Vertical velocities associated with open-ocean deep Session OS2.2: B987 Advances in understanding of in the NW as indirectly observed by gliders the multi-disciplinary dynamics of the Southern European Seas Anthony Bosse1 [[email protected]], P. Testor1, G. Legland2 L. Mortier2, L. Houpert1, L. Prieur3 (Mediterranean and Black Sea) Mediterranean Ocean Observing System for the Environment 1LOCEAN/IPSL, CNRS-Université Pierre et Marie Curie, Paris, France; 2 ENSTA-Paristech, Palaiseau, France; 3 LOV, CNRS-UPMC, Villefranche/mer, France.

Context of the Study Extracting vertical velocities from glider data: Methodology Open-ocean deep convection is a key process of the , as it renews the deep For a set of ight parameters ( ), once Following Merckelbach et al [2010], we wrote the CD0 , , Vg α waters of the oceans. This phenomenon is triggered by strong dry winds blowing in winter over a few is computed, one can get w Gliders are autonomous underwater vehicles pro- dierent forces working on the glider (F : buoyancy, U = mod/sin(θ + α) known places : the NW Mediterranean Sea, the and the . b from the force balance considering the glider to be pelled by buoyancy. They acquire a vertical veloc- Fg gravity, Fl lift and Fd drag) following: ity by changing their volume thanks to an oil pump. in quasi-steady ight. The open-ocean deep convection is classically described as: Large-scale cyclonic gyre (~100km) Their wings then convert it into a horizontal mo- Fb = gρ(Vg[1 − P + αT (T − T0)] +∆ Vg) As some parameters are not accurately known and tion. In this way, they glide along a saw-tooth trajec- ˆ occurring within the center of a large-scale cyclonic Fg = mgg might change during a deployment (biofouling, tory from the ocean surface (where they send their change in oil volume pumped, ...), an optimiza- gyre. The geostrophy implies a doming of the isopyc- 1 U2 real-time data) to a maximum depth of 1000m. A Fl = ρS aα tion of three critical parameters ( ) is per- nals resulting in a large-scale preconditioning to deep 2 CD0 , , Vg mixing; dive/ascent cycle is performed in about 4h, while the 1 formed using the following cost function: glider horizontally moves at the about speed of 20- F = ρSU2(C + C α2) d 2 D0 D1 X 2 X 2 ˆ being modulated by the mesoscale eld, which 30cm/s (⇔ 4km/cycle). By inferring their the- 2 wmod w C + C α J(CD0 , , Vg) = (dP/dt− ) = water can enhance or reduce the intensity of the mixing; α = D0 D1 oretical displacement through the water thanks to a a tan(θ + α) ˆ taking place in small convection plumes (L 1km), glider ight model in absence of mean vertical wa- The optimization is done within 1-day intervals ∼ Fig. 5: Schematic view of a glider moving in a with in green the parameters recorded by the glider Fig. 1: The dierent scales implicated where strong negative velocities ( 10cm/s) are pre- ter displacement, it is possible to extract the vertical allowing the parameters to slightly change during ∼ vertical plane [from Merckelbach et al., 2010]. and in red the unknown parameters to be optimized in open-ocean deep convection [from Mar- vailing (positive vertical velocities occur around these velocities of the water. the deployment and the water velocity to be about (the others are prescribed). shall and Schott, 1999]. plumes and are prescribed by mass conservation). zero in average. Being able to accurately observe these strong vertical velocities with the sucient time and spatial resolution remains a big challenge to characterize the convective plumes and get a deeper understanding Results from a glider deployment during winter 2013 of their role in deep open-ocean convection. 2 → The glider experienced several severe winter storms Heat flux [W/m ] with heat ux peaking at <-1000W.m-2; 500 The deep convection event in the NW Mediterranean Sea in winter 2013 0

