Portrait of the ice at Ilulissat

Signs of earlier advances and retreats of the Inland Ice margin are widespread in the Ilulissat Icefjord area. The present day ice-free strip of land bordering the Inland Ice was glaciated at the culmination of the last ice age about 21.000 years ago.

Around Kangia signs of glaciation date mainly from the last ice age, which started 115,000 years ago and ended 11,550 years ago. There are also many traces of the subsequent deglaciation, when the ice margin retreated from the present coastal areas. The retreat ended about 5000 years ago, and since then the ice margin has shown a slight advance. The cool period culminated in the Little Ice Age 500-100 years ago, when the ice margin reached its most recent maximum. The expansion occurred in a series of waves, each pulse resulting in an increase in the area covered by ice.

During the last ice age the Inland Ice probably covered the continental shelf and the whole of the Ilulissat area. Loss from the margin of the ice sheet occurred by way of major ice streams that produced calf ice. One of the largest ice streams occupied a trough more than 800 metres deep, between Qeqertarsuaq and Aasiaat in the south-west part of Disko Bugt.


Reconstruction of ice margin positions in the Ilulissat area 9500, 8000, 5000-4000 and 150 years ago. The change in the ice margin between the Little Ice Age maximum and the current extent of the ice is shown in brown for land areas and shaded for the floating part of Sermeq Kujalleq.


Water depths in Disko Bugt and Davis Strait. The Egedesminde Dyb trench between Qeqertarsuaq and Aasiaat marks the position of a former ice stream.

The margin of the ice sheet started to break up and retreat in the period from 17,000 to 12,000 years ago. For a 1000 years or more, starting 9500 years ago, the margin of the ice sheet was quite stable, and this standstill may in part have been determined by climate. However, evidence from the moraines at the mouth of Kangia indicate that the retreat of the ice margin slowed because a large part of the ice margin was then resting on land and did not calve into Disko Bugt. The subsequent recession reached a maximum 5000-4000 years ago, when the front of Sermeq Kujalleq was 15-20 kilometres east of its present position.

The following colder period, perhaps associated with increased precipitation, led to a re-advance of the ice margin, which culminated in the maximum position of the ice in the 19th century. More recently many of the glaciers have again started to retreat. The ice front of Sermeq Kujalleq retreated 26 km between 1851 and 1950.


Map showing the retreat of Sermeq Kujalleq from 1850 to 2006. Click on the image to enlarge it.

Portrait of Sermeq Kujalleq

Sermeq Kujalleq can be characterised as a river of crevassed ice. The catchment area that feeds Sermeq Kujalleq is estimated to be 6.5% of the area of the Inland Ice, or 110,000 square kilometres.

The actual ice stream appears as a narrow, 3-6 km wide, well-defined channel that can be followed from the present glacier front about 80 km into the ice sheet. At an altitude of 1350-1400 m above sea level the ice stream splits into three branches, which gradually disappear into the ordinary surface of the ice sheet. Following the course of the ice stream downwards, the surface is relatively smooth until about 50 km from the glacier front, when it gradually becomes more and more crevassed, and lakes and water-filled crevasses disappear. The ice stream has fairly sharp borders against the less-fractured surrounding ice. The marginal crevasse zone of the main stream extends 5 km or more on both sides of the ice stream. About 45 km inland the crevasses and the pattern of flow lines show a funnel-shaped convergence of movement towards the main stream of ice.

Prior to the sudden retreat of the glacier front in 2002, a smaller ice stream flowed south into Sermeq Kujalleq about 10 km behind the glacier front. This northerly branch was separated from the main ice stream by an ice rumple that probably rested on the part of a barrier under the ice. This barrier may have contributed to the prolonged and stable location of the glacier front since the 1950s.

Today we have a reasonably good idea of the topography beneath the ice. The ice stream occupies a very deep trough, and is 1900 m thick at the grounding line, the point where the ice stream loses contact with its substratum and begins to float. As the ice surface is here 400 m above sea level, about 1500 m is below sea level. About 40 km behind the grounding line the ice stream is 2500 m thick, and since the surface of the ice is 1000 m above sea level, the bottom of the trough is 1500 m below sea level. About 50 km east of the grounding line the trough flattens out, and 120 km east of the grounding line no marked trough is discernible on radar profiles.


