Turbulence and mixing within anticyclonic eddies in the Mediterranean sea during BOUM experiment

Transcription

1 Turbulence and mixing within anticyclonic eddies in the Mediterranean sea during BOUM experiment Yannis Cuypers, Pascale Bouruet-Aubertot, Claudie Marec, Jean-Luc Fuda

2 BOUM Objectives Biogeochimie de l'oligotrophie a l'ultra-oligotrophie Mediterraneenne radiale XBT tourb C centre en item 5 A main goal: The representation of the interactions between planktonic organisms and the 0 22 cycle of biogenic elements, considering scales from single process to the whole Mediterranean Sea (Moutin et al submitted) Vertical transport by turbulent mixing has a transverse impact on biogeochemical processes studied in BOUM brings nutrient to the depleted euphotic zone of the oligotrophic Mediterranean sea waters fuels primary production and impact carbon export But turbulent mixing is poorly documented in the central Mediterranean sea 900 To our knowledge only one campaign of microstructure measurements (Woods & Wiley longitude Est 1972) Vertical temperature section (But recent measurements in gulf of Lion, Petrenko et al (LATEX) and cycladic plateau Gregg et al ) Effort made during BOUM to characterize vertical mixing focus on 3 anticyclonic eddies=>isolated environments=> importance of vertical transport, intrinsic physical processes upwelling, internal waves trapping (Ledwell 2008, Kunze 1995)

3 BOUM Measurements Classical fine-scale measurements: Repeated CTD/LADCP profiles ~ every 3 h for 3 days at station A, B and C (within eddies) Salinity and Temperature at 1 m resolution Horizontal currents at 8 m resolution Temperature microstructure measurements at station A, B and C =>dissipation ε at 1 m resolution

4 Outline Stratification and internal waves within the eddies Turbulent dissipation in the first 100m and validation of a fine scale parameterization Parameterized dissipation and estimation of vertical diffusivity over full eddies depth ranges Conclusions & Perspectives

5 Stratification within eddies A, B and C Eddy A 0 Eddy B Log10(N2(rad/s)2) Eddy C pres Pressure(db) Core Core Core days 3 days Shallow seasonal pycnocline 15 m Low stratification within homogeneous eddy cores 3 days -8

6 0 ζ -0.2f Zonal velocity Eddies A, B and C Eddy A ζ -0.15f Eddy B ζ -0.15f Eddy C U(m/s) 0.2 pres(db) Core Core Core days 3 days 3 days Strong near inertial shear at the top and base of Eddy A and C Mechanism generating Strong near inertial shear at depth: Trapping of subinertial waves (f eff =f+1/2ζ) and energy increase at a critical layer at the eddy base (Kunze 1985)? Baroclinic adjustment of the eddy? -0.2

7 Kinetic Energy Spectra Spectra of KE (St.A) Spectra of KE (St.B) Spectra of KE (St.C) 10-3 Data GM 10-3 Data GM 10-3 Data GM f m 2 /(s 2.cpd) 10-4 f M2 M4 f 10-5 M freq. (cpd) freq. (cpd) freq. (cpd) Dominant near inertial peak at St C and st A (resolution too coarse to state wheter it is subinertial) M2 internal tides at Station C, (M4 at Station B?) Spectra leveal sligthly below canonical Garret and Munk (1976) level for station A and C slightly above Garret and Munk level for station B

8 Temperature Microstructure measurements Temperature gradient spectrum (a scalar) in a turbulent homogeneous is modeled by the Batchelor spectrum This spectrum rolls off at the Batchelor scale where generation of sharp temperature gradient by turbulent strain is balanced by diffusion. l k = ε ( ) 1 1/4 B 2 νκt SCAMP small free fall microstructure profiler (max depth 100m) with vertical resolution of mm=>resolve betchelor scale for ε<10-5 W/kg Maximum Likelihood Estimate of the Batchelor spectrum with k B as the variable parameter k b

9 Temperature Microstructure Measurements Station C Log10(ε(W/kg)) Mixed Layer Background in gray scale=strain ξ z with ξ isopycnal displacement T inertial Strong variability of dissipation : <ε< W/kg, High values in the seasonal pycnoncline (10-20)m: ε mean = W/kg Moderate values below the seasonal pycnoncline (z>20m) ε mean = W/kg Influence of internal waves strain (Alford 2010, Alford Pinkel 2000) (important to take into account in a parameterization)

10 Dissipation distribution How to average ε? Simple arithmetic mean

11 Fine scale parameterization of dissipation Assuming a Garret and Munk spectrum, nonlinear wave wave interactions models predict a scaling as ε E GM2 N 2 (D Asaro and Lien 1999, Henyey et al 1985) Gregg (1989) proposed a popular incarnation of this scaling expressed with shear and taking into account deviation from GM level Several studies (Alford 2010, Alford and Gregg2001) and models (Kunze (2006), Gregg 2003, Polzin) suggest to take into account the influence of strain. We consider strain through the function h(rw) (Kunze 2006), where Rw is the shear variance to strain variance ratio

12 Parameterized ε vs measured ε Good agreement between measurements and parameterization that falls within the 95% CI over 80% of the profile length The dissipation level is comparable to GM below the seasonal pycnocline (20m depth) but nearly two order of magnitude higher above Parameterization should be considered with much caution above 20 m depth because comparison with GM may not be valid there (proximity of surface boundary)

13 Parameterized Kz vs experimental Kz Kz is estimated from (Shih et al 2005) - for ε /( υn Kz = ΓεN for ε /( υn 2 ε Kz = υ 2 υn ) < 100 (Osborn 1980) ) > 100 1/ 2 Kz is comparable to GM below the seasonal pycnocline (20m depth) but one order of magnitude higher in the pycnocline, this suggests important exchange with the mixed layer Values slightly smaller than found within upper 100 m in other anticyclonic eddies with similar shallow seasonal pycnocline in Sargasso sea (Ledwell 2008) or in North Atlantic (Dae Oak et et al 2005) but with stronger wind forcing N2 fluxes computed from these Kz may explain only a small fraction of N2 fixation rate (Bonnet et al BGD 2011)

14 ε estimates over eddies full depth range C B A Increase trend of ε at the base of eddy C and A and at the top of eddy C where maximum near inertial shear is observed Low dissipation within eddies core

15 Kz estimates over eddies full depth range C B A Lower dissipation partly balanced for Kz by the lower stratification within the eddies Average Kz is generally higher by a factor 2 to 3 to GM values at depth despite of low internal waves energy sources (weak winds in summer and weak tides) =>Trapped near inertial waves?

16 Conclusions and perspectives Microstructure measurements: High ε values in the seasonal pycnocline and relatively high Kz, This suggests that the seasonal pycnocline may be pemeable to exchange between mixed layer and stratified deeper water ε and Kz estimates are comparable to GM below the seasonal pycnocline for (z<100m) Fine scale parameterization is in good agreement with direct measurements (0<z<100m) high ε param values at the base of eddies associated with inertial shear Kz values higher than GM level at large depth (z>100m) resulting from strong shear at the eddy base or weak stratification within eddies Future work: Determine the mechanism of strong near inertial waves generated at the base of eddies A and C. Possible candidate wave trapping at the eddy base decrease of group velocity and increase of energy (Kunze 1985, Lee and Niler 1998) Impact of turbulent mixing on mixed layer budget Rough estimate of vertical advection