Modelling the evolution of the hydrothermal structure of polythermal glaciers in Svalbard

Svalbard is among the fastest warming regions on Earth, with mean air temperatures rising several times faster than the global average. Approximately 57% of the archipelago remains glacierized, and most of these glaciers are polythermal, containing both cold and temperate ice layers. Understanding their response to ongoing and future climate change requires physically-based thermomechanical modelling capable of capturing the evolution of internal ice temperatures and cold–temperate transitions.

In this study, we apply the Instructed Glacier Model (IGM), an open-source, Python-based glacier model that integrates climate-driven surface mass balance, ice-flow and heat transfer processes. IGM further employs physics-informed machine learning and GPU acceleration to efficiently resolve high-order ice-flow dynamics, enabling large-scale simulations at high spatial resolution.

Svalbard benefits from extensive ground-penetrating radar (GPR) datasets, providing rare observational constraints on the cold–temperate transition surface (CTS). We exploit multi-epoch GPR observations to evaluate the ability of IGM thermodynamics to reproduce the observed CTS evolution. As a first step in a broader PhD project aiming to simulate the evolution of all land-terminating Svalbard glaciers under different greenhouse gas emission scenarios, we focus on Werenskioldbreen, a well-instrumented glacier with repeated GPR surveys (1998, 2008, 2016, 2024) and long-term mass-balance records. This work provides a crucial benchmark for improving thermomechanical modelling of polythermal glaciers and contributes to reducing uncertainties in projections of Svalbard glacier change.