Exploring grid sensitivity in an ice sheet model: A case study of the Amery Ice Shelf

Abstract

The dynamics of the Antarctic ice sheet are key factors affecting global climate change. To project future sea level changes, ice sheet models are developed based on a discrete grid system, which profoundly impacts the accuracy of numerical simulations. To comprehensively explore the sensitivity of ice modeling performance to grid resolution, this study focused on the Amery Ice Shelf (AmIS), the largest glacial flow system in east Antarctica. Using ten grid resolutions (nine sets of structured grids ranging from 2 to 30 km and one set of adaptive unstructured grid from 1 to 20 km), we conducted a series of inversion experiments and diagnostic perturbation tests. Our findings reveal a high sensitivity of both inversion parameters, the rate factor $A$ and basal slipperiness $C$, as well as Grounding Line Flux (GLF) and Volume Above Floatation (VAF), to the grid resolution. We observed that for the AmIS, grid resolutions coarser than 10 km in our model introduce considerable noise and reduce the ability to capture realistic dynamic processes. Additionally, we found that the inversion parameters were transferable between grid systems of different resolutions, and the AmIS exhibited a consistent response to idealized collapse scenarios across these grid systems, particularly for grids with a resolution of 10 km or finer. This study extends the previous ice sheet model intercomparison experiments from synthetic topography to realistic geometry and provides insights for future ice modeling studies on grid systems configurations.