The influence of double-diffusive processes on the melting of ice in the Arctic Ocean: laboratory analogue experiments and their interpretation

This study has been motivated by two oceanographic observations: an increased rate of melting of sea ice in the Arctic Ocean, and the advance of an anomalously warm tongue of Atlantic water across the Arctic below the halocline over the last few decades. A series of laboratory experiments has been carried out in order to explore the physical principles underlying these phenomena, and the possibility that the extra heating at depth is responsible for the enhanced melting rate. A tank was filled with salt solution having various constant vertical density gradients. A block of ice one third of the length of the tank was floated on the surface at one end, and the rest of the surface and the walls of the tank were insulated. When no extra heat was supplied the melting rate (loss of weight of the ice in I h) systematically decreased as the stratification was changed from homogeneous fluid to increasingly large density gradients, while keeping the salinity of the solution in contact with the ice constant. An analogue of the intruding Atlantic water was produced by heating the lower portion of the vertical end wall at the end of the tank opposite to the ice end, keeping its temperature constant, and using the same range of salinity gradients as in the unheated experiments. Again the melting rate decreased as the density gradient was increased, but for low gradients it was larger than that in the unheated experiments. Above a certain intermediate gradient there was no significant difference in melting rate between the unheated and heated runs. The melting data were supplemented by photographs and vertical temperature and salinity profiles. The upward transfer of heat from the body of the fluid to melt the ice was clearly double-diffusive: overturning layers, separated by 'diffusive' interfaces, were visible on shadowgraphs, and the thickness of the layers decreased as the density gradient increased. The mean thickness of the layers through the depth of the tank also systematically decreased as the density gradient increased. With weak gradients an extra heat flux to the ice came from the intruding heated layer, but at large gradients this tongue of warm water at depth did not add to the flux near the surface. Though they were obtained in a simple, arbitrary and fixed geometry, we believe that the results of these experiments can be used as the basis for a better physical understanding of the melting rates of ice in the Arctic under various conditions. (C) 2003 Elsevier B.V. All rights reserved.