A Physics-based Framework to Infer Firn Properties on Antarctic Ice Shelves from ASCAT Observations

Abstract. The stability of Antarctic ice shelves is closely linked to the properties of the firn layer, which regulates meltwater retention and influences ice shelf vulnerability to hydrofracturing. Firn Densification Models (FDMs) provide valuable insights into the firn structure, but key properties such as grain size are often parameterized using simple approximations, leading to significant uncertainties, especially in regions lacking in-situ validation. Here, we leverage 15 years (2007–2021) of active microwave observations from the C-band Advanced Scatterometer (ASCAT) to infer an effective firn grain-size parameter across Antarctic ice shelves. Within this framework, we use the Institute for Marine and Atmospheric Research Utrecht Firn Densification Model (IMAU-FDM) to prescribe the state of the firn layer (layer thickness, density, temperature, and liquid water content) and couple it with the Snow Microwave Radiative Transfer (SMRT) model to simulate radar backscatter. Grain size is treated as an unknown microstructural parameter and is optimized by minimizing the misfit between FDM–SMRT simulations and ASCAT observations. The framework is further used to examine how the sensitivity of ASCAT backscatter varies across firn regimes, and how this influences the interpretation of backscatter in terms of FAC. Our results show broad consistency between optimized effective grain size and IMAU-FDM estimates in high-Firn Air Content (FAC) regions, where ASCAT backscatter is most sensitive to interannual variability in grain size. In contrast, larger discrepancies emerge in intermediate- to low-FAC regions, particularly on the Amery Ice Shelf, where ice saturation in the firn (pore-space depletion) influences grain growth. The sensitivity experiments indicate that inversion constraints are regime-dependent, with the strongest sensitivity to grain size in intermediate firn regimes and weaker constraints in strongly depleted firn. Furthermore, by statistically reducing grain-size-driven scatter after inversion, we present a proof of concept for a more interpretable backscatter–FAC relationship. These findings provide a basis for improving firn model parameterization and advancing large-scale monitoring of firn evolution across Antarctic ice shelves in a warming climate.