Testing applicability of homogenization techniques for understanding macroscopic subglacial soil properties from microscale porewater-sediment interactions

Permeability is a crucial macroscopic quantity to describe porous flows, but can be difficult to estimate through first principles, especially for geophysical granular media due to its high spatiotemporal variability at many scales. Upscaling techniques can offer a generalizable framework of interpretable models that connect flow laws; for example, homogenization of Stokes flow through grains leads to Darcian flow. However, these techniques are often overlooked in geophysics in favor of laboratory measurements on samples or empirical parametrizations. Empirical techniques have limited interpretability, meaning that they leave uncertainty in their generalizability both spatially and temporally.

In this work, we investigate how accurate and applicable estimates from sequential homogenization of first-principles physics can be for the dynamics of subglacial soils when tested against laboratory measurements. We approximate the grain size distribution of subglacial soils with a few scale-separated modes with consideration of real grain shape and packing distributions. We test model predictions of permeability against laboratory measurements in bimodal packings of glass beads, and test against subglacial soil samples to assess the generalizability of upscaling idealized first-principles intuition. We then apply model results to understand how pore-altering and sediment-crushing macroscopic forces like shears or compactions can affect permeability and soil strength.