Bedrock topography beneath Sermeq Kujalleq, showing trough up to 1500 m deep.


Surface heights, and flow speeds and directions, for the ice around and in Sermeq Kujalleq. The normal decrease in flow speed from the equilibrium line on the Inland Ice towards the ice margin is not seen in the ice stream. In contrast, a dramatic increase in the flow speed occurs, and the surface of the ice changes from smooth and undulating to the chaotic, broken and pinnacled surface of the ice stream.

Before the 2002 glacier collapse, the outermost 12 km or so of Sermeq Kujalleq were afloat. The floating segment of the glacier partially fill the inner part of Kangia and moves up and down with the tide, which has a maximum height difference of 3 metres; the tidal effect falls off towards the grounding line. As the tide rises and falls the actual grounding line changes slightly in position, such that there is in effect a grounding zone. Movements of the ice, felt as tremors, vary in strength with the tidal cycle, and can be detected up to 8 km inland from the grounding zone.


Estimates of the annual production of calf ice from Sermeq Kujalleq rely primarily on knowledge of the flow velocity of the ice and the thickness and width of the calving front.

Major calving events, in which large segments of the glacier front detach, take place only a few times in the course of the summer. On the other hand, there is almost continuous calving of smaller pieces of ice from the ice front, and this can be observed all year round.

The process of calving is not yet fully understood. The most important trigger is probably the expansion of the crevasses in the bottom of the glacier, which develop as the glacier detaches from its substratum and begins to float in the fjord. At the same time the glacier is deformed by the effects of the tide on the floating segment, and the tide also induces de-coupling of the glacier from the cliffs on both sides of the fjord. The pressure of water filling the glacier crevasses further strengthens the loosening process.

In the period from March to October 1982, calving and iceberg production was closely followed using data from satellite remote sensing. Three major calving events occurred. The largest produced 14 tabular icebergs, with sizes up to 2 km by 1 km. This event resulted in the estimated discharge of 8.4 cubic kilometres of ice and a retreat of the glacier front by about 2 km, and was associated with the break-up of the sea ice in the inner part of Kangia. Another pronounced calving, with the detachment of large tabular icebergs, was observed in July 1985 when the outer 2 km of the floating part of Sermeq Kujalleq disintegrated in the course of just 45 minutes.

After calving in a normal summer, the glacier front gradually advances during the winter, since the front is supported and stabilised by the sea ice in the fjord. The seasonal movements of the glacier front were studied in the period 1962 to 1996, and the typical difference between the frontal positions in the summer and the winter was 2-3 km.


Calving icebergs once caused waves in Ilulissat harbour.

In earlier times, a strange phenomenon - ‘kaneling’ - could create problems in Ilulissat harbour. This is a description from 1921:

“The water in the harbour suddenly becomes quite rough, and can even become foaming white, and grass and seaweed from the bottom come up to the surface. However, there will rarely be any danger for ships or boats there as long as their moorings are solid. It all lasts only a few minutes before calming down again. In recent years, ‘kaneling’ seems to be on the decline and must now be said to be quite without importance”.

The phenomenon was investigated when a new pier was constructed in Ilulissat harbour. ‘Kaneling’ is related to pressure waves, formed during calving at the ice front, and which become stronger because of resonance in the harbour basin. That the problem declined at the beginning of the 20th century was probably due to the considerable retreat and thinning of Sermeq Kujalleq.

When large icebergs overturn in Disko Bugt, major waves are formed that can be dangerous for boats in the bay and for people on the coast. At Sermermiut, tourists have been inundated by waves from overturning icebergs. This type of wave is not related to ‘kaneling’.

Another phenomenon connected with the iceberg bank (Isfjeldsbanken) was also described in 1921: the formation of an ice-dam across the iceberg bank. The formation of such a dam can cause the water-level in the icefjord to rise above the water level in Disko Bugt. According to the description, the ice-dam breaks when it can no longer withstand the pressure from the dammed waters in the icefjord, and ‘the icefjord shoots out’.

