Data-Driven Inference of the Mechanics of Slip Along Glacier Beds Using Physics-Informed Neural Networks: Case Study on Rutford Ice Stream, Antarctica

Reliable projections of sea-level rise depend on accurate representations of how fast-flowing glaciers slip along their beds. The mechanics of slip are often parameterized as a constitutive relation (or sliding law) whose proper form remains uncertain. Here, we present a novel deep learning-based framework for learning the time evolution of drag at glacier beds from time-dependent ice velocity and elevation observations. We use a feedforward neural network, informed by the governing equations of ice flow, to infer spatially and temporally varying basal drag and associated uncertainties from data. We test the framework on 1D and 2D ice flow simulation outputs and demonstrate the recovery of the underlying basal mechanics under various levels of observational and modeling uncertainties. We apply this framework to time-dependent velocity data for Rutford Ice Stream, Antarctica, and present evidence that ocean-tide-driven changes in subglacial water pressure drive changes in ice flow over the tidal cycle. Plain Language Summary The relation between slip of glaciers along their beds and the level of basal drag at the ice-bed interface is a critical component of ice dynamics for fast-flowing glaciers and ice streams. However, uncertainty surrounding the proper form of this relation, often referred to as the sliding law, has hindered efforts to reliably project the contribution of the Greenland and Antarctic ice sheets to future sea-level rise. Here, we utilize the tools of physics-informed deep learning to learn the evolution of drag at glacier beds from time-dependent ice velocity and elevation observations. By training a neural network with both data reconstruction losses and ice physics-based losses, we are able to reconstruct the evolution of drag for glaciers and ice streams undergoing changes in flow speed and surface elevations. Thus, we can investigate the relation between slip and basal drag without specifying the form of the sliding law. We use this approach to present observational evidence that ocean-tide-driven changes in flow speed for Rutford Ice Stream, Antarctica are driven by changes in subglacial water pressure. Ultimately, this approach provides a natural way to integrate our existing knowledge of ice flow physics with remote sensing data in order to improve flow models.

Automated summary

Reconstructing the evolution of drag at glacier beds from time-dependent ice velocity and elevation observations using deep learning provides a new way to investigate the relationship between slip of glaciers and basal drag, without needing to specify the form of the sliding law. This approach integrates existing knowledge of ice flow physics with remote sensing data to improve flow models and provides insights into the mechanisms driving changes in ice flow dynamics.