Geographic distribution

All major ice caps in Iceland contain surge-type outlet glaciers, and surge-type glaciers occur almost exclusively as outlets of the major ice caps (Fig. 3). These ice caps are widely distributed across the country, and span the full range of climatic conditions represented in Iceland, from the warm and wet southeast coast (Vatnajökull, the largest non-polar ice cap) to the cool and drier northwest peninsula (Drangajökull). All glaciers in Iceland are warm-based, so we can rule out any surge mechanism for these glaciers that relies on a thermal transition. Several surge-type glaciers exist outside of the major ice caps, three of which can be found in the mountains of north central Iceland. Altogether 26 surge-type glaciers ranging in size from 0.5 to 1500 km² have been identified in Iceland. About 80 surge advances have been recorded, ranging from tens of metres up to 10km.

All of Vatnajökull’s major outlets are surge-type glaciers. Surge-affected areas of Vatnajökull are delineated in Figure 3a and occupy approximately 75% of the ice cap. The steep and active valley glaciers draining southeastern Vatnajökull andthe northwestern flank of Vatnajökull near the ice-capped volcano Bárðarbunga are not known to surge. Surges that have led to advances of the glacier terminus are listed in Table 1. However, glaciers commonly experience uplift, crevasse formation and propagation of surface bulges over a period of weeks that only occasionally lead to a significant advance of the glacier terminus (personal communications from R. Stefansson, 1991 and S. Björnsson, 2002). The local people have recorded these events as “gangur” (e.g. Reference BjörnssonBjörnsson, 1956), derived from the Icelandic word meaning “walk”. These events have been documented on the small, steep outlets of the ice-capped volcano Öræfajökull in southern Vatnajökull, the steep southern outlets of Mýrdalsjökull ice cap and on surge-type glaciers such as Skeiðarárjökull and Breiðamerkurjokull.

Table 1.

Reported surge advances in Iceland

Surges occur in the southern outlets of Langjökull (western and eastern Hagafellsjokull; Fig. 3b), but have not been reported in the steep western and eastern outlets. Surges occur in the relatively flat ablation area of Hofsjökull (Fig. 3c) and in the gently sloping northern outlets of Myrdalsjokull (Sléttjökull and Öldufellsjökull; Fig. 3d). Hofsjökull and Mýrdalsjökull are the only major ice caps in which surges have not been observed to affect glaciers all the way to their central ice divides. The upper reaches of Hofsjökull and Myrdalsjokull are relatively steep because the ice caps are draped over a single steep central volcano, and surges tend not to propagate up-glacier past sharp increases in the ice surface slope. Average surface slopes of surge-type glaciers generally fall in the range 1.6–4°, and surges in glaciers steeper than 3° typically only affect the low-sloping ablation areas. Most glaciers in Iceland are steeper than this. The average mean slope of the 45 non-surge-type glaciers in the Iceland Glaciological Society database is 11.8°, ranging from 2.9° to 25.7°. A statistical analysis of an Iceland glacier inventory (253 glaciers, 19 surge-type) found that median surface slope was about 7° for non-surging glaciers, and about 2° for surge-type glaciers (Reference HayesHayes, 2001). Surge-type glaciers are generally absent in the smaller ice caps (e.g. the individual ice-covered volcanoes such as Eyjafjallajökull and Snæfellsjökull). Exceptions to the low-sloping collection of surge-type glaciers are found only in hard-bedded areas of the Tertiary basaltic region: the three surge-type outlets of the ice cap Drangajökull have slopes around 7°, and the three known surge-type cirque glaciers in north central Iceland (Bægisárjökull, Burfellsjokull and Teigardalsjokull) have slopes exceeding 10°. In total, known surges affect 60–70% of the ice-covered area in Iceland, with surge-type glaciers outsizing non-surge-type glaciers on average and having no discernible orientational bias.

Subglacial geology and topography

The geological substrate of surge-type glaciers in Iceland is as varied as the geology of Iceland itself, and includes both hard and soft beds, with substantial subglacial mountain ridges extending both along and transverse to the direction of flow. These substrates range from Tertiary basalts in northwest and eastern Iceland to Quaternary and Holocene basalts and hyaloclastites in the active volcanic zone (Reference Jόhannesson, Sæmundsson and JakobssonJόhannesson and others, 1990). Holocene basalt lavas are highly permeable (hydraulic conductivity >10^–5ms–¹) but other substrates less so (Reference SigurðssonSigurðsson, 1990). Tertiary lava is almost impermeable. We infer subglacial bed types and their relative permeabilities from the geology exposed at the ice margins and from glacier drainage characteristics. The flanks of the ice caps are believed to rest on unlithified deposits, while the central regions of the ice caps probably overlie stiff till or solid rock.

The largest ice cap, Vatnajökull, partially overlies the active volcanic zone and covers several of Iceland’s largest volcanoes. It straddles a distinct geological boundary coincident with the margin of the active volcanic zone (see Fig. 1). Porous lava beds primarily underlie western Vatnajökull, while the main outlets of eastern Vatnajökull rest on impermeable unconsolidated till (Reference BjörnssonBjörnsson, 1988; Reference Jόhannesson, Sæmundsson and JakobssonJόhannesson and others, 1990; Reference SigurðssonSigurðsson, 1990; Reference Björnsson and EinarssonBjörnsson and Einarsson, 1991). The steep non-surging outlets along the southeastern flank of Vatnajökull rest on impermeable bedrock. The other large ice caps, Langjökull, Hofsjökull and Mýrdalsjökull, lie entirely within the active volcanic zone (see Fig. 1). Of those, the northern outlets of Mýrdalsjökull and southern outlets of Langjökull may be partially underlain by porous lavas. The beds of surge-type valley glaciers in northern and northwestern Iceland (Bægisárjökull, Burfellsjokull, Teigardalsjökull and outlets of Drangajökull) are composed of hard impermeable Tertiary basalts. Despite the presence of numerous geothermal areas beneath the ice caps in Iceland, surge-type glaciers are not spatially correlated with these basal heat sources. Subglacial meltwater produced by geothermal heating tends to accumulate locally and drain episodically, rather than contribute to the ambient basal drainage regime.

All of the major ice caps in Iceland have been mapped by radio-echo sounding (Reference BjörnssonBjörnsson,1988; Reference Björnsson, Pálsson and GuðmundssonBjörnsson and others, 1991,Reference Björnsson, Pálsson and Guðmundsson1992, Reference Björnsson, Pálsson and Guðmundsson2000). Bedrock overdeepenings of 50–100m (relative to the proglacial outwash plain) are common beneath the large surge-type outlet glaciers. Skeiðarárjökull and Breiðamerkurjokull, the most active southern outlets of Vatnajökull, excavated troughs 200–300m below sea level during their advance from 1400 to 1900 (Reference BjörnssonBjörnsson, 1996). Our recent hydrologic modelling of Vatnajökull suggests a correlation between the extent of several surge-affected areas and a predisposition for low basal effective pressures. This is probably related to the spoon-shaped basins described by Reference ThorarinssonThόrarinsson (1964) and their ability to retain water.

Surge frequency and duration

Major surges, with return periods ranging from several years to a century, have occurred in all of the large lobate outlets on the northern, western and southwestern flanks of Vatnajökull (Fig.3a). The same applies to some of the broad lobes of Langjökull and Hofsjökull, and to the northern outlets of Myrdalsjokull (Fig. 3b–d). The timing and duration of surges that have caused observed advance of the glacier terminus are summarized in Table 1. Surges do occur that do not lead to glacier terminus advance, but because they are more difficult to distinguish in the historical record they are excluded from this list. Reported surge intervals should therefore be taken as maximum estimates.

Observations suggest that certain glaciers surge at fairly regular intervals. Síðujökull has surged at 30 year intervals since the 1930s, Dyngjujokull at 20–30year intervals, Brúarjökull every 80–100 years and Mulajokull about every 10 years. The recorded surge history of other glaciers points to a variety of surge intervals. Breiðamerkurjökull, one of two major southern outlets of Vatnajökull, illustrates this point: it is known to have surged 11 times between 1794 and 1969, at intervals ranging from 6 to 38 years. The western branch of Skeiðararjökull, the other large southern outlet of Vatnajökull, surged four times at increasing intervals of 25– 118 years between 1787 and 1991. Its middle branch surged four times between 1857 and 1991 at decreasing intervals of 72–6 years. This simultaneous surge-interval increase in the western branch and surge-interval decrease in the middle branch suggests a peculiar relationship between neighbouring glaciers (see Table 1). All southern outlets of Vatnajökull have experienced a similar mass balance during the 20th century (Reference BjörnssonBjörnsson, 1980; Reference Björnsson, Pálsson, Guðmundsson and HaraldssonBjörnsson and others, 1998), but their dynamics have apparently been differently affected. Surges in southern Langjökull since the 1970s occurred after at least 40 years of quiescence and maybe a response to positive mass balance from the 1960s to 1980s.

Fieldwork on glaciers in Iceland since the early 1990s has revealed that the surge process can take 2–3 years from the first signs of increased sliding and the subsequent down-glacier propagation of a surge wave. Prior to regular monitoring of glacier surface velocity, surge duration was generally considered to be less than half a year, as surge onset was associated with increased turbidity in glacier rivers and advance of the glacier terminus. The advance of the terminus, however, lasts in most cases about 2–3 months regardless of the size of the glacier and the distance of the advance. Lingering effects of a surge can often be detected in the accumulation area in the form of crevassing and surface lowering several years after the terminus has stopped advancing.