Dynamics of Sermeq Kujalleq

Glaciers normally move down the steepest surface slope, and it is thus the topography under the ice that determines the course of valley glaciers in mountainous regions all over the world. However, the ice sheets in the polar regions are not forced to follow the underlying topography, but instead overrun entire landscapes and create their own pattern of movement. The flow of the ice is also directed downwards here, but it is controlled in particular by the surface topography of the ice sheet, and the flow pattern does not necessarily follow the underlying topography.

In most of the marginal areas of ice sheets the ice moves slowly, and there is little variation in lateral velocity. In other marginal areas the flow can be faster and concentrated in stream-like channels. Tongues of ice that extend down into valleys and fjords are called outlet glaciers, no matter whether they end on land or in water. When a glacier runs into a fjord or a lake, the flow velocity normally increases because of the reduced friction at the bottom. The large ice sheets in Greenland and in Antarctica have outlet glaciers that can locally move at velocities of several kilometres per year; this happens when the glaciers discharge through deep troughs where the basal friction is reduced.

The Inland Ice flow slowly down over following the direction of the ice surface inclination. Near the ice margin, the ice movement pattern is govern by the bedrock topography. (Foto Henrik Højmark Thomsen)

Sermeq Kujalleq is a classic example of an ice stream and its high velocity is due to ice from a large drainage area becoming concentrated into a narrow stream. The crucial question is the nature of the climatic conditions that create a permanent fast-flowing ice stream. The large production of calf ice and the high velocity indicate that Sermeq Kujalleq reacts quickly to climatic changes, but the reasons for the permanently high rate of flow are still under debate.

Traditionally, the term used by glaciologists to explain the high discharge of ice is known as the ‘Jakobshavn effect’. This term derives from the now obsolete Danish name ‘Jakobshavn Isbræ’ formerly used for Sermeq Kujalleq glacier. The ‘Jakobshavn effect’ appears to be a combination of crevasse formation due to increased melting, and increasing acceleration of the disintegrated current of ice. Surface melting is greatly enhanced by extensive surface crevasses. When large quantities of melt water freeze internally in the glacier, considerable latent heat is released and this makes the ice softer and more fluid. Meltwater that reaches the bottom of the glacier acts as a lubricant that helps the ice masses to flow more quickly over the underlying rock.

Large crevasses in a chaotic cracked surface of Sermeq Kujalleq (Jakobshavn Isbræ). (Photo: Henrik Højmark Thomsen)

It has been suggested that the movement of Sermeq Kujalleq started to accelerate about 1850, when rising temperatures after the end of the Little Ice Age led to increased surface melting of the lower parts of the ice sheet. Meltwater that ran down into crevasses and moulins, may have warmed the ice and lubricated the bedrock surface, and thus initiated Sermeq Kujalleq’s fast rate of flow. The high velocity in its turn contributed to the disintegration of the ice into a chaotic mass, so that the surface area exposed to melting became three times as large. The consequences were increased melting, more meltwater to heat up the ice and lubricate the underlying bedrock, and increased transport of ice from the inner part of the Inland Ice towards the fjord. However, there are no records of the glacier’s movements before 1875, and the earliest historical descriptions of the iceberg bank indicate there was a high production of icebergs long before 1850. Whether the acceleration of the glacier began about 1850 is thus an open question.

Classic glaciological theory suggests the ‘Jakobshavn effect’ sustains Sermeq Kujalleq’s permanent high speed. However, other proposals question the necessity of this effect to explain Sermeq Kujalleq’s continued fast flow.


Estimated maximum (blue) and minimum (turquoise) calf ice product­ion from glaciers along the west coast of Greenland. The production of calf ice from Sermeq Kujalleq is in a class of its own. Click on the image to enlarge it.

Dramatic break-up in 2002

In the autumn of 2002, a number of detached parts of the glacier front were observed at Isfjeldsbanken; their characteristic feature was a dark, pinnacled surface. The quantity of such fragments was particularly striking in the summer of 2003, and flights over Kangia and Sermeq Kujalleq showed that the fjord was closely packed with broken up segments of the glacier front. The boundary between these fragments and the glacier front was at times difficult to determine precisely, unlike the situation during the ‘stable’ period of the last half-century, when the frontal position of the glacier in Kangia was usually clearly defined.

