On the Along-Slope Heat Loss of the Boundary Current in the Eastern Arctic Ocean Schulz,

Dyfyniad o'r fersiwn a gyhoeddwyd / Citation for published version (APA): Schulz, K., Janout, M. A., Lenn, Y-D., Castillo, E. R., Polyakov, I., Mohrholz, V., Tippenhauer, S., Reeve, K., Holemann, J., Rabe, B., & Vredenborg, M. (2021). On the Along-Slope Heat Loss of the Boundary Current in the Eastern Arctic Ocean. Journal of Geophysical Research: Oceans, 126(2), [e2020JC016375]. https://doi.org/10.1029/2020JC016375

ity was bounded within the M2 frequency band (1.9-2 cycles per day) and was removed 174 from the time series using a 100-hour running average. is not the same as the AW layer, which is often defined as the layer between the 0 • C isotherms 190 (Polyakov et al., 2017), or based on potential density (1027.70-1027.97 kg m −3 ) and po-191 tential temperature (> 2 • C) (Rudels et al., 2000). Thermohaline staircases, which are 192 found the the depth of the AW thermocline at some stations, were visually identified.

193
The upper ocean heat content (in J m −2 , displayed in Fig. 7) is calculated accord- where θ f is the (salinity and pressure dependent) freezing temperature, ρ 0 = 1027 kg m −3 196 is the seawater density and c p ≈ 3991.9 J kg −1 K −1 the specific heat capacity of sea- In the post-processing of the microstructure profiler data, signals from the respec- on the cable when the profile is terminated before reaching the sea floor). In each raw 215 data channel, data points that exceed 3 times the standard deviation, calculated over 216 40 data points, were identified as outliers, removed and linearly interpolated.

217
-9-manuscript submitted to JGR: Oceans The dissipation rate ε is calculated independently from each shear sensor by fit-218 ting Nasmyth's universal turbulence spectrum (Nasmyth, 1970) to the power spectrum 219 of subdivided sections of 512 data points, after removing the linear trend from each sub-220 section. The results derived from the two shear sensors were subsequently averaged, where 221 again data points were discarded when the individual dissipation estimates differed by 222 a factor of 5. All data were subsequently averaged to 1 m vertical resolution.

223
Unfortunately, no direct current velocity measurements are available contempora- where κ = 0.41 denotes the von Kármán constant and z the height above bottom. As 230 this relation is only valid in the well-mixed near-bottom layer, only the lowermost two 231 bins, corresponding to the lowermost 2 m of the water column, were used for the calcu-232 lation.

233
The calculation of vertical heat fluxes from the microstructure data requires the 234 turbulent diffusivity where ε denotes the dissipation rate and N the buoyancy frequency.

267
Using the mixing efficiencies discussed above (Γ = 1 in the AW thermocline, Γ = 268 0.2 otherwise), the turbulent heat flux is then calculated as where θ denotes the conservative temperature, ρ 0 and c p are again the sea water den-270 sity and the specific heat capacity of sea water, respectively, and the z coordinate is ori-  -13-manuscript submitted to JGR: Oceans

334
To quantify the apparent heat loss from west to east along the ABC pathway, we   ing north again (Fig. 10A). The drift track roughly follows the contemporary modeled timates, therefore includes regional over-and underestimations. Nevertheless, consid-533 ering the robustness of the heat content calculations discussed above, and the high co-534 efficient of determination (R 2 =0.98) of the linear regression (Fig. 7), we are confident   , et al., 2020). Thermohaline staircases were identified in some CTD pro-587 files presented in this study, but they did not exist throughout the (deeper parts) of the 588 study region (Fig. 5). While thermohaline staircases are not expected near the energetic 589 shelf break, their absence in the deeper part of the 126 • E transect might be a first sign for the above mentioned change in conditions, but more observational data is needed to 591 confirm this hypothesis.

592
The mean vertical heat flux at the upper AW interface of 10 W m −2 in the offshore 593 (based on the 24 hour station) and 3.7 W m −2 in the onshore regions are larger than pre- fraction of the heat loss is thus attributed to mixing with ambient cold water in the con-761 tinental slope region (Fig. 12). There, the observed dissipation rates were highest but 762 heat fluxes (4 W m −2 ) were lower than in the deep basin, which is due to weaker tem-763 perature gradients as a result of the enhanced mixing. Heat fluxes were strongly elevated 764 in the near-bottom region above the slope, where deep warm water intersects the tur-765 bulent bottom boundary layer, as well as on the lee side of a topographic sill, as was ob-766 served during a 10 hour-microstructure survey from a freely drifting ship. Our observa-767 tions indicate that diapycnal mixing prevails above the slope, while the basin regions are 768 dominated by lateral homogenization of the AW layer through isopycnal mixing (Fig. 12), 769 which agrees with the general perception that basin-wide diapycnal mixing is to first or-770 der determined by boundary mixing, while lateral (isopycnal) mixing dominates the calmer 771 interior regions (Stigebrandt, 1979;Goudsmit et al., 1997;Ledwell & Bratkovich, 1995;772 Holtermann et al., 2012). Other processes such as winter ventilation that could poten-773 tially contribute to AW heat loss, are unlikely to play a dominant role in the present east-

789
The episodic nature of turbulence is a major source of uncertainty for heat budgets as 790 well as for nutrient fluxes, and therefore requires enhanced efforts to develop and improve 791 mooring-based methods to measure turbulent mixing year-round.

792
Acknowledgments 793 We would like to thank the crew and participants of the Akademik Tryoshnikov cruise.

794
Financial support was received from the German Federal Ministry for Science and Ed-  Hydrographic data used in this study is available at: