Return to Table of Contents


Laboratory observations of ice-floe processes made during long-term drift and collision experiments

Susan Frankenstein and Hayley Shen, Department of Civil and Environmental Engineering, Clarkson University, Potsdam, New York 13699-5710
 Experiments at a laboratory in New Hampshire are helping scientists understand the annual freezing of the southern oceans and the dynamics of floes at the ice edge.
This article describes visual observations made during long-term multifloe drift experiments which were carried out in the refrigerated wave tank at the U.S. Army Cold Regions Research and Engineering Laboratory. The tank's length, width, and height are 36.58 meters (m), 1.22 m, and 0.61 m, respectively. A paddle spanning the width of the tank was at the far upstream end, and a gravel beach with a 1:10 slope was at the downstream end. The purpose of these tests was to determine the drift velocity and collision frequency of individual ice floes and how these factors influenced the formation of a solid ice cover. Three different wave conditions were tested, each chosen to minimize the wave reflection from the beach. The resulting wave periods ranged from 1.71 to 2.73 seconds. The air temperature during these tests was -12C to -5C.

The ice floes used for these experiments were cut from a seeded ice sheet, grown under still-water conditions. The ice sheet was composed of randomly oriented, nearly spherical crystals whose average diameter was 1.1 millimeter (mm). The average thickness of the ice sheet was 1.26 centimeters (cm). These parameters were chosen to mimic newly formed pancake ice floes. Once the ice sheet was stiff enough to handle, individual floes with a width of 20 cm and a length spanning the width of the tank were cut. Seven long-term drift experiments were performed. The motion of the floes was recorded using two time-lapse videocameras. The tests lasted 10-12 hours. The number of floes cut for each test varied from 23 to 50. The average number of floes per wavelength ranged from 16 to 28. The water surface conditions in front of and behind the floes were also varied. For the first two tests, there was open water in front of and behind the cut floes. For the next test, there were random floes in front of and behind the cut floes. For the final four tests, there were random floes in front of and open water behind the cut floes. The test conditions are summarized in the table.
Test conditions
Test number Wave period (seconds) Water depth (m) Number of floes Front condition Back condition Average ice thickness (cm)
1 2.73 0.460 23 Open water Open water 1.10
2 2.73 0.445 30 Open water Open water 1.18
3 2.73 0.440 39 Random floes Random floes 1.41
4 1.71 0.430 33 Random floes Open water 1.17
5 1.71 0.437 47 Random floes Open water 1.14
6 1.71 0.450 42 Random floes Open water 1.68
7 2.16 0.435 50 Random floes Open water 1.15
To begin each test, the wave paddle was started, and the wave field was allowed to set-up. The floes were then held parallel to the wave front and as steady as possible to achieve an initial velocity of 0. After being released, the floes tended to twist slightly so that they were at an angle to the wave. The floes were not all oriented the same though. A floe's position relative to the wave was influenced by its neighbors unless there was open water between them. Thus, floe orientation tended to occur in groups. Occasionally, some floes were observed to rotate 90. These floes did not follow the wave surface because they were too rigid to bend. Some floes pivoted about their center. Most, though, began to oscillate back and forth. Thus, neighboring floes would come together and move apart. This caused water to be pumped onto the floes' surfaces resulting in the ice's surface becoming softer. This phenomenon was also observed in the field by Henderson (1962). If these floes were left undisturbed, this softer surface refroze, causing the floes to thicken.

Upon release, the floes were seen to drift downstream, toward the beach. At the same time, frazil formed in the open-water areas. Because some reflection occurred, there was always an open-water area at the beach where new frazil formed before being pushed upstream. Thus, the frazil was thinnest at the beach and thickest next to the floes. The force on the floes caused by this frazil growth was stronger than the wave drift force. This resulted in the floes being slowly pushed backward toward the paddle. This backward drift was not constant. Periods of no drift were interspersed with periods of drift as high as 0.00175 meters per second. The average drift velocity was 1-2·10-4 meters per second. The magnitude of the drift and the pattern of drift vs. no drift was not related to the initial test conditions.

