A LABORATORY EXPERIMENT OF ICE MELT BY PRESSURIZED TURBULENT WATER FLOW (Invited Presentation)

As water flows in contact with an ice surface, the heat carried by the water, in addition to that created by viscous dissipation, can melt the ice. This process is of critical importance for glaciology as it controls glacial hydrology and melt at the ice–ocean interface. Glacial hydrology is pivotal to ice dynamics, while melt at the ice–ocean interface may hold the key to understanding the stability of marine-terminating glaciers. Our physical understanding of the heat exchange between turbulent water flow and ice is, however, quite poor. From current theories, we expect that ice melt rate scales with water temperature and velocity, but such theories are unable to reproduce field observations.

We built a laboratory experiment emulating pressurized water flow at the bed of a glacier to explore the physical controls of heat exchange between flowing water and ice. We designed an ice flume in which we created and maintained a block of ice ~3m long and 0.3 m high and wide. A chiller recirculated a glycol mixture below freezing and, paired with extensive insulation, kept the ice block frozen. We emulated the glacier bed with fixed roughness elements glued onto a lid. The lid was fixed and sealed on top of the ice, and water flowed under pressure at the interface between the two surfaces. We ran six experiments to test the effect of different water temperatures and lid roughnesses. In every experiment, the water flow started by melting scallops which size was commensurate with that of the roughness elements. The size of the scallops then grew, and the scallops migrated down-flow. When the lid was smooth, scallops only developed spawning from local imperfections in the experimental set-up. Our findings confirm that ice melt rates scale linearly with water temperature, and that water temperature controls melt rates to the first order. However, we found that water velocity fails to explain variations in temperature-normalized melt rates. Instead, we found that ice melt rates are a function of skin friction boundary stress and a roughness-height Reynolds number owing to the rough boundary from the scalloped ice surface. We suggest that including a rough boundary layer in theories of ice melt will help bridge the current gap between model predictions and observations.