Search

Glider observations were essential components of the observational program in the Autonomous Ocean Sampling Network (AOSN-II) experiment in the Monterey Bay area during summer of 2003. This paper is focused on the impact of the assimilation of glider temperature and salinity observations on the Navy Coastal Ocean Model (NCOM) predictions of surface and subsurface properties. The modeling system consists of an implementation of the NCOM model using a curvilinear, orthogonal grid with 1–4 km resolution, with finest resolution around the bay. The model receives open boundary conditions from a regional (9 km resolution) NCOM implementation for the California Current System, and surface fluxes from the Coupled Ocean–Atmosphere Mesoscale Prediction System (COAMPS) atmospheric model at 3 km resolution. The data assimilation component of the system is a version of the Navy Coupled Ocean Data Assimilation (NCODA) system, which is used for assimilation of the glider data into the NCOM model of the Monterey Bay area. The NCODA is a fully 3D multivariate optimum interpolation system that produces simultaneous analyses of temperature, salinity, geopotential, and vector velocity. Assimilation of glider data improves the surface temperature at the mooring locations for the NCOM model hindcast and nowcasts, and for the short-range (1–1.5 days) forecasts. It is shown that it is critical to have accurate atmospheric forcing for more extended forecasts. Assimilation of glider data provided better agreement with independent observations (for example, with aircraft measured SSTs) of the model-predicted and observed spatial distributions of surface temperature and salinity. Mooring observations of subsurface temperature and salinity show sharp changes in the thermocline and halocline depths during transitions from upwelling to relaxation and vice versa. The non-assimilative run also shows these transitions in subsurface temperature; but they are not as well defined. For salinity, the non-assimilative run significantly differs from the observations. However, the glider data assimilating run is able to show comparable results with observations of thermocline as well as halocline depths during upwelling and relaxation events in the Monterey Bay area. It is also shown that during the relaxation of wind, the data assimilative run has higher value of subsurface velocity complex correlation with observations than the non-assimilative run.

This article is the second of a two-part paper on ELCIRC, an Eulerian-Lagrangian finite difference/finite volume model designed to simulate 3D baroclinic circulation across river-to-ocean scales. In part one (Zhang et al., 2004), we described the formulation of ELCIRC and assessed its baseline numerical skill. Here, we describe the application of ELCIRC within CORIE, a coastal margin observatory for the Columbia River estuary and plume. We first introduce the CORIE modeling system and its multiple modes of simulation, external forcings, observational controls, and automated products. We then focus on the evaluation of highly resolved, year-long ELCIRC simulations, using two variables (water level and salinity) to illustrate simulation quality and sensitivity to modeling choices. We show that, process-wise, simulations capture well important aspects of the response of estuarine and plume circulation to ocean, river, and atmospheric forcings. Quantitatively, water levels are robustly represented, while salinity intrusion and plume dynamics remain challenging. Our analysis highlights the benefits of conducting model evaluations over large time windows (months to years), to avoid significant localized biases. The robustness and computational efficiency of ELCIRC has proved invaluable in identifying and reducing non-algorithmic sources of errors, including parameterization (e.g., turbulence closure and stresses at the air-water interface) and external forcings (e.g., ocean conditions and atmospheric forcings).

