Publication

2004-06

English

http://archimer.ifremer.fr/doc/2004/publication-670.pdf (3.05 Mo)

2002 Elsevier Science Ltd. All rights reserved

Thierry Virginie, Treguier Anne-Marie, Mercier Herle

IFREMER, Ctr Brest, CNRS, UBO,Lab Phys Oceans, F-29280 Plouzane, France.

Ocean Modelling (1463-5003) (Elsevier), 2004-06 , Vol. 6 , N. 1 , P. 1-30

28

Mathematics-Computer Science, Physical Oceanography

Wind driven circulation, Seasonal variability, Equatorial waves

Using idealized and realistic high-resolution primitive equation models, we explore the annual and semi-annual period fluctuations in the deep equatorial Atlantic Ocean. The deep seasonal variability derives its energy from the equatorial zonal wind stress component and the wind energy reaches the deep layers by means of vertically propagating Kelvin and Rossby waves. The first meridional mode (l = 1) annual Rossby wave reaches the western boundary while the other Rossby waves directly propagate toward the bottom. A deep Kelvin wave is generated by the reflection of the incoming l = 1 annual Rossby wave at the western boundary. Due to the complexity of the surface layers in the non-linear simulations, one cannot separate the waves generated at the boundaries and the waves directly forced by the wind. Below the thermocline, the waves characteristics agree well with the linear theory when considering the Kelvin and the l = 1 Rossby wave but discrepancies occur for the l = 3 and 5 Rossby waves. Non-linearities explain that minima reached by the zonal velocity anomaly is larger, in absolute value, than maxima. An important result of that study concerns the enhancement of the semi-annual cycle compared to the annual signal in the equatorial Atlantic Ocean. A resonance phenomenon occurs even when realistic coastline geometry, dissipation and non-linear terms are taken into account. Along with the wind-stress amplitude, the forcing period and the basin width, the horizontal dissipation controls the amplitude of the deep seasonal fluctuations. The waves are damped during their propagation when the lateral dissipation coefficient is set to a high value (v(h) = 2000 m(2) s(-1)). In an "inviscid" regime (when the waves reach the other side of the basin before being damped), the amplitude of the deep reponse is not much influenced by the parameterization chosen. Finally, bottom reflections are not necessary for the set up of the deep response and the Mid-Atlantic Ridge does not change the nature of the deep signal.