Where, when, and why do frozen landscapes erode faster than unfrozen ones?

A fundamental research question of periglacial geomorphology is whether and how ground ice interacts with surface water flow to change grain-scale sediment transport dynamics, and how those processes translate to the landscape-scale, potentially changing the timing and intensity of erosion events. Polar deserts in the Canadian High Arctic serve as ideal field laboratories to isolate the effects of ground ice on particle transport due to a lack of vegetation and by being largely undisturbed since the last glacial maximum.

During the Summer 2024 field season at the Flying Squirrel polygon field on Devon Island, we observed a complex of pools interconnected by channelised polygon troughs and relatively steeper relict gravel deposits, as well as evidence of recent transport of gravel- and sand-sized particles. However, despite visiting during a storm event, we did not observe active transport and only limited surface water run-off. As such, the timing and magnitude(s) of the flow events  that caused the gravel deposits are unclear, nor do we know the thermal state of the bed during the time of transport. 

To investigate this research gap, we conducted flume experiments with an initially frozen bed under rarefied transport conditions to investigate at what thermal state sediment transport is favoured and compare the bed behaviour to unfrozen experiments. Particle flux is maximised at the start of the frozen experiments before decaying to an approximate steady-state background flux similar to the unfrozen experiments, following a power law with the relationship . At early stages of the frozen experiments, hydraulic jumps develop in concert with variations of the local thaw depth, which result in enhanced particle entrainment and relatively rapid thawing downstream, as the hydraulic jumps migrate upstream. Beneath hydraulic jumps, we observe forced injections of water into the partially-frozen bed, which can spread laterally along an evolving thaw front. Depending on the thaw front depth, the combined effects of locally-intensified melting, increased pore pressure and mechanical disruption of the bed can enhance particle entrainment locally and increase the overall erosion rate compared to unfrozen experiments. Enhanced rates of particle entrainment continue until hydraulic jump activity diminishes and the injected surface water no longer penetrates to the thaw front. Accordingly, we develop a maximum injection depth, which is strongly dependent on the local permeability, as well as the jump height.

Our experimental results show that the thermal state of the bed can have a strong influence on the local entrainment rate at the grain-scale with entrainment being promoted at a shallow thaw depth. We hypothesise that this sensitivity also translates to the landscape-scale, where all water has to travel as overland flow when the active layer is thin, whereas much of the water supply can be compensated as subsurface flow late in the summer, minimising particle entrainment. This could explain the lack of active erosion at the Flying Squirrel polygon field during the late summer when the active layer was approximately 1m deep.