Although it is nominally a tropical locale, the semiannual wind reversals associated with the Monsoon system of the Arabian Sea result annually in two distinct periods of elevated biological activity. While in both cases monsoonal forcing drives surface layer nutrient enrichment that supports increased rates of primary productivity, fundamentally different entrainment mechanisms are operating in summer (Southwest) and winter (Northeast) Monsoons. Moreover, the intervening intermonsoon periods, during which the region relaxes toward oligotrophic conditions more typical of tropical environments, provide a stark contrast to the dynamic biogeochemical activity of the monsoons. The resulting spatial and temporal variability is great and provides a significant challenge for ship-based surveys attempting to characterize the physical and biogeochemical environments of the region. This was especially true for expeditions in the pre-satellite era. Here, we present an overview of the dynamical response to seasonal monsoonal forcing and the characteristics of the physical environment that fundamentally drive regional biogeochemical variability. We then review past observations of the biological distributions that provided our initial insights into the pelagic system of the Arabian Sea. These evolved through the 1980s as additional methodologies, in particular the first synoptic ocean color distributions gathered by the Coastal Zone Color Scanner, became available. Through analyses of these observations and the first large-scale physical–biogeochemical modeling attempts, a pre-JGOFS understanding of the Arabian Sea emerged. During the 1990s, the in situ and remotely sensed observational databases were significantly extended by regional JGOFS activities and the onset of Sea-viewing Wide Field-of-View Sensor ocean color measurements. Analyses of these new data and coupled physical–biogeochemical models have already advanced our understanding and have led to either an amplification or revision of the pre-JGOFS paradigms. Our understanding of this complex and variable ocean region is still evolving. Nonetheless, we have a much better understanding of time–space variability of biogeochemical properties in the Arabian Sea and much deeper insights about the physical and biological factors that drive them, as well as a number of challenging new directions to pursue.

A high resolution, data assimilating ocean model of the Monterey Bay area (ICON model) is under development within the framework of the project “An Innovative Coastal-Ocean Observing Network” (ICON) sponsored by the National Oceanographic Partnership Program. The main objective of the ICON model development is demonstration of the capability of a high resolution model to track the major mesoscale ocean features in the Monterey Bay area when constrained by the measurements and nested within a regional larger-scale model. This paper focuses on the development of the major ICON model components, including grid generation and open boundary conditions, coupling with a larger scale, Pacific West Coast (PWC) model, atmospheric forcing etc. Impact of these components on the Model's predictive skills in reproducing major hydrographic conditions in the Monterey Bay area are analyzed. Comparisons between observations and the ICON model predictions with and without coupling to the PWC model, show that coupling with the regional model improves significantly both the correlation between the ICON model and observed ADCP currents, and the ICON model's skill in predicting the location and intensity of observed upwelling events. Analysis of the ICON model mixed layer depth predictions show that the ICON model tends to develop a thicker than observed mixed layer during the summer time, and while assimilation of sea surface temperature data is enough for development of observed thin mixed layer in the regional larger-scale model, the fine-resolution ICON model needs variable heat fluxes as surface boundary conditions for the accurate prediction of the vertical thermal structure. The paper targets researchers involved in high-resolution numerical modeling of coastal areas in which the dynamics are determined by the complex geometry of a coastline, variable bathymetry and by the influence of complex water masses from a complicated hydrographic system (such as the California Current system).

The evolution of the upper ocean in the strong seasonally forced Arabian Sea, as observed by a mooring deployed in 1994–1995, is investigated using the Naval Research Laboratory Layered Ocean Model (NLOM). Model simulations were sensitive to the choice of surface wind products used for forcing, and results are reported for simulations forced by monthly mean climatologies and 12 hourly 1994–1995 wind products from two operational atmospheric forecast models, the European Centre for Medium-Range Weather Forecast model and the Navy Operational Global Atmospheric Prediction System model of Fleet Numerical Meteorology and Oceanography Center (FNMOC). The NLOM yields the best prediction of sea surface temperature (SST) and mixed layer depth when using FNMOC forcing. Surface cooling is found to be responsible for the seasonal SST minimum during the NE monsoon. Heat advection is found to be important for supporting the surface cooling during the second half of the NE monsoon. Strong entrainment and appreciable advective cooling are responsible for the SST minimum of the SW monsoon. The NLOM wind experiments strongly suggest that thermal convection may be important in the central Arabian Sea during the winter months.

