Inverting for Firn Aquifer Properties from Ice-Penetrating Radar Data

Perennial firn aquifers may play an important role in modulating the surface mass balance and ice dynamics of the Greenland Ice Sheet by storing surface meltwater in a porous firn layer. Ice-penetrating radar has been shown to be an effective tool for mapping the spatial extent and thickness of these subsurface water bodies. However, quantitative estimates of aquifer properties from radar observations require extensive a priori knowledge of the local firn structure and electrical properties of the aquifer water. Here, we develop a Markov Chain Monte Carlo joint inversion of radar reflectivity and attenuation to simultaneously estimate firn structure and aquifer properties and their associated uncertainty. We show that using this inversion method, the only additional field measurement required to robustly and accurately constrain the total water storage is the conductivity of the aquifer water. However, the inversion is largely insensitive to firn structure parameters, and aquifer properties like thickness and bulk porosity are poorly constrained on an individual basis due to the large number of solutions that map to the same total water storage. Regardless, our results suggest that, when combined with field measurements of water conductivity, ice-penetrating radar surveys can be an effective method for remotely characterizing kilometer-scale spatial and temporal variations in firn aquifer storage capacity.