How Does Arctic Sea Ice Form and Decay?

Peter Wadhams
Professor of Ocean Physics, Scott Polar Research Institute, University of Cambridge, UK
(from 1 Jan 2003: Dept. of Applied Mathematics and Theoretical Physics)

Sea ice in Winter in the Beaufort Sea
Sea ice in Winter in the Beaufort Sea

Sea ice occupies about 7% of the area of the world ocean, and is of enormous importance climatically because it reflects most of the solar radiation that falls on it, affecting the average albedo of the earth, and also because it interposes a solid layer between the ocean and the atmosphere which reduces the free transfer of heat and moisture between the two. Observational evidence at the moment tells us that the sea ice in the Arctic (although not in the Antarctic) is retreating and thinning, and computer models predict that by the 2080s the ice cover will completely disappear in summer, so it is important for us to understand the mechanisms by which sea ice forms and decays.

Why does ice float?
We first have to account for the fact that ice floats on water at all, since ice is one of very few substances where the solid is less dense than its molten form. This is because the structure of normal ice, called ice I, is an open hexagonal structure. Each oxygen atom is at the centre of a tetrahedron with four other O atoms at the apices. The O atoms are concentrated close to a series of parallel planes that are known as the basal planes. The principal axis, or c-axis, of the crystal unit cell lies perpendicular to the basal plane. The whole structure looks much like a beehive, composed of layers of slightly crumpled hexagons. The net of O atoms is held together by hydrogen bonds. The H atoms lie along these bonds. It is the length of the hydrogen bond that creates the open structure of ice; when ice melts, some of the bonds are broken, causing a disordered structure with a higher density. But even in liquid water some short-range order remains, with a few water molecules retaining the crystal-like bonded structure until this is destroyed by thermal motion; this causes a curious density behaviour in fresh water, where there is a maximum density at 4°C.

Cooling the water down
Consider a fresh water body being cooled from above, for instance a lake at the end of summer experiencing subzero air temperatures. As the water cools the density increases so the surface water sinks, to be replaced by warmer water from below, which is in its turn cooled. This creates a pattern of convection through which the whole water body gradually cools. When the temperature reaches 4°C, the lake reaches its maximum density. Further cooling results in the colder water becoming less dense and staying at the surface. This thin cold layer can then be rapidly cooled down to the freezing point, and ice can form on the surface even though the temperature of the underlying water may still be close to 4°C. Thus a lake can experience ice formation while considerable heat still remains in the deeper parts.

This does not apply to sea water. The addition of salt to the water lowers the temperature of maximum density, and once the salinity exceeds 24.7 parts per thousand (most Arctic surface water is 30-35), the temperature of maximum density disappears. Cooling of the ocean surface by a cold atmosphere will therefore always make the surface water more dense and will continue to cause convection right down to the freezing point - which itself is depressed by the addition of salt to about -1.8°C for typical sea water. It may seem, then, that the whole water column in an ocean has to be cooled to the freezing point before freezing can begin at the surface, but in fact the Arctic Ocean is composed of layers of water with different properties, and at the base of the surface layer there is a big jump in density (known as a pycnocline), so convection only involves the surface layer down to that level (about 100-150 metres). Even so, it takes some time to cool a heated summer water mass down to the freezing point, and so new sea ice forms on a sea surface later in the autumn than does lake ice in similar climatic conditions.

How ice forms in calm water
In quiet conditions the first sea ice to form on the surface is a skim of separate crystals which initially are in the form of tiny discs, floating flat on the surface and of diameter less than 2-3 mm. Each disc has its c-axis vertical and grows outwards laterally. At a certain point such a disc shape becomes unstable, and the growing isolated crystals take on a hexagonal, stellar form, with long fragile arms stretching out over the surface. These crystals also have their c-axis vertical. The dendritic arms are very fragile, and soon break off, leaving a mixture of discs and arm fragments. With any kind of turbulence in the water, these fragments break up further into random-shaped small crystals which form a suspension of increasing density in the surface water, an ice type called frazil or grease ice. In quiet conditions the frazil crystals soon freeze together to form a continuous thin sheet of young ice; in its early stages, when it is still transparent, it is called nilas. When only a few centimetres thick this is transparent (dark nilas) but as the ice grows thicker the nilas takes on a grey and finally a white appearance. Once nilas has formed, a quite different growth process occurs, in which water molecules freeze on to the bottom of the existing ice sheet, a process called congelation growth. This growth process yields first-year ice, which in a single season in the Arctic reaches a thickness of 1.5-2 m.