Besides drifting, the floes were also seen to collide. If the floes had rough edges, the floes would stick together instead of bouncing apart after contacting one another. The measured restitution coefficient averaged 0.14 for the clean collisions. The collisions appeared to coincide with the peak of the wave. Thus, collisions would be seen progressing downstream from floe to floe much like the metal balls in a Newton's cradle. The frequency of collisions was approximately the wave frequency. Often, a floe was seen to collide with one neighbor several times then collide with its other neighbor before returning to the first neighbor.

Floes were also seen to raft onto one another. This was caused by the floe field's need to relieve the pressure build up that resulted from the wave drift force being opposed by the expanding frazil. As stated earlier, many of the floes' surfaces were softened due to water being pumped onto them. When two floes began to raft, the bottom of the upper floe would push its way onto the lower floe. This motion was oscillatory. Thus, the surface of the lower floe would be scraped off ahead of the upper floe. This would cause the lower floe to become further submerged, softening it further. This process continued until the upper floe had totally rafted over the lower floe. Multiple rafting involving three or four floes was often seen. Occasionally, the rafting process would result in one of the floes breaking. The crack occurred perpendicular to the point of contact and seemed to be a fatigue problem.

It was also seen initially that some floes were not colliding and had open water surrounding them. In these cases, the frazil that formed in the open water adhered to the floe. Over time, this new growth hardened and thickened. Frazil that formed in the open-water area between the floes and the beach was seen to be pushed under the floes, adhering to the floes' bottoms. Along the width of a floe, this adhesion of frazil crystals was thinnest at the edges and thickest in the middle, creating a parabolic shape. Up to 6 cm of edge and bottom growth as a result of frazil adhesion was observed in a period of 2-3 hours. These same phenomena were observed in the Weddell Sea (Wadhams, Lange, and Ackley 1987; Wadhams and Holt 1991).

As has been mentioned previously, frazil formed in the open-water area between the beach and the floes. The floes essentially formed a solid barrier. Frazil could move beyond this only if it got swept underneath the floes, as was discussed in the previous paragraph. Otherwise, the frazil collected against the flow farthest downstream. As new frazil was being added at the beach, the older frazil abutting the floes became denser. The progressive wave action caused the frazil slurry to coalesce into small clumps or pancakes. These pancakes gradually grew in size and became stiffer. At this point, collisions between neighboring pancakes resulted in the formation of raised edges around the perimeter of the pancake due to the pumping of frazil crystals and water onto its surface. These raised edges formed only if there was some stiffness to the floe. Stiffer floes were observed to have higher edges. These pancakes were observed to freeze together and to raft, creating a thicker, larger, more solid ice cover. This ice cover was thinnest and softest at the beach where the wave action was greatest. These results suggest that the fact that the floes created a boundary beyond which the frazil could not move was important to the formation of the pancakes. They also suggest that wave action was important to this process.

The formation of pancakes from frazil and the eventual development of an ice cover from the pancakes have been observed to be the main ice-cover formation process in the southern oceans (Wadhams et al. 1987; Lange et al. 1989; Wadhams and Holt 1991). Other phenomena seen in the field, such as the pumping of water onto a floe's surface, were also successfully reproduced in the laboratory. New insight concerning the rafting and collision processes between neighboring floes was gained through these tests. Thus, it is seen that simple laboratory experiments are important tools in increasing our understanding of the freezing of the southern oceans and floe dynamics at the ice edge.

Thanks are given to John Gagnon for his technical assistance during the experiments. A Research Experience for Undergraduates student, Chris Moore, also assisted with the laboratory work. This study was supported by National Science Foundation grant OPP 92-19165.


Henderson, J.A. 1962. Study of natural forces acting on floating ice fields (U.S. Naval Civil Engineering Laboratory Contract Report NBy-32215). U.S. Naval Civil Engineering Laboratory: Port Hueneme, California.

Lange, M.A., S.F. Ackley, G.S. Diekmann, H. Eikenn, and P. Wadhams. 1989. Development of sea ice in the Weddell Sea Antarctica. Annals of Glaciology, 12, 92-96.

Wadhams, P., and B. Holt. 1991. Waves in frazil and pancake ice and their detection in Seasat synthetic aperture radar imagery. Journal of Geophysical Research, 96(C5), 8835-8852.

Wadhams, P., M.A. Lange, and S.F. Ackley. 1987. The ice thickness distribution across the Atlantic sector of the antarctic ocean in midwinter. Journal of Geophysical Research, 92(C13), 14535-14552.