12.8 12.9 13 13.1 13.2 13.3 In the NW Mediterranean Sea, the deep convec- 500 tion takes place in the center of a cyclonic gyre 1000 → Coastal turbid waters in the NC near the shelf + 44.5 whose northern part is the Northern Current (NC) 0 Enhanced turbidity signal during a strong con- 25 Feb 2013 Turbidity 44 and its southern part is associated with the North- vective event (right after the Feb. 25) in the mid- 0.1 dle of the deep water formation zone: could be Balearic Front (NBF). 500 43.5 0.05 the sign of suspended particles due to bottom- N During winter 2013, a multi-platform experiment 0 Depth [m] reached vertical mixing (also strong horizontal 43 was conducted to thoroughly study the deep con- 1000 -1 O2 [ mol.L ] currents ∼50cm/s over the whole water column); 42.5 vection event and the development of the spring 200 LION bloom: two R/V cruises (∼100 CTD stations in → Oxygenation of the mixed patch + Vertical gra- 42 winter and spring) + deployment of a eet of 500 180 160 dient;

gliders and release of bio-Argo oats. Depth [m]

Latitude 41.5 NC These data altogether represent an unprecedented 1000 Chl [mg.m-3] Vertical dilution of the phytoplankton in the 41 NBF → S observational network and enable a good delimi- 0.5 mixed patch + only signicant signals are located 40.5 tation of the deep convection area only from in 500 in frontal areas; situ proles (see the blue area in gure 2). 0 40 Depth [m] 1000 -3 Density fronts at the northern (:NC) and southern 39.5 ϑ [kg.m ] → → The individual dots represent all the proles (:NBF) boundaries of the mixed patch + signals of 39 carried out by gliders, Argo oats and R/V CTD 500 29.12 mesoscale eddies detaching from the rim current + 2 3 4 5 6 7 8 9 29.1 Longitude casts in a ±25 days temporal windows. Ar- density increase of the mixed patch from the begin- Depth [m] Fig. 2: Potential temperature averaged from rows represent the mean circulation. Tempera- 1000 ning to the end of the deep convection; 400m to 600m. The coloured contours is the ob- ture <13°C can be associated with deep convec- S [‰] jectively analysed eld on February 25, 2013. → tion (MLD>∼2000m, see gure 3). 500 38.55 → Slight salinity (θ) increase (decrease) of the mixed 38.5 38.45 patch from the beginning to the end of the deep Depth [m] convection due to atmospheric forcing + evidence 1000 Observation at the LION mooring ϑ [°C] of submesoscale activity resulting in vertical ex- The LION mooring is a highly instrumented mooring line (11 Seabird microcats + 5 Aquadopp current- 13.5 changes at the fronts; 500 meters), which is maintained since 2008. Its location is perfectly suited to observe the deep convection. 13

The winter 2013 can be characterized as a strong convective year. The convection reached the Depth [m] 1000 bottom (∼2500m) at LION by about mid-February (also observed by CTD casts with MLD >2500m). w [cm/s] → Vertical velocities ∼ 0 during period of weak at- The convection ceased on the 20th of March after a quick re-stratication followed by a last winter 0.1 mospheric forcing or in stratied conditions + storm that made the MLD reach ∼2000m. During the deep convection phase, ADCPs measured down- 500 0 vertical velocities > ±10cm/s when net heat loss -2 and upward vertical velocities of O(10cm/s) every 30 minutes. 0.1 <-500W.m ; Depth [m]

-1 Deep convection phase Pot. Temperature [°C] Horizontal currents (cm.s ) 1000 0 13.3 40 01/26 01/31 02/05 02/10 02/15 02/20 02/25 03/02 30 Date → Position of the glider (north/south sections) from 20 S 13.2 N 500 S N point N to S, see gure 2. 10

0 13.1 150m 250m 506m 1000m 2315m 1000 80 Distribution of vertical velocities Vertical velocities at the plume scale, a closer look on Feb. 24 Vertical currents (cm.s-1)

13 60 Surface waters are denser 1500 → LION mooring LION mooring 40 0.25 0 than below due to the in- -3 2 ϑ [kg.m ] 12.9 20 0.1 0.2 N <0 tense surface heat loss (<- 2000 -2 2 Deep convection phase 500W.m ): N <0 0 0.15 29.126

