesoscale roughly doubles the coupling strength. Coupling between latent heat flux and SST is found to be significant throughout the entire seasonal cycle in both fully coupled mode and large-scale coupled mode, with peak coupling during winter months. The atmospheric response to the oceanic mesoscale SST is also studied by comparing the fully coupled run to an uncoupled atmospheric model forced with smoothed SST prescribed from the coupled run. Precipitation anomalies are found to be forced by surface wind convergence patterns that are driven by mesoscale SST gradients, indicating the importance of the ocean forcing the atmosphere at this scale.     

Dynamics and predictability of a low-order wind-driven ocean–atmosphere coupled model

The dynamics of a low-order coupled wind-driven ocean–atmosphere system is investigated with emphasis on its predictability properties. The low-order coupled deterministic system is composed of a baroclinic atmosphere for which 12 dominant dynamical modes are only retained (Charney and Straus in J Atmos Sci 37:1157–1176, 1980) and a wind-driven, quasi-geostrophic and reduced-gravity shallow ocean whose field is truncated to four dominant modes able to reproduce the large scale oceanic gyres (Pierini in J Phys Oceanogr 41:1585–1604, 2011). The two models are coupled through mechanical forcings only. The analysis of its dynamics reveals first that under aperiodic atmospheric forcings only dominant single gyres (clockwise or counterclockwise) appear, while for periodic atmospheric solutions the double gyres emerge. In the present model domain setting context, this feature is related to the level of truncation of the atmospheric fields, as indicated by a preliminary analysis of the impact of higher wavenumber (“synoptic” scale) modes on the development of oceanic gyres. In the latter case, double gyres appear in the presence of a chaotic atmosphere. Second the dynamical quantities characterizing the short-term predictability (Lyapunov exponents, Lyapunov dimension, Kolmogorov–Sinaï (KS) entropy) displays a complex dependence as a function of the key parameters of the system, namely the coupling strength and the external thermal forcing. In particular, the KS-entropy is increasing as a function of the coupling in most of the experiments, implying an increase of the rate of loss of information about the localization of the system on its attractor. Finally the dynamics of the error is explored and indicates, in particular, a rich variety of short term behaviors of the error in the atmosphere depending on the (relative) amplitude of the initial error affecting the ocean, from polynomial (at ² + bt ³ + ct ⁴) up to exponential-like evolutions. These features are explained and analyzed in the light of the recent findings on error growth (Nicolis et al. in J Atmos Sci 66:766–778, 2009).     

Current status of ENSO prediction skill in coupled ocean–atmosphere models

The overall skill of ENSO prediction in retrospective forecasts made with ten different coupled GCMs is investigated. The coupled GCM datasets of the APCC/CliPAS and DEMETER projects are used for four seasons in the common 22 years from 1980 to 2001. As a baseline, a dynamic-statistical SST forecast and persistence are compared. Our study focuses on the tropical Pacific SST, especially by analyzing the NINO34 index. In coupled models, the accuracy of the simulated variability is related to the accuracy of the simulated mean state. Almost all models have problems in simulating the mean and mean annual cycle of SST, in spite of the positive influence of realistic initial conditions. As a result, the simulation of the interannual SST variability is also far from perfect in most coupled models. With increasing lead time, this discrepancy gets worse. As one measure of forecast skill, the tier-1 multi-model ensemble (MME) forecasts of NINO3.4 SST have an anomaly correlation coefficient of 0.86 at the month 6. This is higher than that of any individual model as well as both forecasts based on persistence and those made with the dynamic-statistical model. The forecast skill of individual models and the MME depends strongly on season, ENSO phase, and ENSO intensity. A stronger El Niño is better predicted. The growth phases of both the warm and cold events are better predicted than the corresponding decaying phases. ENSO-neutral periods are far worse predicted than warm or cold events. The skill of forecasts that start in February or May drops faster than that of forecasts that start in August or November. This behavior, often termed the spring predictability barrier, is in part because predictions starting from February or May contain more events in the decaying phase of ENSO.     

