One of the major challenges in modelling climate change is to accurately predict the melting rate of icebergs and glacier, as this is connected to the albedo, the ocean currents, and ocean level. Current predictions for the melting of glaciers can be off by a factor of 100, and different melting models show inconsistencies—no general consensus of the cryospheric modeling has been reached yet.
The difficulties in describing the entire melting process stems from its multiscale nature (micrometers to kilometers) and the interaction between thermal, solutal, and viscous boundary layers and their complex interplay with the continuously reshaping (due to melting) boundary. To add, it is commonly believed that melting always smooths the shape. However, from examples in nature and from theoretical analysis, it is clear that flows around melting objects can create a rough (dimpled or 'scalloped') surface, dramatically increasing the difficulty of accurate predictions due to the complicated shape-flow interplay.
The complexity of such a system is immense. The heat-flux (describing the melting rate) depends on many factors and fully simulating the entire process using computers is not possible for the foreseeable future. The best one can do is to either study select naturally-occurring instances of icebergs and glaciers by means of field measurements—with its concomitants imperfections and uncontrollabilities—or to do the largest-attainable well-controlled experiments in the laboratory and to use the latest direct numerical simulations, to understand the fundamental underlying mechanisms, and from that predict to larger scales. This bottom-up approach has been highly successful in pipe and wind-tunnel flows, thermal convection, and rotating turbulence, and we believe this approach will also be successful here, and will allow us to predict transitions and scaling so desperately needed to predict these types of processes.
The aim of this research is therefore to investigate the effect of roughness on the melting process. And we investigate this using well-controlled laboratory experiments complemented by direct numerical computer simulations of these problems.
