First-year land-fast Antarctic sea ice as an archive of ice shelf meltwater fluxes

• A validated method to reconstruct Antarctic near-surface water δ18O values is presented in this paper.
• The range of sea ice effective fractionation coefficients derived for oxygen-18 was + 1.84 ‰ to + 2.21 ‰.
• The method is an improvement on previous approaches; uncertainty in reconstructed δ18O was ± 0.2 ‰.
• The method is of limited utility to detect ice shelf melt due to reconstruction uncertainty.
• The heat flux parameterisation requirement means that the method cannot be entirely retrospective.

Sampling beneath Antarctic ice shelves is sparse; therefore, tracking changes in ocean δ18O composition adjacent to ice shelves holds promise as an indicator of ice shelf basal melting. Sea ice archives of ice shelf–ocean interaction in particular could be important tools for future observational climate studies. Ocean δ18O values near the McMurdo Ice Shelf were reconstructed using observational data (sea ice δ18O, snow depth, and ice formation dates) from McMurdo Sound, Antarctica, by combining a recently revised version of an isotope fractionation model with an established thermodynamic sea ice model, resulting in improvements compared to previous approaches. Growth rates from the thermodynamic sea ice model were validated using direct growth rate measurements. That validation and supporting analysis indicated that a change was needed in ocean heat flux assumption from 0 W m− 2 to around − 13 W m− 2 part way through the sea ice growth season. A well-constrained range (+ 1.84 ‰ to + 2.21 ‰) of effective fractionation coefficients for sea ice was derived, along with a mean of 1.99 ‰. For the first time, reconstructed ocean δ18O values were validated using winter-long measurements of Antarctic near-surface water δ18O. Taking uncertainties into account, the reconstructed ocean δ18O values generally agreed to within ± 0.2 ‰ with the measured ocean δ18O mean values. Results indicated an overall decrease in measured ocean δ18O during the winter, but this was not statistically significant given the uncertainties in the measurements. Although the method works, it currently has limited utility for determining the presence and scale of any step-changes in ocean δ18O composition associated with present day ice shelf basal melting. This is because the uncertainty of the reconstructed values (± 0.2 ‰) is of the same magnitude as the expected change. Also, the requirement to parameterise the ocean heat flux is a barrier to the method being an entirely retrospective method (i.e., one requiring only data from the end of the sea ice growth season). In a future Antarctic scenario of increased basal melting of the ice shelves, the method may become more valuable in an Antarctic context. The method developed in this paper will be useful currently in the Arctic, because Arctic waters exhibit much larger fresh water fluxes.