Velocity and property observations were made during February and March 1995 as part of the World Ocean Circulation Experiment (WOCE) Hydrographic Program (WHP) expedition in the Indian Ocean. The observed circulation in the upper 300 m of the ocean during the northeast monsoon is compared to the output of a high-resolution, 3-layer, nonlinear model forced by European Center for Medium Range Weather Forecasting (ECMWF) winds. The data identify several new features in the Bay of Bengal: a shelf-break coastal current in the northeast corner, an eddy and subsurface jet near the South Preparis Channel, and an eastward countercurrent extending from 80°E to the eastern boundary. The countercurrent transports high-salinity surface water from the northwest Indian Ocean into the bay, and separates the historically observed North Equatorial Current into two currents with separate sources and water properties. The data also detail the spatial structure of the South Equatorial Current and Countercurrent, and show the Equatorial Undercurrent in the eastern half of the Indian Ocean. The model compares well enough with the data to suggest that such realistic models may provide a useful temporal context for the WHP snapshot.

Accurate, year-long time series of winds, incoming shortwave and longwave radiation, air and sea temperatures, relative humidity, barometric pressure, and precipitation were collected from a surface mooring deployed off the coast of Oman along the climatological axis of the Findlater Jet from October 1994 to October 1995. Wind stress, heat flux, and freshwater flux were computed using bulk formulae. The Northeast Monsoon was characterized by steady but moderate winds, clear skies, relatively dry air, and two months, December and January, in which the ocean, on average, lost 45 W m-2 to the atmosphere. The Southwest Monsoon had strong winds, cloudy skies, and moist air. Because of reduced latent and longwave heat loss, it was accompanied by sustained oceanic heat gain, with the strongest monthly mean warming, 147 W m-2, in August. Large differences are found between the observations and older climatologies. Recent climatologies agree better with the observations. The means of the Southampton Oceanography Center climatology for 1980–1995 are close to the buoy monthly means. Monthly means from that climatology show that 1994–1995 was in general a typical year, with surface meteorology and air–sea fluxes within one standard deviation of the long term means. Concurrent data from the NCEP, ECMWF, and FNMOC show that the models provide realistic surface winds. FNMOC winds show that the timing and character of the onset of the Southwest Monsoon in 1995 differed from 1994 and 1996 when variability within one month is resolved. The models fail to replicate other observed surface meteorology and to produce realistic heat fluxes. Annual and monsoonal mean net heat fluxes from the models differed from those of the buoy by 50 to 80 W m-2. Because of these differences, some care is warranted in selecting and using air-sea flux fields in studies of the Arabian Sea.

[1] The Oregon Coastal Transition Zone (OCTZ) extends several hundred kilometers offshore where shelf flows interact with the northern California Current. A primitive-equation numerical ocean model is used to study the upwelling circulation in this region from 1 May to 1 November 2001. This OCTZ model obtains initial and boundary conditions from a larger-scale model of the California Current System and forcing from a regional atmospheric model product. The model results are compared with extensive in situ and remotely sensed data, and the model is found to provide a realistic representation of flows both over the shelf and in the broader OCTZ. Simulation of coastal sea level and shelf currents over the complex topography of the central Oregon coast is improved quantitatively relative to previous regional models. A particularly significant qualitative improvement is realistic representation of coastal jet separation and eddy formation offshore of Cape Blanco. Three-dimensional Lagrangian analysis of water parcel displacement shows that the surface waters inshore of the separated jet are upwelled from near the bottom along the shelf as far north as 45.5°N. A large eddy, which incorporates some of this upwelled water and carries it farther westward, forms offshore in the late summer. Ensemble simulations show a distinction between the strongly deterministic response to wind forcing over the shelf and the more unstable, less predictable jet separation and offshore eddy formation processes in the region near Cape Blanco.

Sp. Iss. SI; During the Global Ocean Data Assimilation Experiment (GODAE), numerical modeling and prediction in coastal and shelf seas benefited from development of state-of-the-art, data-assimilative, and data-validated large-scale models that can supply initial and boundary conditions to nested domains. Rather than attempting an exhaustive synthesis, this article illustrates the progress in coastal ocean modeling and prediction made possible by GODAE, either directly by providing estimates, or more subtly by rendering coastal forecasting more feasible and its applications more obvious.

