27.12.2013 Views

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

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

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

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

4.4. TILTED CONVECTIVE PLUMES IN NUMERICAL EXPERIMENTS 95<br />

104 A. Wirth, B. Barnier / Ocean Mo<strong>de</strong>lling 12 (2006) 101–111<br />

layer at every time-step. The ocean is cooled in a circular region attached to the surface, using a<br />

Gaussian profile given by<br />

!<br />

H ¼ 1.28 W m exp x 2 þ y 2 z 2<br />

ð4Þ<br />

3 ð200mÞ 2 ð70mÞ 2<br />

leading to a surface heat flux of<br />

!<br />

H s ¼ 80 W m exp x 2 þ y 2<br />

2 ð200mÞ 2<br />

The total heating is thus P ¼ R R 3HdV ¼ 1.0 107 W and the buoyancy flux is F 0 ¼ ðPagÞ=<br />

ðcqÞ ¼ 5.03 10 3 m 4 =s 3 . The source term in Eq. (3) is given by<br />

S ¼ H=ðcqÞ<br />

The numerical resolution in the horizontal is 128 · 128 grid points, there are 192 grid points in the<br />

vertical direction and the time-step is Dt = 60 s.<br />

ð5Þ<br />

tel-00545911, version 1 - 13 Dec 2010<br />

3. Tilted plumes<br />

The surface heat flux of 80 W/m 2 is smaller than the maximum values of heat exchange during<br />

specific convection events in the Labrador Sea, Greenland Sea and the Mediterranean (Marshall<br />

and Schott, 1999). We express the non-alignment of rotation and gravity by giving the latitu<strong>de</strong> h<br />

of the corresponding situation, that is: h = 90° N for an alignment of rotation and gravity on the<br />

North Pole (NP), h = 60° N for the situation in the Labrador Sea (LS), h = 45° N for the situation<br />

in the Golf of Lions (GL) and h = 0° for the situation at the equator (EQ) (rotation being perpendicular<br />

to gravity). We are aware of the fact that there is no <strong>de</strong>ep convection at the equator but<br />

inclu<strong>de</strong>d the case for compl<strong>et</strong>eness of the dynamical picture. We also performed calculations for<br />

the case with vanishing rotation (NR) in this case the horizontal expansion of the plume is not<br />

arrested, as it is in the rotating case. One-and-a-half days after the ons<strong>et</strong> of cooling the NRplumeÕs<br />

horizontal extension is such that the plume dynamics is strongly influenced by the finite<br />

(periodic) domain size (Fig. 1(a)). In the cases with rotation, the horizontal expansion of the<br />

plume is arrested and it only spans a fraction of the horizontal domain size.<br />

The most conspicuous feature in tilted convection is that the plumes extend in the direction of<br />

the axis of rotation rather than gravity (see Fig. 1). This finding, which has been explored by<br />

Sherem<strong>et</strong> (2004) in laboratory experiments for the oceanic context, is well known to researches<br />

consi<strong>de</strong>ring convection in rotating spheres (see e.g. Busse <strong>et</strong> al., 1998). The elongation of dynamical<br />

structures along the axis of rotation is attributed to the Taylor–Proudman theorem, which<br />

states that, in flows dominated by rotation the velocity vector is constant along the axis of<br />

rotation.<br />

For the first 5–8 h of the convection experiment the influence of rotation is not visible and all<br />

experiments look i<strong>de</strong>ntical during this time the plume has dropped by about 500 m, being in<br />

perfect agreement with the scaling in time t 1 = 2.4/X and distance h c1 = 3.3(F 0 /X 3 ) 1/4 proposed<br />

by Fernando <strong>et</strong> al. (1998) (based on laboratory experiments). After that period the plume starts

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!