Evaluation of coupled ocean–atmosphere simulations of the mid-Holocene using palaeovegetation data from the northern hemisphere extratropics

We have used the BIOME4 biogeography–biochemistry model and comparison with palaeovegetation data to evaluate the response of six ocean–atmosphere general circulation models to mid-Holocene changes in orbital forcing in the mid- to high-latitudes of the northern hemisphere. All the models produce: (a) a northward shift of the northern limit of boreal forest, in response to simulated summer warming in high-latitudes. The northward shift is markedly asymmetric, with larger shifts in Eurasia than in North America; (b) an expansion of xerophytic vegetation in mid-continental North America and Eurasia, in response to increased temperatures during the growing season; (c) a northward expansion of temperate forests in eastern North America, in response to simulated winter warming. The northward shift of the northern limit of boreal forest and the northward expansion of temperate forests in North America are supported by palaeovegetation data. The expansion of xerophytic vegetation in mid-continental North America is consistent with palaeodata, although the extent may be over-estimated. The simulated expansion of xerophytic vegetation in Eurasia is not supported by the data. Analysis of an asynchronous coupling of one model to an equilibrium-vegetation model suggests vegetation feedback exacerbates this mid-continental drying and produces conditions more unlike the observations. Not all features of the simulations are robust: some models produce winter warming over Europe while others produce winter cooling. As a result, some models show a northward shift of temperate forests (consistent with, though less marked than, the expansion shown by data) and others produce a reduction in temperate forests. Elucidation of the cause of such differences is a focus of the current phase of the Palaeoclimate Modelling Intercomparison Project.     

Projected climate change scenario over California by a regional ocean–atmosphere coupled model system

This study examines a future climate change scenario over California in a 10-km coupled regional downscaling system of the Regional Spectral Model for the atmosphere and the Regional Ocean Modeling System for the ocean forced by the global Community Climate System Model version 3.0 (CCSM3). In summer, the coupled and uncoupled downscaled experiments capture the warming trend of surface air temperature, consistent with the driving CCSM3 forcing. However, the surface warming change along the California coast is weaker in the coupled downscaled experiment than it is in the uncoupled downscaling. Atmospheric cooling due to upwelling along the coast commonly appears in both the present and future climates, but the effect of upwelling is not fully compensated for by the projected large-scale warming in the coupled downscaling experiment. The projected change of extreme warm events is quite different between the coupled and uncoupled downscaling experiments, with the former projecting a more moderate change. The projected future change in precipitation is not significantly different between coupled and uncoupled downscaling. Both the coupled and uncoupled downscaling integrations predict increased onshore sea breeze change in summer daytime and reduced offshore land breeze change in summer nighttime along the coast from the Bay area to Point Conception. Compared to the simulation of present climate, the coupled and uncoupled downscaling experiments predict 17.5 % and 27.5 % fewer Catalina eddy hours in future climate respectively.     

Upper-bound general circulation of coupled ocean–atmosphere: Part 1. Atmosphere

We consider the general atmospheric circulation within the deductive framework of our climate theory. The preceding three parts of this theory have reduced the troposphere to the tropical and polar air masses and determined their temperature and the surface latitude of their dividing boundary, which provide the prior thermal constraint for the present dynamical derivation. Drawing upon its similar material conservation as the thermal property, the (columnar) potential vorticity (PV) is assumed homogenized as well in air masses, which moreover has a zero tropical value owing to the hemispheric symmetry. Inverting this PV field produces an upper-bound zonal wind that resembles the prevailing wind, suggesting that the latter may be explained as the maximum macroscopic motion extractable by random eddies – within the confine of the thermal differentiation.With the polar front determined in conjunction with the zonal wind, the approximate leveling of the isobars at the surface and high aloft specifies the tropopause, which is colder and higher in the tropics than in the polar region. The zonal wind drives the meridional circulation via the Ekman dynamics, and the preeminence of the Hadley cell stems from the singular Ekman convergence at the equator that allows it to supply the upward mass flux in the ITCZ demanded by the global energy balance.     