[1] Three years of Sea-viewing Wide Field-of-view Sensor (SeaWiFS) ocean color data were combined with three-dimensional thermal fields generated by the U.S. Navy's Modular Ocean Data Assimilation System (MODAS) in order to examine the interdependencies between bio-optical fields and their relationship to seasonal and mesoscale changes in upper ocean thermal structure. The combined data set suggests that the oceanic boundary layer within the Gulf of Mexico may be broadly defined by two seasonally occurring bio-thermal periods. A winter mixing period, characterized by net heat losses to the atmosphere, deepening of the isothermal layer depth, and annual maxima of satellite-estimated colored detrital matter (CDM) absorption coefficients and surface pigment concentration, was followed by a thermally stratified period characterized by net surface ocean heating, reduced isothermal layer depths, and annual minima in surface bio-optical fields. Variability in the interdependencies of ocean color products was used to diagnose an attendant shift in the size-structure of surface phytoplankton communities as well as identify CDM as the constituent responsible for the majority of blue-light absorption in Gulf of Mexico surface waters. The mesoscale circulation, as resolved by MODAS thermal fields into cold and warm-core eddies, appears to significantly modulate the seasonal bio-optical cycle of CDM absorption and surface pigment concentration. An empirical model was developed to describe CDM absorption as a function of upper ocean thermal energy. The model accounted for nearly half the variance in the satellite-estimate of this bio-optical variable. Large mismatches between the model and satellite data implied episodes of shelf water export to the deep Gulf of Mexico.

Prediction System (COAMPS TM )(COAMPS is a registered trademark of the Naval Research Laboratory). Issues investigated in this study are: NCOM-based model simulations of upwelling and relaxation events, coupling to COAMPS, use of sigma versus hybrid (sigma-z) vertical grids, and coupling with a larger-scale model on the open boundaries. The NCOM simulations were able to reproduce the observed sequence of the upwelling and relaxation events, which can be attributed, in part, to the good agreement between the observed and COAMPS winds. Comparisons with the mooring observations show that COAMPS overestimates shortwave radiation values, which makes the NCOM modeled SSTs too warm in comparison with observations. The NCOM runs forced with different resolution atmospheric forcing (3 versus 9 km) do not show significant differences in the predicted SSTs and mixed-layer depths at the mooring locations. At the same time, during the extended upwelling event, the model runs forced with 3 and 9 km resolution COAMPS fields show differences in the surface circulation patterns, which are the most distinct in the southern portion of the model domain. The model run with 9-km forcing develops a northward flow along the coast, which is not present in the run with 3-km forcing and in observations (for example, HF radar-derived radials). Comparison of the wind patterns of the 3- and 9-km products shows a weakening of the 9-km wind stress along the southern coast of the NCOM model domain, which is responsible for the development of the artificial northward flow in the NCOM run with 9-km forcing.

The specific objectives of this research are to determine the impact on coastal ocean circulation models of open ocean boundary conditions from Global Ocean Data Assimilation Experiment (GODAE) Pacific Ocean models, and to address closely related issues of uncertainty and predictability in coastal ocean models.

The Lombok Strait, a gap in the lower Indonesian Archipelago second in cross sectional area only to the Timor passages, provides a major pathway for the Pacific to Indian throughflow. A global reduced gravity model, corroborated by dynamic height climatology from the Generalized Digital Environmental Model, predicts annual mean sea levels 15-20 cm higher at the Pacific entrance to the Indonesian Seas than in the Indian Ocean south of the archipelago straits. Consistent with this regional pressure gradient, Pacific core layers of the Northern Subtropical Central Water and the North Pacific Intermediate Water are traced southward from the Makassar Strait into the Lombok Strait. Maps of temperature, salinity, and density distributions and sea surface dynamic heights in the Lombok Strait from January, June, and September 1985 also indicate a persistent southward flow of appreciable magnitude. Geostrophic speeds, however, are clearly too large by a factor of two or more. Current meter arrays in the north strait (January 1985 – March 1986) provide direct measurements of southward currents which persist through most of the year and are concentrated in the upper few hundred meters consistent with Wyrtki’s (1987) analysis of the regional pressure gradient. Maximum sustained speeds of over 70 cm/sec occur from July to September with a long period of weak currents from mid-October 1985 through January 1986. Tropical cyclones in the Timor Sea (December-April) force strong northward flow reversals which can persist for ten days. The wind-forced numerical model identifies the strong westward wind stresses in the Timor Sea during the southeast monsoon as the major cause of the annual cycle of current in the Strait.

