Data used in the manuscript entitled "Turbulent heat flux dynamics along the Dotson and Getz ice-shelf fronts (Amundsen Sea, Antarctica)"

Data files used in the analysis in the manuscript entitled "Turbulent heat flux dynamics along the Dotson and Getz ice-shelf fronts (Amundsen Sea, Antarctica)". Data were collected during the RV NB Palmer NBP2202 cruise, during the 2022 TARSAN campagine in the Amundsen Sea. Underway data provides daily files from the underway and meteorology sensors in JGOFS format. CTD data collected from the cruise. Information about sensors and data formats is included in the data report. Glider data was processed through the UEA Seaglider Toolbox (https://bitbucket.org/bastienqueste/uea-seaglider-toolbox/src/toolbox) and is provided in Matlab format. Manuscript abstract: In coastal polynyas, where sea–ice formation occurs, it is crucial to have accurate estimates of heat fluxes in order to predict future rates of sea–ice formation. The Amundsen Sea Polynya is the fourth largest coastal polynya around Antarctica, yet remains poorly observed because of its remoteness. Consequently, we rely on models and reanalysis that are unvalidated to study the effect of atmospheric forcing on polynya dynamics. We use summer ship-board data from the NBP22/02 cruise to understand the turbulent heat flux dynamics in the Amundsen Sea Polynya and evaluate our ability to represent these dynamics in ERA5. We show that cold and dry air outbreaks from Antarctica enhance air–sea temperature and humidity gradients, triggering episodic heat loss events. The heat loss is larger along the ice shelves, and it is also where the ERA5 turbulent heat flux exhibits the largest biases, underestimating the flux by up to 141~W~m$^{-2}$ due to its coarse resolution and misrepresentation of ice-shelf location. By reconstructing a turbulent heat flux product from ERA5 variables using a nearest neighbour approach to obtain sea surface temperature, we decrease the bias to 107 W m$^{-2}$. Using a 1D-model, we show that the mean co-located ERA5 heat loss underestimation of -28~W~m$^{-2}$ led to an overestimation of the summer evolution of sea surface temperature (heat content) by +0.76~°C (+8.2e+07~J) over 35-days. By obtaining the reconstructed flux, the reduced heat loss bias (12 W~m$^{-2}$) reduced the seasonal bias in sea surface temperature (heat content) to -0.17~°C (-3.30e+07~J) over the 35-days. This study shows that caution should be applied when retrieving ERA5 turbulent flux along the ice shelves, and that a reconstructed flux using ERA5 variables shows better accuracy.