Interannual variations of the boreal summer intraseasonal variability predicted by ten atmosphere–ocean coupled models

The reproducibility of boreal summer intraseasonal variability (ISV) and its interannual variation by dynamical models are assessed through diagnosing 21-year retrospective forecasts from ten state-of-the-art ocean–atmosphere coupled prediction models. To facilitate the assessment, we have defined the strength of ISV activity by the standard deviation of 20–90 days filtered precipitation during the boreal summer of each year. The observed climatological ISV activity exhibits its largest values over the western North Pacific and Indian monsoon regions. The notable interannual variation of ISV activity is found primarily over the western North Pacific in observation while most models have the largest variability over the central tropical Pacific and exhibit a wide range of variability in spatial patterns that are different from observation. Although the models have large systematic biases in spatial pattern of dominant variability, the leading EOF modes of the ISV activity in the models are closely linked to the models’ El Nino-Southern Oscillation (ENSO), which is a feature that resembles the observed ISV and ENSO relationship. The ENSO-induced easterly vertical shear anomalies in the western and central tropical Pacific, where the summer mean vertical wind shear is weak, result in ENSO-related changes of ISV activity in both observation and models. It is found that the principal components of the predicted dominant modes of ISV activity fluctuate in a very similar way with observed ones. The model biases in the dominant modes are systematic and related to the external SST forcing. Thus the statistical correction method of this study based on singular value decomposition is capable of removing a large portion of the systematic errors in the predicted spatial patterns. The 21-year-averaged pattern correlation skill increases from 0.25 to 0.65 over the entire Asian monsoon region after applying the bias correction method to the multi-model ensemble mean prediction.     

Simulation of the Atlantic meridional overturning circulation in an atmosphere–ocean global coupled model. Part I: a mechanism governing the variability of ocean convection in a preindustrial experiment

A preindustrial climate experiment was conducted with the third version of the CNRM global atmosphere–ocean–sea ice coupled model (CNRM-CM3) for the Intergovernmental Panel on Climate Change Fourth Assessment Report (IPCC AR4). This experiment is used to investigate the main physical processes involved in the variability of the North Atlantic ocean convection and the induced variability of the Atlantic meridional overturning circulation (MOC). Three ocean convection sites are simulated, in the Labrador, Irminger and Greenland–Iceland–Norwegian (GIN) Seas in agreement with observations. A mechanism linking the variability of the Arctic sea ice cover and convection in the GIN Seas is highlighted. Contrary to previous suggested mechanisms, in CNRM-CM3 the latter is not modulated by the variability of freshwater export through Fram Strait. Instead, the variability of convection is mainly driven by the variability of the sea ice edge position in the Greenland Sea. In this area, the surface freshwater balance is dominated by the freshwater input due to the melting of sea ice. The ice edge position is modulated either by northwestward geostrophic current anomalies or by an intensification of northerly winds. In the model, stronger than average northerly winds force simultaneous intense convective events in the Irminger and GIN Seas. Convection interacts with the thermohaline circulation on timescales of 5–10 years, which translates into MOC anomalies propagating southward from the convection sites.     

Impact of a realistic river routing in coupled ocean–atmosphere simulations of the Last Glacial Maximum climate

The presence of large ice sheets over North America and North Europe at the Last Glacial Maximum (LGM) strongly impacted Northern hemisphere river pathways. Despite the fact that such changes may significantly alter the freshwater input to the ocean, modified surface hydrology has never been accounted for in coupled ocean–atmosphere general circulation model simulations of the LGM climate. To reconstruct the LGM river routing, we use the ICE-5G LGM topography. Because of the uncertainties in the extent of the Fennoscandian ice sheet in the Eastern part of the Kara Sea, we consider two more realistic river routing scenarios. The first scenario is characterised by the presence of an ice dammed lake south of the Fennoscandian ice sheet, and corresponds to the ICE-5G topography. This lake is fed by the Ob and Yenisei rivers. In the second scenario, both these rivers flow directly into the Arctic Ocean, which is more consistent with the latest QUEEN ice sheet margin reconstructions. We study the impact of these changes on the LGM climate as simulated by the IPSL_CM4 model and focus on the overturning thermohaline circulation. A comparison with a classical LGM simulation performed using the same model and modern river basins as designed in the PMIP2 exercise leads to the following conclusions: (1) The discharge into the North Atlantic Ocean is increased by 2,000 m³/s between 38° and 54°N in both simulations that contain LGM river routing, compared to the classical LGM experiment. (2) The ice dammed lake is shown to have a weak impact, relative to the classical simulation, both in terms of climate and ocean circulation. (3) In contrast, the North Atlantic deep convection and meridional overturning are weaker than during the classical LGM run if the Ob and Yenisei rivers flow directly into the Arctic Ocean. The total discharge into the Arctic Ocean is increased by 31,000 m³/s, relative to the classical LGM simulation. Consequentially, northward ocean heat transport is weaker, and sea ice more extensive, in better agreement with existing proxy data.     