12.8 500 gravitationally unstable ; 01 Nov 15 Nov 01 Dec 15 Dec 01 Jan 15 Jan 01 Feb 14 Feb 01 Mar 15 Mar 01 Apr 15 Apr 01 May 07/2012 09/2012 11/2012 01/2013 03/2013 05/2013 07/2013 0.05 0.1 29.124

Fig. 3: (left) Potential temperature observed at the LION mooring with the Depth Depth [m] 0.05 29.122 → θ/S/σθ exhibits small

(MLD) in yellow; (left) horizontal and vertical velocities measured by the ve current meters of the Probabilty density function Probabilty density function inhomogeneities mooring line. 0 0 1000 −20 −15 −10 −5 0 5 10 15 20 −20 −15 −10 −5 0 5 10 15 20 S [‰] within the mixed layer: 0.1 0.5 38.5 O(0.01°C)/O(0.001 %) 500 glider glider -3 0.08 0.4 500 38.498 /O(0.001 kg.m ); old DW 38.496 ] 0.06 0.3 Depth [m] fresher/colder deep waters 2 0 → are associated with the old 0.04 0.2 1000 ϑ [°C] −500 deep waters; 0.02 0.1 12.93 Probabilty density function Probabilty density function → Downward convective latent 0 0 12.92 Heat fux [w/m −1000 sensible −20 −15 −10 −5 0 5 10 15 20 −20 −15 −10 −5 0 5 10 15 20 500 plumes are clearly identi- w [cm/s] w [cm/s] old DW 12.91 -1 net able: w -15/-10cm.s

Depth [m] ∼ Fig. 6: Distribution of vertical velocities measured by the 4 current meters in the rst 1000m at −1500 12.9 within small structures 01/01 01/15 02/01 02/15 03/01 03/15 the LION mooring (upper panel) and estimated by the glider (lower panel): (left) during periods of 1000 (0.51km); strong heat loss (<-500W.m-2) and unstratied conditions, (right) during periods of weak heat loss w [cm/s] 0.1 Fig. 4: Surface sensible and latent heat loss from the AROME-WMED atmospheric model [Seity et (>-100W.m-2) and/or stratied conditions. → Strong negative velocities transport warmer/saltier al., 2011] at the mooring location (42°N, 4°E40') during the deep convection phase. 500 0 ˆ Similar pdfs of the two estimates with a sharper (broader) pdf of the glider measurements under waters down;

weak (intense) buoyancy heat loss; Depth [m] The region of the endured several episodes of strong heat loss (<-500W.m-2): 5 in L~0.5-1km 0.1 → No signicant vertical ve- February and 1 in March. The diapycnal mixing is enhanced during these strong wind events, as shown ˆ Asymmetric distribution with a predominance of stronger downward velocities (up to 1000 ∼ 13:00 15:00 17:00 19:00 locities are associated with by the vertical velocities measured at the mooring. -15cm/s) than upward velocities (up to +10cm/s) presence of convective plumes. ∼ → Date the old fresher/colder deep waters;

References Perspectives: Conclusion:  Seity, Y., Brousseau, P., S. Malardel, G. Hello, P. Bénard, F. Bouttier, C. Lac and V. Masson, 2011 : The AROME-France convective ˆ Extend the analysis to the other gliders at sea during the deep convection (up to 5 gliders in 2013) scale operational model, Mon. Wea. Rev.  Marshall, J. and F. Schott, 1999 : Open-ocean convection: Observations, theory, and models, Rev. Geophys. and to other convective year to have an extensive panel of situations; ˆ The methodology presented by Merckelbach et al. [2010] provides good estimates of the vertical  Merckelbach, L., Smeed, D. and G. Griths, 2010 : Vertical Water Velocities from Underwater Gliders, JAOT velocities, especially during episodes of intense deep convection; ˆ Assess the horizontal length scale of convective plumes and its relation to the atmospheric forcing. ˆ Downward vertical velocities appear to occur within plumes of scale 0.51km, as previously re- ˆ Understand the nescale processes involved in the large-scale deep convection. ported [Marshall and Schott, 1999].