Resolution requirements for grounding-line modelling: sensitivity to basal drag and ice-shelf buttressing

Simulations of grounding-line migration in ice-sheet models using a fixed grid have been shown to exhibit poor convergence at achievable resolutions. We present a series of 'shelfy-stream' flowline model experiments using an idealized set-up. We assess the performance of a range of grounding-line parameterizations (GLPs) over a large input space by varying bedrock gradient, rate factor, basal drag coefficient and net accumulation. The relative performance of GLPs is similar to Gladstone and others (2010a) except at low basal drag, in which case the grounding-line errors are very small for all GLPs. We find that grounding-line errors are far more sensitive to basal drag than to the other inputs or to choice of GLP. We then quantify grounding-line errors as a function of resolution while varying basal drag and channel width (using a parameterization to represent buttressing). Reducing either basal drag or channel width reduces the errors associated with the grounding line. Our results suggest that a structured fixed-grid shelfy-stream ice-sheet model would need to run at a horizontal resolution of similar to 1-2 km to accurately simulate grounding-line positions of marine ice-sheet outlet glaciers such as Pine Island Glacier, Antarctica.