[1] A 2-dimensional variational method is used to blend the satellite scatterometer measured (QuikSCAT) and regional mesoscale atmospheric model simulated (COAMPS) surface vector winds for coastal central California. The approach is distinct from existing methods in that it considers errors from both measurements and models. When compared with independent in situ observations, the blended wind product shows consistently higher correlation and smaller RMS errors than QuikSCAT or COAMPS winds. The proposed algorithm can be implemented over any part of the world ocean. It should be a valuable tool for describing small-scale atmospheric processes in coastal zones and for forcing high-resolution coastal ocean models.

[1] This paper introduces and evaluates a new atmospheric model reanalysis product for the U.S. west coast from the Navy's operational Coupled Ocean/Atmosphere Mesoscale Prediction System (COAMPS™) implemented on a triply nested 81/27/9 km grid. Wind measurements from Monterey Bay Aquarium Research Institute (MBARI) moorings are used to evaluate a two year time series of the model output from October 1998 to October 2000. The model accurately reproduced both the landward decrease in the large-scale wind field and the shoreward intensification of the diurnal cycle during the summer season. This new reanalysis product will be required for accurate high resolution (km) coastal (0–10 km) simulations of the U.S. west coast by ocean models.

Monsoons over the Arabian Sea—the oceanic basin that separates the Arabian peninsula from the Indian subcontinent—follow seasonal cycles, reversing directions twice a year, in summer and winter. Rather than spreading across the expanse of the sea, the southwest (summer) monsoon is often concentrated into a jet over the central Arabian Sea. Evidence suggests that variations in wind stress force substantial upwelling in the ocean to the west of the jet, and weaker upwelling or even downwelling to the east. This upwelling provides nutrients to the euphotic zone and enhances biological productivity.

Numerical simulations of the remotely forced ocean response to westerly wind bursts prior to and during the 1991 - 1992 El Nino are examined; the models are forced by wind stress from the U.S. Navy's atmospheric global operational analysis/forecast system. The study focuses on (1) the relative response of the first and second internal modes to a single episode of westerly bursts; (2) the role of westerly bursts in producing the eastern Pacific sea level variations from October 1990 to February 1992; and (3) the relative importance of the remotely forced sea level responses generated by central and western Pacific wind anomalies. The simulation use the Naval Research Laboratory global multilayer formulation; the suite of experiments includes hydrodynamic simulations that use both one- and three- active-layer reduced gravity models as well as an experiment that also includes thermodynamic effects. The models are executed from January 1, 1990, to March 1, 1992, a period that includes 10 significant westerly wind bursts or burst clusters and the 1991 - 1992 El Nino event. The numerical experiments reveal an ability to accurately simulate the eastern Pacific sea level variations during this period. In response to a single burst, the three-layer hydrodynamic simulation reveals that the second internal mode Kelvin wave yields a sea level change at the eastern boundary that is approximately one third that of the first mode and a surface velocity signature that is equivalent to the first mode. During the onset of the El Nino event the inclusion of higher modes also produces a more realistic representation of the observed eastern boundary sea level signal. Furthermore, by comparing the model response to particular wind bursts with the observed sea level at Baltra, Galapagos, a value of 2.5 - 2.6 m/s is suggested as the most appropriate mean speed for the first internal mode Kelvin wave during the onset phase. A scenario for the onset of the 1991 - 1992 El Nino is given.

