Southern ocean overturning in an eddying model

Transcription

1 Southern ocean overturning in an eddying model Anne Marie Treguier (LPO, Brest) Matthew England (University of New South Wales, Sydney) Steve Rintoul (CSIRO, Hobart) Julien le Sommer, Jean Marc Molines (LEGI, Grenoble) Gurvan Madec (LOCEAN, Paris) Surface eddy kinetic energy Global ocean-ice model, 14km at 60 S, the DRAKKAR group

2 The meridional circulation in the Southern Ocean: a few ideas 1) The meridional circulation is most meaningful when integrated in density classes 2) The residual meridional circulation is the relevant one, sum of eddy and mean 3) Recent theories suggest un simple relationship between the diabatic surface forcing and the residual meridional circulation

3 Doos and Webb, 1994: Deacon cell? Downwelling at 40 S down to 3000m???? Wind driven Ekman transport Upwelling of deep water North South Manabe et al, 1990 GFDL coupled model

4 The downwelling flux does not cross isopycnals Mean isopycnals Meridional overturning calculated in density coordinates : Upwelling occurs along isopycnals; no deep Deacon cell. Doos and Webb, JPO 1994

5 In density coordinates, the meridional circulation has an eddy component Meridional circulation in density coordinates: v h = v h + v h v R = v + v h / h Non-acceleration theorem (Andrews and McIntyre, 1976): v R = 0 (if no diabatic forcing) In the presence of diabatic forcing, a RESIDUAL circulation remains: the diabatic Deacon cell (Speer et al, 2000)

6 The meridional circulation in the Southern Ocean: a few ideas 1) The meridional circulation is most meaningful when integrated in density classes 2) The residual meridional circulation is the relevant one, sum of eddy and mean 3) Recent theories suggest a simple relationship between the diabatic surface forcing and the residual meridional circulation

7 Residual meridional circulation and buoyancy forcing David Marshall, 1997, J. Mar. Res: zonal channel geometry Residual streamfunction at the ML base (-h) D = surface density flux σ y = meridional density gradient Ψ R = ƒ -h v R = D/ σ y -H Ψ R = Ψ+ Ψ = -τ /ρ 0 f - κσ y/ σ z Mean meridional circulation = Ekman Eddy component: GM parameterization

8 Evaluating Ψ R and D Karsten and Marshall, JPO τ /ρ 0 f Ψ R The ACC is non-zonal: follow streamlines Mean > eddy Ψ =-κσ y/ σ z Eddy > mean

9 Residual circulation? South Speer, Rintoul et al 2000: Near the surface no need for eddies, mean Ekman advection balances buoyancy forcing (freshening and heating). Eddy fluxes at depth. Karsten and Marshall: Eddies are important near the surface and overcome the Ekman component north of 54 S. South

10 Models: confusing view Lee et al, JPO 2007, new OCCAM model total eddy South Strong residual circulation (12 Sv) opposite to Ekman transport South Eddies are important only at depth (12 Sv at 2500m)

11 Aim of this study Use the DRAKKAR global ¼ model: 14 km grid at 60 S, long integration (47 years, ), well balanced air-sea fluxes. 1. Explore the residual mean circulation in the surface layers, following mean streamlines 2. Does the residual circulation balance the buoyancy forcing?

12 ORCA025 global DRAKKAR configuration Numerical characteristics: NEMO/OPA-Lim numerical code, Tripolar ORCA grid, Model domain broken down into 186 ocean processors. General validation: Barnier et al, ocean dynamics 2006; Penduff et al, ocean science 2007 (partial steps, momentum advection scheme). CORE experiment - Climatological forcing (Large and Yeager, 2004): NCEP and satellite precipitations and shortwave fluxes - 10 years run Reference experiment: interannual forcing, ERA40 instead of NCEP - modified TKE scheme

13 Model EKE Validation and model intercomparisons: Barnier et al, 2006, Ocean Dynamics Surface EKE TOPEX/ERS Surface EKE ORCA025 model

14 SSH variability SR3 section Altimetry ORCA025 ¼ Section from Autralia to Antarctica (Sokolov et Rintoul, 2007) Meridional gradient of SSH as a function of time (3 years): persistent jet-like structures

15 SR3 section Observations (Rintoul et al, january 1995) Model (ORCA025)

16 The ORCA025 model is well adapted to study the upper ocean circulation Next step: define streamlines. Proposed definition of the ACC region: Area bounded by the streamlines of time-mean barotropic transport that go through Drake Passage.