Bi-decadal variability excited in the coupled ocean–atmosphere system by strong tropical volcanic eruptions

Decadal and bi-decadal climate responses to tropical strong volcanic eruptions (SVEs) are inspected in an ensemble simulation covering the last millennium based on the Max Planck Institute—Earth system model. An unprecedentedly large collection of pre-industrial SVEs (up to 45) producing a peak annual-average top-of-atmosphere radiative perturbation larger than ?1.5 Wm^?2 is investigated by composite analysis. Post-eruption oceanic and atmospheric anomalies coherently describe a fluctuation in the coupled ocean–atmosphere system with an average length of 20–25 years. The study provides a new physically consistent theoretical framework to interpret decadal Northern Hemisphere (NH) regional winter climates variability during the last millennium. The fluctuation particularly involves interactions between the Atlantic meridional overturning circulation and the North Atlantic gyre circulation closely linked to the state of the winter North Atlantic Oscillation. It is characterized by major distinctive details. Among them, the most prominent are: (a) a strong signal amplification in the Arctic region which allows for a sustained strengthened teleconnection between the North Pacific and the North Atlantic during the first post-eruption decade and which entails important implications from oceanic heat transport and from post-eruption sea ice dynamics, and (b) an anomalous surface winter warming emerging over the Scandinavian/Western Russian region around 10–12 years after a major eruption. The simulated long-term climate response to SVEs depends, to some extent, on background conditions. Consequently, ensemble simulations spanning different phases of background multidecadal and longer climate variability are necessary to constrain the range of possible post-eruption decadal evolution of NH regional winter climates.     

Seasonal prediction of global sea level anomalies using an ocean–atmosphere dynamical model

Advanced warning of extreme sea level events is an invaluable tool for coastal communities, allowing the implementation of management policies and strategies to minimise loss of life and infrastructure damage. This study is an initial attempt to apply a dynamical coupled ocean–atmosphere model to the prediction of seasonal sea level anomalies (SLA) globally for up to 7 months in advance. We assess the ability of the Australian Bureau of Meteorology’s operational seasonal dynamical forecast system, the Predictive Ocean Atmosphere Model for Australia (POAMA), to predict seasonal SLA, using gridded satellite altimeter observation-based analyses over the period 1993–2010 and model reanalysis over 1981–2010. Hindcasts from POAMA are based on a 33-member ensemble of seasonal forecasts that are initialised once per month for the period 1981–2010. Our results show POAMA demonstrates high skill in the equatorial Pacific basin and consistently exhibits more skill globally than a forecast based on persistence. Model predictability estimates indicate there is scope for improvement in the higher latitudes and in the Atlantic and Southern Oceans. Most characteristics of the asymmetric SLA fields generated by El-Nino/La Nina events are well represented by POAMA, although the forecast amplitude weakens with increasing lead-time.     