The dynamics and influence of the connection between the Pacific and Indian Oceans in the Indonesian Archipelago (PACIO) on the upper tropical Pacific and Indian Oceans are investigated using a reduced gravity global model. Two different model geometries are used; the first has an open PACIO region (open case) that permits normal transport, the second has a closed PACIO region (closed case) that allows no transport. The model is forced by either monthly 1980–1989 European Centre for Medium-Range Weather Forecasting wind stresses or by an imposed Rossby wave pulse. The average upper layer thickness of the Indian Ocean is significantly greater and more variable because of the influence of PACIO transport. The effect in the Pacific Ocean is much smaller, and there appears to be no appreciable effect of PACIO transport on interannual variability in the eastern tropical Pacific Ocean. RMS differences between the open and closed cases show areas of low-frequency (> ½ year) variability over most of the southern Indian Ocean, the western boundary of the Indian Ocean, the region of the Mindanao Eddy, and the eastern equatorial Pacific waveguide. Areas of significant high-frequency variability are the East African Coastal Current, the Mindanao Eddy region, the Coral Sea Basin, and in the eastern equatorial Pacific waveguide. The large effect on the southern Indian Ocean is linked to off-equatorial Rossby waves generated by transport through the PACIO region.

The seasonal formation of a large (500–800 km diameter) anticyclonic eddy in the upper 300–400 m of the eastern Arabian Sea during the northeast monsoon period (December-April) is indicated from hydrographic and satellite altimetry sea level observations, as well as from numerical model experiments. The center of the eddy circulation is approximately 10°N, 70°E, just to the west of the north-south Laccadive Island chain. In this paper the eddy is called the Laccadive High (LH). In some ways it is a mirrorlike counterpart to the Great Whirl, which develops during the southwest monsoon off the Somali coast (western Arabian Sea). The LH occurs at the same latitude but on the opposite side of the basin during the reversed monsoon. It is different from the Great Whirl, however, in its formation process, its intensity, and its decay. The hydrographic data obtained from surveys all during a single season give sufficiently close station spacing to allow reasonable contouring of the geopotential surfaces and of the properties within and around the LH region with minimum time aliasing. The Geosat altimeter record extends over 4 years, during which the seasonal variability of the LH indicates a dynamic relief of approximately 15–20 cm, which is in good agreement with the hydrographic observations. The altimetry time series also suggests a westward translation of the LH by January with a subsequent dissipation in midbasin. The model used is a wind-forced three-layer primitive equation model which depicts a LH in agreement with the timing, position, and amplitude of both the hydrographic and altimetric measurements. The numerical simulation includes a passive tracer located in the western Bay of Bengal; the western advection of the tracer around the south coasts of Sri Lanka and India in December and January is consistent with the appearance of low-salinity water observed to extend into the Arabian Sea during this period. The modeling studies suggest that both local and remote forcing are important in formation of the LH.

The relative effects of islands and topography on the seasonal circulation of the southern Indian Ocean are examined. The numerical simulations use a one-active layer reduced gravity model and a two-layer model with full bottom topography forced by the Hellerman and Rosenstein wind stress climatology. The reduced gravity experiments show that the absence of a seasonal cycle of transport off the north coast of Madagascar is not caused by the presence of islands to the east. The two-layer experiments indicate that the presence of abrupt topography associated with the Mascarene Ridge and the Providence Bank north of Madagascar decrease the annual barotropic signal of the North Madagascar Current and increase the seasonal variation of the Southeast Madagascar Current. Furthermore, the two-layer experiments predict the existence of a significant seasonal variation of transport north of the Saya de Malha Bank along the Mascarene Ridge; the signal (12–16 Sv in amplitude) is sufficiently large that the model results may be tested by future observations. Finally, the differences between the seasonal variation of the Indian and North Atlantic oceans are discussed.