17 Streamline definition (3) Karsten and Marshall: smoothing, stop at 67S (no ssh data under ice) ORCA025-G70, well defined Weddell and Ross gyres, stop at 80 S: interpolate linearly between 117 Sv line and the Antarctic coast.

18 Streamline-averaged ACC

19 Streamline-averaged forcings Heat contribution = cooling Freshwater = freshening Density flux: Black = reference, buoyancy gain (freshening) red = CORE; buoyancy loss (cooling) Ekman transport

20 Residual meridional circulation and buoyancy forcing Residual streamfunction at the ML base (-h) D = surface density flux σ y = meridional density gradient Ψ R = ƒ -h v R = D/ σ y -H Residual circulation averaged over ACC region: 3 Sv (reference case) -1 Sv (CORE case)

21 Streamline and zonal mean differ! Streamline view: similar to Speer et al Zonal average: opposite cell in the upper layers, due to time-mean rather than eddy, resolution-dependent

22 Meridional circulation in σ 0 Southern gyres ACC 100m depth 500m depth Total: northward, 12 Sv Upwelling occurs along isopycnals or in the mixed layer.

23 Meridional circulation in σ 0 Eddy circulation: opposite to mean, a little weaker Total: northward, 12 Sv Total CORE run: 7-8 Sv

24 Meridional circulation in σ 0 50 S 57 S Compare Reference (thick lines) and CORE (thin lines) experiments At what depth should the theory apply?

25 Inhomogeneity of the mixed layer depth along the ACC Africa Tasmania Drake

26 Theory and 3D model Residual streamfunction at the ML base (-h) D = surface density flux σ y = meridional density gradient Ψ R = ƒ -h v R = D/ σ y -H At what depth should this relation apply? Ψ R = Ψ+ Ψ = -τ /ρ 0 f - κσ y/ σ z

27 Theory and 3D model Ψ R = Ψ+ Ψ Ψ - τ /ρ 0 f Ψ =? -κσ y/ σ z The Eke structure depends on topography : instability is mixed bartropic/baroclinic and/or modified by bathymetry. The transient part includes seasonal cycle and effect of transient air-sea fluxes: it may not be simply related to baroclinic instability Ψ seems compatible with GM parameterization, but not with constant coefficient, 500 m²/s² rather than 2000 m²/s²

28 Conclusions The meridional circulation averaged zonally is confusing The meridional circulation across the ACC (in streamline coordinates) is in qualitative agreement with theory. Eddy fluxes and eddy transports are large. Quantitatively, there is no simple relation with the buoyancy forcing. Factors to take into account: cabbelling, 3D structure of the ACC (for eddy structure as well as vertical mixing) Paper available in Ocean Science discussions (EGU journal see web site)

29 Inhomogeneity of eddies along the ACC Surface EKE averaged over the ACC band at each longitude Different regimes, role of topography (and probably barotropic instability) A zonally averaged mixing coefficient does not reflect this complexity

30 Balances in the ACC region Between the two black streamlines (southern and northern limits), surface to bottom: freshwater, heat, salt, density

31 Zonally averaged streamfunction ORCA025-G70 OCCAM (Lee et al, 2007)

32 Model resolution Model grid Dy (60 S) Reference POPCM ¼ Lon,lat 44 km Jayne and Marotzke FRAM 1/3 Lon,lat 28 km Doos and Webb OCCAM ¼ Lon,lat 28 km Lee and Coward POP 1/6 Mercator 16 km Olbers, Ivchenko ORCA025 Mercator 14 km Barnier et al MESO 1/6 Mercator 9 km Hallberg OCCAM 1/12 Lon,lat 9 km Lee et al POP 1/10 Mercator 6 km Maltrud McClean

33 Eddy Kinetic Energy Satellite T/P+ERS ORCA-1/4 EEN+PS OCCAM-1/4 Flux+PS ORCA-1/4 ENS+FS POP-1/10 Flux+FS CLIPPER-1/6 ENS+FS

34 Transport adjustments Drake passage transport slows down from 150 to 118 Sv over 12 years; Stable afterwards.

35 Acc path definition (2) Subantarctic front Polar front

36 The meridional circulation in the Southern Ocean: a few ideas 1) The meridional circulation is most meaningful when integrated in density classes 2) The residual meridional circulation is the relevant one, sum of eddy and mean 3) Recent theories suggest un simple relationship between the diabatic surface forcing and the residual meridional circulation

37 ACC path definition (2) Between Antarctica and the ACC: Ross and Weddell gyres ACC region ACC belt

38 Meridional circulation in σ 0 Time mean flow: Ekman transport (northward), Deacon cell, mean upwelling across isopycnals Southern gyres ACC 100m depth 500m depth South 20 Sv