Imagine you are standing by a small stream. On the channel bed you might see individual sediment grains moving downstream. This is what river erosion looks like at the grain-scale and by counting how many particles move at a given point in space and time, you can measure how quickly that stream is eroding the landscape. Now imagine the water stops flowing, the channel bed freezes and then water begins flowing over the now frozen bed again. Do you think more or less grains will move? And how do you think this will change as the bed thaws?
This thought experiment inspired our initial laboratory experiments to quantify how pore ice in a riverbed changes the erodibility of its sediment. Polar and high mountain cold regions on Earth are undergoing a strong seasonal cycle where the ground freezes fully during the winter and thaws from the surface downward every summer. In this project, we are interested in how this seasonal cycle is changing when, where and how new river channels form.

The coupled-thaw entrainment model

As a channel bed thaws, the frozen-unfrozen interface in the bed (thaw front) acts as an impermeable boundary, trapping any water flow within the top layer. In the early thaw season, when the thaw front is shallow, surface water injections hit the thaw front and are redirected upwards, resulting in overturning convective stirring within the bed. This convection has two effects. First, it disrupts the bed structure generating a buoyancy force that helps pick up more sediment particles, enhancing erosion. Second, it delivers heat directly to the thaw front, locally enhancing thawing which causes the thaw front to become undulated. These undulations generate hydrostatic pressure gradients that localise erosion and drive the formation of stepped topography on the bed surface. These undulations continue to grow until the thaw front is deep enough that surface water injections do not reach it anymore. However, the undulations continue to localise erosion for the remainder of the season.