Choice of Vertical Coordinates Impacts Ice-Shelf Basal Melting
Comparing results from several ocean models that include the cavities under Antarctic ice shelves has shown significant, worrying differences in how much melting each model produces. In this work, we use three models that span a range of different vertical grids to explore why models with a high vertical resolution near the ice-ocean interface may produce significantly lower melt rates than their lower-resolution counterparts, depending on how the ice-ocean boundary is parameterized.
Understanding ice shelf-ocean interaction is fundamental to projecting the Antarctic Ice Sheet response to a warming climate as well as feedbacks of increased melting back on the climate. This work shows that more research is needed in order to derive a robust, physically based boundary-layer parameterization for the ice shelf-ocean interface.
Results from the Marine Ice Sheet-Ocean Model Intercomparison Project (MISOMIP) suggest that ocean models with different vertical discretizations produce significantly different melt rates in idealized simulations of the cavities under ice shelves. This work explores how current parameterizations for ice shelf-ocean thermodynamic interaction perform in three different models. Each model uses a different variant of a common parameterization, with important differences in how the temperature and salinity that drive melting are sampled and how meltwater fluxes are distributed. While some models assume a relatively thick (e.g. ∼20 m), spatially invariant thickness, others use finer vertical resolution and rely on explicit mixing over the resolved boundary layer. All models either suffer from an implicit dependency between melting and the vertical resolution of the model or an arbitrary, constant choice of the flux mixing and tracer sampling distances with no physical basis. Further investigation is required to understand the physics and processes that govern this transfer of heat and salt from the ocean outside the boundary layer into the ice. Improved parameterizations must better capture the transfer of heat and salt across both resolved and unresolved portion of the ice-ocean boundary region, without dependence on model vertical resolution.