Satelite photo

Satellite image of Sermeq Kujalleq and Kangia recorded 7. July 2001. Data recorded by the Landsat 7 satellite from an altitude of 705 kilometres. The colours are artificial. Areas with vegetation are brown or red-coloured and rock surfaces are grey-coloured. The ice margin is located in the position, which was stable from the 1950´s until the break-up in 2002.

Studies of satellite images have enabled the position of the glacier front to be determined for the period May 2001 to May 2003, supplemented by data from an aircraft that flew over the area in August 2003. The images show that the last winter advance of the glacier tongue occurred about March 2002, and that the subsequent disintegration of the floating glacier tongue has resulted in the position of the glacier front retreating 12 km eastwards.

The large ice bay, Tissarissoq, on the southern side of the former glacier front, is now isolated from the main glacier and was showing signs of extensive melting in the summer of 2003. Numerous finds of marine bivalve shells in the moraines around the bay show that Tissarissoq was formerly a marine bay, and it may be that the marine bay may redevelop once again.

satelite map

Satellite image from 28 May 2003. Data from the ASTER sensor on board the TERRA satellite.

The icefjord

Kangia has only been investigated to a limited extent. However, there is no doubt that the inflow of melt water and the ice from Sermeq Kujalleq have a crucial influence on the salinity, currents and temperatures in the fjord. The large freshwater input from the ice sheet and the melting icebergs also has a major influence on the waters beyond the icefjord.

The size of the icebergs indicates that the deepest part of the fjord is between the glacier front in the east and the iceberg bank in the west at the mouth of the fjord. Some measurements made from the sea ice in the fjord indicate depths of over 1000 m in the central parts, but systematic depth profiles along the fjord have not yet been made.

Kangia is normally full of ice the whole year round, and the seasonal variations in the ice cover have been studied in detail using satellite images. From June to September the fjord is normally filled with brash ice and drifting icebergs, while sea ice covers the water between the icebergs the remainder of the year. In recent years, however, the outer parts of the fjord have often been ice-free for long periods in the autumn and early winter. The steady advance of the glacier front, however, means that cracks are continually forming and re-freezing in the sea ice during the autumn, winter and spring.

satelite map

3-dimensional satellite image of Kangia and the surrounding region, seen from the west. The colours are artificial. Areas with vegetation are brown or red-coloured and rock surfaces are grey-coloured. The vertical scale is exaggerated to enhance the differences in elevation.
(Satellite interpretation: Tapani Tukiainen, 3D model: Carsten Egestal Thuesen)

The iceberg bank

Isfjeldsbanken, the iceberg bank, lies at the mouth of the icefjord. Icebergs that are calved from the front of Sermeq Kujalleq drift slowly towards the mouth of the icefjord, a process that takes between 12 and 15 months. The depth of water over Isfjeldsbanken is only about 200-225 m, and here most of the larger icebergs run aground.

The iceberg bank is a curved, low-water threshold at the mouth of Kangia, and stretches from Sermermiut in the north to Avannarliit in the south. There are moraines on each side of the fjord, which are composed of pebbles and boulders and formed at the margin of the Inland Ice 9500-8000 years ago. It is assumed that the line of moraines formerly crossed the mouth of Ilulissat Icefjord and formed the top of the iceberg bank. The morainic sediments have probably been largely eroded by the icebergs subsequently, and it is probable that a sill of bedrock occurs beneath the remnants of the moraine, as is the case in other threshold fjords in Greenland.

A few soundings at Isfjeldsbanken, given on nautical charts, show depths of 200- 225 metres. This shallow threshold separates the deep trough of the icefjord from the 200-400 m deep sea bottom in the inner part of Disko Bugt. Further towards the south-west a deep trench known as Egedesminde Dyb is found, where water depths exceed 800 metres. This trench was formed by the activity of another ice stream during the ice ages.
Large icebergs go aground on the iceberg bank throughout the year. They remain there until they have either melted so much that they can float over the threshold, or the pressure from the ice in the icefjord becomes so great that the icebergs are forced across the threshold and into the deeper water of Disko Bugt.