El Nino may be defined oceanographically as a massive influx of warm water into the coastal region of Ecuador and Peru. We have tested the hypothesis that these rare events occur after a substantial reduction of the atmospheric trade winds over the central Pacific Ocean. An idealized nonlinear, two-layer, equatorial beta-plane ocean is spun-up with easterly winds for 50 days after which the wind is relaxed over several days. The relaxation of the wind initiates internal Kelvin wave fronts at both sides of the ocean at the equator. The eastern wave fronts propagate poleward and the western ones eastward. Internal Rossby waves are generated which propagate westward from the eastern boundary. As the Kelvin wave fronts move poleward along the eastern boundary, strong downwelling occurs and the coastal currents reverse direction and become poleward. The rapid downwelling and the sudden reversal of the coastal current are consistent with observations during El Nino. This downwelling is much more rapid ...

A common observation in major coastal upwelling regions is a reversal of the mean longshore current with depth during active upwelling. A numerical model and a simple analytical model applied to a two-layer ocean on a beta-plane indicate that this undercurrent cannot exist for upwelling off an east-west (zonal) coastline. The models are applied to the coastal upwelling occurring on the Campeche Bank during the period from March to September. The beta-plane solutions for upwelling along an east-west coastline resemble f-plane solutions for a meridional (N–S) coastline. In particular, the hypothesis explaining the undercurrent of Hurlburt and Thompson cannot hold for upwelling along a zonal coastline. The results in this paper form a new hypothesis which could be tested by measuring currents on the Campeche Bank during active upwelling.

The excitation of equatorial Rossby solitons during El Nino is studied numerically using a one-layer reduced-gravity formulation. The model basin - an idealization of the tropical Pacific - extends 15,000 km zonally and 3,000 km meridionally. Forcing is applied as a patch of zonal wind stress representing the relaxation of the equatorial easterlies preceding El Nino. It is shown that a relaxation of the equatorial winds generates internal Rossby solitons subsequent to the reflection of an equatorial Kelvin wave front from the eastern boundary. The meridional structure of the soliton is that of a first latitudinal mode Rossby wave. The simple El Nino simulations result in the excitation of two or three solitons separated by approximately 3,000 km. The results of this study suggest that internal Rossby solitons are most likely to be observed in the central and western Pacific during a major El Nino year.

The circulation of the western Indian Ocean is examined using a reduced-gravity model with one active layer and realistic basin geometry for the entire Indian Ocean north of 30°S. The Hellerman and Rosenstein monthly mean wind stress climatology is used to force the model. The numerical simulations reproduce the observed (Luyten and Roemmich, 1982) 26-day waves along the equator and the 50-day oscillations (Mysak and Mertz, 1984; Schott et al. 1988) between the equator and Madagascar. The 25- to 28-day oscillations of the model meridional velocity component agree with observed values of period, amplitude, wavelength, group velocity, and phase of the seasonal modulation. The model oscillations, which are excited in August and persist into February–March, are shown to be the result of Yanai waves generated between the western boundary and 50°E. During the southwest monsoon, the Yanai waves are initiated by a complex barotropic instability associated with the southern gyre. During the early stages of the northeast monsoon, the 26-day Yanai waves are generated by resonant forcing due to the intrusion into the equatorial waveguide of a standing, 800- to 900-km-wavelength meander of the eastward flow fed by the East African Coastal Current. Hence the simulation reveals that the 26-day oscillations in the equatorial Indian Ocean are excited by mechanisms significantly different than that believed to be responsible for the 20- to 30-day oscillations in the equatorial Atlantic and Pacific oceans. The numerical simulation also shows a 50-day oscillation between the equator and Madagascar west of 50°E. This periodicity is due to Rossby waves generated by a barotropic instability associated with the East African Coastal Current beginning about April each year. No evidence of the 50-day period oscillation is found in a corresponding linear simulation. Hence the barotropic instability of the oceanic currents in this region is offered as an alternative to direct wind forcing as the generating mechanism for the observed 40- to 60-day oscillations in the western Indian Ocean.