How ice forms in rough water
If the initial ice formation occurs in rough water, for instance at the extreme ice edge in rough seas such as the Greenland or Bering Seas, then the high energy and turbulence in the wave field maintains the new ice as a dense suspension of frazil, rather than forming nilas. This suspension undergoes cyclic compression because of the particle orbits in the wave field, and during the compression phase the crystals can freeze together to form small coherent cakes of slush which grow larger by accretion from the frazil ice and more solid through continued freezing between the crystals. This becomes known as pancake ice because collisions between the cakes pump frazil ice suspension onto the edges of the cakes, then the water drains away to leave a raised rim of ice which gives each cake the appearance of a pancake. At the ice edge the pancakes are only a few cm in diameter, but they gradually grow in diameter and thickness with increasing distance from the ice edge, until they may reach 3-5 m diameter and 50-70 cm thickness. The surrounding frazil continues to grow and supply material to the growing pancakes.

At greater distances inside the ice edge, where the wave field is calmed, the pancakes may begin to freeze together in groups and eventually coalesce to form first large floes, then finally a continuous sheet of first-year ice known as consolidated pancake ice. Such ice has a different bottom morphology from normal sea ice. The pancakes at the time of consolidation are jumbled together and rafted over one another, and freeze together in this way with the frazil acting as "glue". The result is a very rough, jagged bottom, with rafted cakes doubling or tripling the normal ice thickness, and with the edges of pancakes protruding upwards to give a surface topography resembling a "stony field". The rough bottom is an excellent substrate for algal growth and a refuge for krill. The thin ice permits much light to penetrate, and the result is a fertile winter ice ecosystem.

In the Arctic, a key area where pancake ice forms the dominant ice type over an entire region is the so-called Odden ice tongue in the Greenland Sea. The Odden (the word is Norwegian for headland) grows eastward from the main East Greenland ice edge in the vicinity of 72-74°N during the winter because of the presence of very cold polar surface water in the Jan Mayen Current, which diverts some water eastward from the East Greenland Current at that latitude. Most of the old ice continues south, driven by the wind, so a cold open water surface is exposed on which new ice forms as frazil and pancake in the rough seas. The salt rejected back into the ocean from this ice formation causes the surface water to become more dense and sink, sometimes to great depths (2500 m or more), making this one of the few regions of the ocean where winter convection occurs, which helps drive the entire worldwide system of surface and deep currents known as the thermohaline circulation (or "Great Ocean Conveyor Belt").

Growth of the ice
Once a continuous sheet of nilas has formed, the individual crystals which are in contact with the ice-water interface grow downwards by freezing of water molecules onto the crystal face. This freezing process is easier for crystals with horizontal c-axes than for those with c-axes vertical. The crystals with c-axis horizontal grow at the expense of the others, and as the ice sheet grows thicker crowd them out in a form of crystalline Darwinism . Thus the crystals near the top of a first-year ice sheet are small and randomly oriented, then there is a transition to a fabric composed of long vertical columnar crystals with horizontal c-axes. This columnar structure is a key identifier of congelation ice (i.e., ice which has grown thermodynamically by freezing onto an existing ice bottom), and is a striking feature of first-year ice even when viewed by the naked eye. The ions of the salts in sea water cannot enter the crystal structure despite its open nature. One might expect all salt to be rejected, therefore, leading to a sea ice cover composed of pure ice. Such is not the case, however. If you suck on a piece of first-year sea ice it will taste distinctly salty. The water from young sea ice may have a salinity of about 10 parts per thousand, dropping to 1-3 in old ice. How does this salt get into the ice?