Isfjeldsbanken at the mouth of Ilulissat Icefjord.


The large icebergs that ground on Isfjeldsbanken at the mouth of the icefjord act as a barrier that causes smaller icebergs to pile up in the fjord. The icebergs are very different in size and form. They vary from small pieces of ice to giants that tower more than 100 m above sea level. Some icebergs are rounded, while others have extremely irregular shapes, often jagged and with pointed summits.

A boat trip between pinnacled icebergs can be an awe-inspiring experience.
(Photo: Dieter Zillmann/Elke Meissner)

It is often said that only a tenth of an iceberg is above sea level. However, glacier ice contains many air bubbles and the part visible above the water is really about one-seventh of the total volume. The content of air bubbles or moraine material can also affect the depth of an individual iceberg. Icebergs with a heavy load of gravel and boulders can be so heavy that only a small part projects over the water. In addition, water salinity plays a role in deciding the depth at which an iceberg floats. The saltier the water, the higher an iceberg will project above the water surface.

Melting and erosion starts as soon as an iceberg is released from the front of the glacier, but usually progresses slowly in arctic waters. Warm air and the effect of the sun’s rays cause the part of the iceberg that is above water to melt. Beneath the water surface, the iceberg is melted by warm salty water and is eroded by the dashing of waves. The processes occur at different speeds at different places on the iceberg, which can therefore become unstable and overturn.

When icebergs escape across the threshold (Isfjeldsbanken) at the mouth of the icefjord, they drift further out through Disko Bugt to the open sea. Some icebergs drift south around the island of Disko, while the majority drift north around the island and there enter Davis Strait between Greenland and Canada. Here the icebergs meet the West Greenland Current, which carries them northwards along the Greenland coast. Further north, the icebergs are carried by ocean currents westwards towards Canada, where they are caught by the Baffin and Labrador Currents and drift southwards.

Many of the large icebergs drift as far south as the latitude 40°N before they melt. Icebergs from Kangia, and from other icefjords and major glaciers in west and North-west Greenland, have always been a danger to shipping on the North Atlantic routes, and it was probably one of these icebergs that sealed the fate of the Titanic in 1912.

Naval map

Distribution of icebergs in the North Atlantic, and reports (with year) of exceptional sightings. The sinking of Titanic in 1912 occurred south-west of Newfoundland.

After the catastrophe an international conference decided that an ice patrol should be established to observe and track icebergs. The US Coast Guard started monitoring them in 1914, and the task was expanded in 1928 to include investigations of where the icebergs originated. A total of 21 outlet glaciers in West Greenland were identified as possible sources of icebergs, but only a few of them, and Sermeq Kujalleq in particular, were considered as being important.

The interest in monitoring icebergs fell in the 1950s and 1960s, probably due to a pronounced fall in the number of icebergs encountered compared with the preceding decades. This development changed dramatically in the 1970s, and 1972 and 1974 were the most iceberg-rich years ever registered. At the same time the oil crisis had generated interest in exploiting oil along Canada’s coast and offshore West Greenland, which gave iceberg studies a completely new perspective.

Continue your journey through Ilulissat Icefjord

The land below and around the ice

Bubbles and glacier ice

Ice can be formed in two ways:

  • by freezing water or
  • by compression of snow.

Greenland’s Inland Ice is composed almost exclusively of glacier ice, formed by the compression of snow.

A pronounced characteristic of glacier ice is the granular structure, with ice grains of different sizes interlocked into each other like pieces of a jigsaw puzzle. The crystals are formed from the original snow crystals, that in the course of time have grown under pressure at the expense of neighbouring crystals. During the process the air in the snow is trapped as bubbles in the ice. A study of the icebergs released from Greenland glaciers shows that the majority of air bubbles have a pressure of 4-5 atmospheres, but pressures up to 20 atmospheres have been measured.

The white colour of the ice is due to the bubbles. Blue or green bands of pure ice that are frequent in icebergs were formed by freezing of bubble-free melt water in the former crevasses in the glacier.

The generally white colour of the ice is due to the presence of air bubbles. Blue or green bands, often seen in icebergs, are formed by frozen meltwater that filled former glacier crevasses.