Advancing Radar Sounding Attenuation Estimates with Frequency-Based Techniques

Attenuation rates derived from radar data offer valuable insights into subsurface ice sheet conditions, revealing information about the ice sheet temperature, chemical composition, and physical structure. Accurate attenuation estimates are also essential for interpreting basal conditions. However, established methods for estimating attenuation rates perform poorly in certain ice sheet regions, and uncertainties remain in the physical interpretations of attenuation results.

In this study, we develop a novel frequency-based method for deriving ice sheet attenuation rates, adapting techniques from planetary radio science and seismology. We apply this method to airborne radar sounding data collected across multiple Antarctic basins, enabling new interpretations of the englacial and subglacial environment in regions where subsurface information is sparse. Not only do these frequency-based attenuation estimates offer valuable englacial and subglacial insight in new regions of the Antarctic ice sheet, but we show how leveraging the attenuation results to train neural networks can facilitate predictions and constraints on subglacial conditions. Such constraints are useful for better resolving subsurface processes in numerical models. Our study highlights the potential of advancing conventional geophysical methods in combination with AI-driven approaches and model validation to enhance our understanding of ice sheet subsurface conditions and ice dynamics.