Observations of turbulence at a near-surface temperature front in the Arctic Ocean


 <p>Measurements made at an Arctic thermohaline front show turbulence production through convection and forced symmetric instability, a mechanism drawing energy from the frontal geostrophic current. Destabilizing surface buoyancy fluxes from a combination of heat loss to the atmosphere and cross-front Ekman transport by down-front winds reduce the potential vorticity in the upper ocean. The front, located in the Nansen Basin close to the sea ice edge, separates the cold and fresh surface melt water from the warm and saline mixed layer. High resolution temperature, salinity, current and turbulence data were collected in the upper 100 m, on 18 September 2018 across the front from a research vessel and an autonomous underwater vehicle. The AUV was deployed to autonomously collect high resolution data across the front using adaptive sampling. Both front detection and sampling location were decided by a state-based autonomous agent running onboard the AUV, optimizing data collection across and along the front.</p><p>In addition to convection by heat loss to atmosphere and mechanical forcing by moderate wind in the mixed layer, forced symmetric instability contributed with comparable magnitude in generation of turbulence at the front location down to 40 m depth. This turbulence was associated with turbulent heat fluxes of up to 10 W.m<sup>-2</sup>, eroding the warm and cold intrusions observed at respectively 35 and 55 m depth. A similar frontal structure has been crossed by a Seaglider in the same region 10 days after our survey. The submesoscale-to-turbulence scale transitions and resulting mixing can be widespread and important in the Atlantic sector of the Arctic Ocean.</p>


where ρ 0 = 1027 kg m −3 is the seawater density, C p = 3991.9 J kg −1 K −1 is the spe-164 cific heat of seawater, Θ is the background temperature and K ρ is the diapycnal eddy 165 diffusivity. We thus assume that turbulence diffuses the finescale temperature gradient 166 at the same rate as the density gradient. The sign convention is that positive heat fluxes 167 correspond to upward heat fluxes in the water column. An upper bound for diapycnal 168 diffusivity was obtained using the Osborn (1980) relation: with the mixing coefficient set to Γ = 0.2, the recommended value for the oceanic ap- and oxygen values were corrected against water sample analyses.

218
The wind speed, direction and surface air temperature (Figure 2a and c) were recorded 219 every minute during the cruise from the ship's weather station. Outside the cruise dates, 220 the ERA5 reanalysis is used (www.ecmwf.int/en/forecasts/datasets/reanalysis-datasets/era5).

221
Total net surface flux Q (in W m −2 , positive downward into the ocean, Figure 2b) 222 is expressed as Q = Q rad + Q l + Q s , with Q rad the net radiation (sum of net short-223 wave and longwave radiation), Q l the latent heat flux and Q s the sensible heat flux.  , et al., 2017;Meyer, Fer, et al., 2017). In this region, turbulence is mainly driven 255 by wind and by tidal currents over slopes. Turbulent heat fluxes exceeding 100 W m −2

259
During our measurements, sea ice edge was visible from the ship, and a Sentinel 260 1 image confirms that the front was located 4 -5 km from the sea ice edge (Figure 1a).

261
The average wind speed was moderate with 7 m s −1 from northeast, and the tidal cur-262 rents were weak, less than 3 cm s −1 (Figure 2a  the upper 500 m, or 10 km when the deepest measurement was extrapolated to full depth.

278
The mixed layer deformation radius, defined as R ml = N D ml /f where D ml is the mixed 279 layer depth, varied in the range of 1 and 4 km.

299
The adjusted station locations are shown in Figure 3. The near-surface tempera-300 ture is then mapped using an objective interpolation. As the vessel crossed the front, 487 q bc = ∂u ∂z ∂b ∂y .
N 2 is the Brunt Vaisala frequency, u is the along-front velocity and b is the buoyancy 488 (b = − g ρ0 σ 0 ). The geostrophic potential vorticity q g is approximated by: where the geostrophic shear is substituted as

525
Our calculations show that the front is susceptible to develop FSI; however, the ob-526 served dissipation integrated over H is lower than the potential supply from FSI (Fig-527 ure 9b). We do not observe a substantial increase in dissipation rates driven by FSI at 528 the front location. There is no conclusive evidence; however, the turbulence production 529 by FSI could potentially contribute to the elevated dissipation rates beneath the mixed 530 layer (Figure 7, blue profile).

539
We estimated the vertical velocity w using the quasi-geostrophic ω equation (Hoskins   540   et al., 1978): where Q is the divergence of the kinematic deformation: Q = ∇u.∇ gρ ρ0 . We obtained 542 Q using the objectively interpolated composite cross-front sections (density and cross-   Figures 6 and 11). Also, a week before the glider crossed the front, the wind 581 blew steadily for 3-4 days from the northeast (not shown), in approximately down-front 582 direction, resulting in similar atmospheric conditions as encountered during our study.

583
-20-the front crossed by the glider shows similar dynamics (convection due to heat loss to 585 the atmosphere, mechanical forcing by moderate wind, substantial EBF and FSI). As 586 the front is visible in the SST map in a broad region north of Svalbard, one can suggest 587 that the frontal structure is likely active all along the front north of Svalbard. teraction of inertial oscillations and FSI at a front shapes the stratification, shear, and They showed that FSI more efficiently extracts energy from a front via shear produc-616 tion during periods when inertial motions reduce stratification. During our survey, the 617 wind variability was weak and no evidence of inertial motions were seen. However, storms 618 are strong and frequent in the Arctic, and are observed to lead to energetic turbulence 619 and increased heat fluxes (Meyer, Fer, et al., 2017;Graham et al., 2019). The possible 620 contribution of FSI was not addressed in these studies, or in the Arctic fronts in general.

621
FSI is a process that can increase turbulence at the fronts, particularly during storms,