A highly nonlinear coupled mode of decadal variability in a mid-latitude ocean–atmosphere model

This study examines mid-latitude climate variability in a model that couples turbulent oceanic and atmospheric flows through an active oceanic mixed layer. Intrinsic ocean dynamics of the inertial recirculation regions combines with nonlinear atmospheric sensitivity to sea-surface temperature (SST) anomalies to play a dominant role in the variability of the coupled system.Intrinsic low-frequency variability arises in the model atmosphere; when run in a stand-alone mode, it is characterized by irregular transitions between preferred high-latitude and less frequent low-latitude zonal-flow states. When the atmosphere is coupled to the ocean, the low-latitude state occurrences exhibit a statistically significant signal in a broad 5–15-year band. A similar signal is found in the time series of the model ocean's energy in this coupled simulation. Accompanying uncoupled ocean-only and atmosphere-only integrations are characterized by a decrease in the decadal-band variability, relative to the coupled integration; their spectra are indistinguishable from a red spectrum.The time scale of the coupled interdecadal oscillation is set by the nonlinear adjustment of the ocean's inertial recirculations to the high-latitude and low-latitude atmospheric forcing regimes. This adjustment involves, in turn, SST changes resulting in long-term ocean–atmosphere heat-flux anomalies that induce the atmospheric regime transitions.     

Impact of diurnal atmosphere–ocean coupling on tropical climate simulations using a coupled GCM

The impacts of diurnal atmosphere–ocean (air–sea) coupling on tropical climate simulations are investigated using the SNU coupled GCM. To investigate the effect of the atmospheric and oceanic diurnal cycles on a climate simulation, a 1-day air–sea coupling interval experiment is compared to a 2-h coupling experiment. As previous studies have suggested, cold temperature biases over equatorial western Pacific regions are significantly reduced when diurnal air–sea coupling strategy is implemented. This warming is initiated by diurnal rectification and amplified further by the air–sea coupled feedbacks. In addition to its effect on the mean climatology, the diurnal coupling has also a distinctive impact on the amplitude of the El Nino-Southern Oscillation (ENSO). It is demonstrated that a weakening of the ENSO magnitude is caused by reduced (increased) surface net heat fluxes into the ocean during El Nino (La Nina) events. Primarily, decreased (increased) incoming shortwave radiation during El Nino (La Nina) due to cloud shading is responsible for the net heat fluxes associated with ENSO.     

On the mechanisms in a tropical ocean–global atmosphere coupled general circulation model. Part I: mean state and the seasonal cycle

 The mechanisms responsible for the mean state and the seasonal and interannual variations of the coupled tropical Pacific-global atmosphere system are investigated by analyzing a thirty year simulation, where the LMD global atmospheric model and the LODYC tropical Pacific model are coupled using the delocalized physics method. No flux correction is needed over the tropical region. The coupled model reaches its regime state roughly after one year of integration in spite of the fact that the ocean is initialized from rest. Departures from the mean state are characterized by oscillations with dominant periodicites at annual, biennial and quadriennial time scales. In our model, equatorial sea surface temperature and wind stress fluctuations evolved in phase. In the Central Pacific during boreal autumn, the sea surface temperature is cold, the wind stress is strong, and the Inter Tropical Convergence Zone (ITCZ) is shifted northwards. The northward shift of the ITCZ enhances atmospheric and oceanic subsidence between the equator and the latitude of organized convention. In turn, the stronger oceanic subsidence reinforces equatorward convergence of water masses at the thermocline depth which, being not balanced by equatorial upwelling, deepens the equatorial thermocline. An equivalent view is that the deepening of the thermocline proceeds from the weakening of the meridional draining of near-surface equatorial waters. The inverse picture prevails during spring, when the equatorial sea surface temperatures are warm. Thus temperature anomalies tend to appear at the thermocline level, in phase opposition to the surface conditions. These subsurface temperature fluctuations propagate from the Central Pacific eastwards along the thermocline; when reaching the surface in the Eastern Pacific, they trigger the reversal of sea surface temperature anomalies. The whole oscillation is synchronized by the apparent meridional motion of the sun, through the seasonal oscillation of the ITCZ. This possible mechanism is partly supported by the observed seasonal reversal of vorticity between the equator and the ITCZ, and by observational evidence of eastward propagating subsurface temperature anomalies at the thermocline level. Received: 7 April 1997 / Accepted: 15 July 1998