The answer lies in the way that the ice sheet grows. The ice-water interface advances in the form of parallel rows of cellular projections called dendrites. Brine rejected from the growing ice sheet accumulates in the grooves between rows of dendrites. As the dendrites advance, ice bridges develop across the narrow grooves that contain the rejected brine, leaving the brine trapped and isolated. The walls of the "prison" close in through freezing, until the salt is contained in a very small cell of highly concentrated brine, concentrated enough to lower the freezing point to a level where the surrounding walls can close in no further. The cell then remains, a tiny inclusion. They eventually drain out of the ice, by way of a network of brine drainage channels which they create, and as the ice sheet ages the brine concentration drops. These channels have a biological role. Phytoplankton have been observed to live on their walls, and even larger zooplankton such as amphipods have been observed to crawl up the larger channels. Within a channel there is possibly a higher light level than on the ice bottom, because of the waveguide effect of the channel for light penetrating from above, while the oscillating water flow brings nutrients and oxygen to the resident biological community. In addition the tube provides security from larger browsers.

The summer melt period

Melting sea ice in Spring in the Bering Sea
Melting sea ice in Spring in the
Beaufort Sea

In the Arctic, the overlying snow layer typically begins to melt in mid-June and is gone by early July. The meltwater from the snow gathers to form a network of meltwater pools over the surface of the ice. On first year ice, which has a smooth upper surface at the end of winter (except where ridged), the pools are initially very shallow, forming in minor depressions in the ice surface, or simply being retained within surviving snow pack as a layer of slush. As summer proceeds, however, this initial random structure becomes more fixed as the pools melt their way down into the ice through preferential absorption of solar radiation by the water, which reflects only 15-40% of the radiation falling on it compared to 40-70% for bare ice.

As the melt pools grow deeper and wider they may eventually drain off into the sea, over the side of floes, through existing cracks, or by melting a thaw hole right through the ice at its thinnest point or at the melt pool's deepest point. The downrush of water when a thaw hole opens may be quite violent, and on very level ice, such as fast ice, a single thaw hole may drain a large area of ice surface. From the air such thaw holes give the appearance of "giant spiders", with the "body" being the thaw hole and the "legs" channels of melt water draining laterally towards the hole.

The underside of the ice cover also responds to the surface melt. Directly underneath melt pools the ice is thinner and is absorbing more incoming radiation. This causes an enhanced rate of bottom melt so that the ice bottom develops a topography of depressions to mirror the melt pool distribution on the top side. In this way an initially smooth first-year ice sheet acquires by the end of summer an undulating topography both on its top and bottom sides. Some of the drained melt water may in fact gather in the underside depressions to form under-ice melt pools, which refreeze in autumn and partially smooth off the underside, leaving it with bulges but not depressions.

A final and most important role of the melt water is that some of it works its way down through the ice fabric through minor pores, veins and channels, and in doing so drives out much of the remaining brine. This process, called flushing, is the most efficient and rapid form of brine drainage mechanism, and it operates to remove nearly all of the remaining brine from the first-year ice. The hydrostatic head of the surface meltwater provides the driving force, but an interconnecting network of pores is necessary for the flushing process to operate. Given that the strength properties of sea ice depend on the brine volume, this implies that the flushing mechanism creates a surviving ice sheet which during its second winter of existence has much greater strength than in its first winter.

What happens to the ice that survives?
Ice which has survived one or more summer seasons of partial melt is called multi-year ice. In the Arctic, sea ice commonly takes several years to either make a circuit within the closed Beaufort Gyre surface current system (7-10 years) or else be transported across the Arctic Basin and expelled in the East Greenland Current (3-4 years). More than half of the ice in the Arctic is therefore multi-year ice. Growth continues from year to year until the ice thickness reaches a maximum of about 3 metres, at which point summer melt matches winter growth and the thickness oscillates through an annual cycle. This old, multi-year ice is much fresher than first-year ice; it has a lower conductivity and a rougher surface. The low salinity of multi-year ice makes it much stronger than first-year ice and a formidable barrier to icebreakers.

