Effects of Arctic sea-ice concentration on turbulent surface fluxes in four atmospheric reanalyses

A prerequisite for understanding the local, regional, and hemispherical impacts of Arctic sea-ice decline on the atmosphere is to quantify the effects of sea-ice concentration (SIC) on the turbulent surface fluxes of sensible and latent heat in the Arctic.

The best available information in data-sparse regions such as the Arctic is provided by global atmospheric reanalyses. Because each reanalysis uses its own forecast model, data-assimilation system, and often also different atmospheric and surface observations to create the data sets, their atmospheric and surface variables, and boundary conditions often differ. While the differences between reanalyses in variables SIC, latent and sensible heat flux have been demonstrated via comparisons against observations and inter-comparisons between reanalyses, how much these data sets scatter in the effects of SIC on surface turbulent fluxes is not known.

To fill these knowledge gaps, we analyse these effects utilising four global atmospheric reanalyses: ERA5, JRA-55, MERRA-2, and NCEP/CFSR (CFSR and CFSv2), and evaluate their uncertainties arising from inter-reanalysis differences in SIC and in the sensitivity of the turbulent surface fluxes to SIC.

Using daily field means in nine Arctic basins, the magnitude of the differences in SIC is up to 0.15, but typically around 0.05 during all four seasons. Bilateral orthogonal-distance regression analyses indicate that the greatest sensitivity of both the latent and the sensible heat flux to SIC occurs in the cold season, November to April. For these months, using daily means of data, the average sensitivity is 400 W m^-2 for the latent heat flux and over 800 W m^‑2 for the sensible heat flux per unit of SIC (change of SIC from 0 to 1, positive sign referring to the downward flux). The differences between reanalyses are as large as 300 W m^-2 for the latent heat flux and 600 W m^-2 for the sensible heat flux per unit of SIC. The sensitivity is highest for the NCEP/CFSR reanalysis. Comparing two study periods 1980–2000 and 2001–2021, we find that the effect of SIC on turbulent surface fluxes has weakened, due to the increasing surface temperature of sea ice and the sea-ice decline.

Multilateral ordinary-least-square regression analyses show that the effect of SIC on turbulent surface fluxes arises mostly via its effect on atmosphere-surface differences in temperature and specific humidity, whereas the effect of SIC on wind speed (via surface roughness and atmospheric-boundary-layer stratification) partly cancels out in the turbulent surface fluxes, as the wind speed increases the magnitude of both upward and downward fluxes.