Unraveling the uncertainties of bulk-derived heat fluxes: a case study for the Southern Ocean

The icebreaker R/V SA Agulhas II spent 3 months (December-Feb) in the open ocean (35 days) and sea ice (40 days) collecting atmospheric and oceanographic variables required for calculating momentum, sensible heat, and latent heat fluxes. In addition, both longwave and shortwave radiative fluxes were measured by radiometers to provide a full air-sea heat flux budget. These observations were compared against the commonly-used reanalysis product ERA5 to evaluate surface heat flux components in both the Southern Ocean sea ice and open ocean regions during austral summer to better understand air-sea interactions in the region. Both sensible and latent heat fluxes had significant short-term events (less than a day) that reduced the daily mean by 13% and 3% respectively. Wind speed, air temperature, shortwave, latent and sensible heat fluxes were all underestimated by ERA5 in sea ice, while SST and longwave were overestimated. Ship-based sensible heat flux in sea ice exhibited a diurnal phasing with a minimum ocean heat loss during mid-day (-25 Wm⁻². ERA5 had a reversed diurnal phase with a maximum heat loss in mid-day (-23 Wm⁻²). Ship-based latent heat flux varied little (±3.6 Wm⁻² daily range), whereas ERA5 had a diurnal phase similar to sensible heat flux  (-62 Wm⁻²).The total biases in the neat heat flux show that ERA5 underestimates the net heat flux by 65 Wm⁻² in sea ice due to the difference in diurnal phases of turbulent fluxes. Here, the sensible and latent heat flux are underestimated by 34 Wm⁻² and 20 Wm⁻² respectively. In the open ocean, turbulent fluxes agree well between ERA5 and ship observations (<10 Wm⁻² difference). Shortwave and longwave (radiative fluxes) are consistently biased in estimations by ERA5 in both sea ice and open ocean, possibly due to parameterization of clouds. Longwave radiation is overestimated by 28 Wm⁻² by ERA5 in both regions, shortwave is cold biased (underestimated) by 25 Wm⁻² in sea ice and warm biased (overestimated) by 46 Wm⁻² in the open ocean. This in situ evaluation of heat flux components is highly valuable for further improving our understanding of heat fluxes in the Southern Ocean.