Ice doesn't just grow and melt
In this essay we have been concerned with the way in which sea ice forms and changes under thermal processes alone. Yet we know that pack ice is constantly in motion, driven by the wind, and that this produces many important changes to its appearance and development. The two most obvious features that this creates are leads and pressure ridges.

The wind stress which drives the sea ice through frictional drag is integrated over a large area - it has been estimated that in concentrated pack ice a piece of sea ice responds to wind fields integrated over a distance of 400 km upwind. Therefore a large-scale divergent wind field, created by an appropriate pressure pattern, can also create a divergent stress over a large area of icefield. Since ice has little strength under tension, this divergence can open up cracks which widen to form leads. In winter leads rapidly refreeze because of the enormous temperature difference between the atmosphere (typically -30°C) and the ocean (-1.8°C). The heat loss from a newly-opened lead can be so violent (more than 1000 W m-2) that the lead steams with frost smoke from the evaporation and condensation of the surface water. A young ice cover rapidly forms, within hours, as nilas if the surface is calm, and this cuts out the evaporation. When a subsequent wind stress field becomes convergent, the young ice in the refrozen leads forms the weakest part of the ice cover and is the first part to be crushed, building up heaps of broken ice blocks above and below the water line. Such a linear deformation feature is called a pressure ridge, the above-water part being the sail and the below-water part (more extensive) being called the keel. Keels in the Arctic can reach down to 50 m, although most are about 10-25 m deep. Ridged ice in the Arctic makes a major contribution to the overall mass of sea ice; probably about 40% on average and more than 60% in coastal regions.

Because of leads and ridges the "landscape" of the Arctic Ocean is an ever-changing panorama of white fields separated by white hedges and walls (ridges) , and with rivers and streams (leads) appearing at unpredictable locations. The whole array passes through a seasonal cycle of growth and decay which, if present trends continue, will eventually lead to the disappearance of the ice and an open Arctic Ocean.


The crystal structure of ice

The temperature of the density maximum (Tpmax) and the freezing point (Tf) of seawater as a function of salinity S, showing how they meet at 27.4 psu. Density contours of sea water are also shown.

Pancake ice

Consolidated pancake ice in the Antarctic (continuous sheet)

The Odden ice tongue in the Greenland Sea on February 12, 1993, shown by satellite passive microwave imagery. The different colours represent different percentage sea ice covers.

Sheets of brine cells in sea ice, seen under a microscope. Spacing of cells about 0.6 mm.

Geometry of a typical brine drainage channel

Frost smoke from a freshly opened lead in winter

The sail of an Arctic pressure ridge

The keel (underwater part) of an Arctic pressure ridge

Ice Photographs

Ice photographs from a ship in the Ross Sea (Antarctica): Grease ice, pancake ice, and pack ice from the Virginia Institute of Marine Science at the College of William and Mary

KidSat: Sea Ice images from space

Alfred Wegener Institute - Sea ice

Pancake ice in Bering Sea

Closeup of pancake ice in the Bering Sea

Ice floes in the Bering Sea

Closeup of ice floes in the Bering Sea

Spring melt

Winter sea ice in the Beaufort Sea

Sea ice off of Tigvariak Island

Melting ice on Beaufort Sea

Jelly fish floating under arctic ice

Other Essays on Sea Ice

Changes in Arctic sea ice over the past 50 years: Bridging the knowledge gap between the scientific community and the Alaska Native community, Rob Mattlin and Henry Huntington

What changes have occurred in Arctic sea ice volume and dynamics over the past 50 years? Norbert Untersteiner

How might sea level be affected by changes in the Arctic land ice? Roger Barry

What do we know about organisms which thrive in the Arctic sea ice? Christopher Krembs and Jody Deming

What is the difference between rubble ice and pressure ridges? Hajo Eicken


Back to Essays  Back to top