Etudes et évaluation de processus océaniques par des hiérarchies ...

104 CHAPITRE 4. ETUDES DE PROCESSUS OCÉANOGRAPHIQUES
APRIL 2008 W I R T H A N D B A R N I E R 803
Mean Circulation and Structures of Tilted Ocean Deep Convection
A. WIRTH AND B. BARNIER
MEOM/LEGI, Grenoble, France
(Manuscript received 4 January 2007, in final form 17 April 2007)
ABSTRACT
tel-00545911, version 1 - 13 Dec 2010
1. Introduction
Convection in a homogeneous ocean is investigated by numerically integrating the three-dimensional
Boussinesq equations in a tilted, rotating frame (f–F plane) subject to a negative buoyancy flux (cooling)
at the surface. The study focuses on determining the influence of the angle (tilt) between the axis of rotation
and gravity on the convection process. To this end the following two essential parameters are varied: (i) the
magnitude of the surface heat flux, and (ii) the angle (tilt) between the axis of rotation and gravity. The
range of the parameters investigated is a subset of typical open-ocean deep convection events.
It is demonstrated that when gravity and rotation vector are tilted with respect to each other (i) the
Taylor–Proudman–Poincaré theorem leaves an imprint in the convective structures, (ii) a horizontal mean
circulation is established, and (iii) the second-order moments involving horizontal velocity components are
considerably increased.
Tilted rotation thus leaves a substantial imprint in the dynamics of ocean convection.
Most of the World Ocean is known to be in hydrostatic
balance at large scales. However, ocean dynamics
at scales smaller than about 10 km, convection, circulation
in coastal areas, surface mixed layer dynamics,
the breaking of internal waves, and the dynamics of
overflows in straits are instances of ocean dynamics
where nonhydrostatic effects are essential. These instances
of nonhydrostatic ocean dynamics are important
dynamical problems in their own right, but they
also influence the large-scale dynamics of the World
Ocean and the climate system of our planet. A detailed
discussion of nonhydrostatic effects in the ocean dynamics
can be found in Marshall et al. (1997).
A striking feature of open-ocean deep convection is
that, although it governs a substantial part of the poleward
heat transport of the atmosphere–ocean system
through its influence on the thermohaline circulation, it
is nevertheless extremely localized in space and time. In
fact, convection chimneys account only for a tiny fraction
of the World Ocean, and a convection event typically
lasts for only about 1 week. A substantial part of
Corresponding author address: A. Wirth, MEOM/LEGI, BP 53,
38041 Grenoble CEDEX 9, France.
E-mail: achim.wirth@hmg.inpg.fr
the large-scale and long-time ocean and climate dynamics
is thus slaved to what happens at these almost
space–time singularities. For a detailed discussion of
open-ocean convection from an observational, theoretical,
and modeling perspective, we refer the reader to
the review of Marshall and Schott (1999). For a recent
study of the impact of the convection process on the
large- and mesoscale dynamics in the Labrador Sea, we
refer the reader to Straneo (2006) and Chanut et al.
(2008, hereafter ChBa).
Convection chimneys, being by themselves an ensemble
of many convection plumes, typically measure
about 100 km in the horizontal directions. They thus
barely span the area of a single horizontal grid point in
ocean global circulation models (OGCMs) employed in
today’s climate models. Even when much higher resolutions
were computationally feasible, most models will
not be able to explicitly represent the convection process
because they integrate the hydrostatic primitive
equations, neglecting vertical momentum. Numerical
ocean models based on the hydrostatic approximation
are, and will be in the foreseeable future, the major
tools in entangling the ocean dynamics at the global or
basin scale. Such models, however, have to be supplemented
at certain locations to account for the influence
of the nonhydrostatic processes, as, for example, convection.
This can be achieved in two ways: first, by
locally nesting a nonhydrostatic model at the critical
DOI: 10.1175/2007JPO3756.1
© 2008 American Meteorological Society