Changes in surface geometry and mass flux during surges

Surges have a marked impact on the geometry of the ice caps, typically thinning the accumulation area by 25–100m. They also play an important role in the mass flux through the outlet glaciers. During the 1990s, 3000 km² of Vatnajökull (38% of the ice-cap area) were affected by surges, which transported about 40 km³ of ice from the accumulation area to the ablation area. This is approximately 25% of the total ice flux to Vatnajökull’s ablation area during that period. For some outlet glaciers the contribution of surges to mass transport is higher still. During the 1998–2000 surge of Dyngjujökull, 13 km³ of ice were transported to the ablation area (Fig. 4) out of the total 20 km³ of ice accumulated during the preceding 20 year quiescent period.

Fig. 4.

Changes in the surface of Dyngjujökull during the 1998–2000 surge. Upper panel: Interpolated change in surface elevation from the beginning to the end of the surge. Heavy dashed line separates zones of thickening and thinning. Lower panel: Change in surface elevation along profile A–B in upper panel for three different time periods during the surge.

Repeatedly measured surface profiles on Tungnaárjökull in western Vatnajökull show a classic example of the surge-related cycle of mass accumulation and expulsion (Fig. 5). For approximately 50 years following the surge that ended in 1946, Tungnaárjökull thickened in the reservoir area and thinned and steepened in the receiving area. The next surge of Tungnaárjökull in the early to mid-1990s resulted in a readvance of the glacier terminus relative to its measured position in 1992, and surface drawdown in the reservoir area extending 30 km up-glacier from the terminus.

Fig. 5.

Observed surface profile of Tungnaárjökull, 1946–95(a) Surface profiles over 50 years following a surge. (b) Surface profiles between and after two surges. Note the difference in horizontal scales in (a) and (b). The numbered triangles are poles for measuring velocity and mass balance.

Surge onset and propagation

After several years of glacier surface steepening, we observe velocity increases in a zone typically 10 km long (in Vatnajökull and Langjökull) and centred in the upper ablation area. We refer to this area as the enhanced velocity zone. Figure 6a shows 1992 surface stake locations and basal topography along a profile onTungnaárjökull, and Figure 6b shows measured velocity profiles for 1986 (pre-surge) and summer 1992 (at the onset of a surge) compared to the calculated average balance velocity for 1992–93. The 1986 presurge velocity profile shows a clear deficit relative to the computed balance velocity profile, while the 1992 profile shows velocities in excess of balance in the lower ablation area. Velocity profiles along a flowline through this zone are usually bell-shaped. Upstream from the velocity peak along the profile (e.g. between stakes 5 and 6; Fig. 6a), crevasses are formed and a slight surface lowering is observed, while downstream from the velocity peak the glacier thickens. Velocities continue to show seasonal variations but generally increase over 2–3 years, while the maximum velocity along the profile remains within the enhanced-velocity zone (Fig. 6c). Hence, the location of maximum surge velocities appears to be relatively stable, rather than migrating up-or down-glacier. Figure 7 illustrates the dramatic velocity increase experienced in lower Tungnaárjökull during this surge that occurred in the early 1990s.

Fig. 6.

changes in surface velocity associated with a surge of tungnaárjökull. (a) glacier surface (1992) and bed topography along the profile, with 1992 stake locations. (b) measured velocity profiles before (1986) and during (19 92) the surge, compared to the computed average balance velocity profile for 1992–93. the number of velocity stakes on the profile varied from year to year. note the error bars. (c) selected velocity measurements along the profile during the 1992–94 surge. note the difference in vertical scale between (b) and (c).

Fig. 7.

temporal changes in surface velocity along a profile on tungnaárjökull during surge onset. (a) glacier surface and bed topography along the profile, and stake positions. (b–d) velocity records from stakes t3 (b), t4 (c) and t5 (d).

Over a period of a few months, a step-like thickening of the glacier develops in the lower part of the enhanced-velocity zone several kilometres from the terminus. This development does not occur preferentially during any particular season of the year. The first unquestionable sign of a surge is the advance of this bulge, usually tens of metres high, at rates measured at 20–80 m d ^–1 . Propagation of the bulge to the glacier terminus typically requires less than 1 year, and often less than 6 months, during which time its crest remains relatively uncrevassed. Once the bulge reaches the terminus, the glacier begins to advance as a vertical front 20–50 m high. For most surge-type glaciers in Iceland, advance of the terminus lasts several months. Notable exceptions to this are the steep (around 7°) hard-bedded outlets of Drangajökull which advance for 5–6 years during a surge. The maximum advance rate measured during a surge in Iceland is 100 m in 24 hours on Bruarjokull in 1964 (Reference ThorarinssonThόrarinsson, 1969) and Siðujökull in 1994 (Science Institute, unpublished data). Four of the large outlets of Vatnajökull typically advance by about 1km (Skeiðararjökull, Siðujokull, Tungnaárjökull and Dyngjujokull; see Table 1), while Brúarjökull usually advances by 8–10 km.

Surge-related hydrology

The muddying of glacial rivers and the increased number of hydraulic outlets at the glacier terminus (Reference BjörnssonBjörnsson, 1998) are two robust hydrological signatures of surging in both hard- and soft-bedded Icelandic glaciers. The latter is a common manifestation of the conduit-to-distributed transition in the basal hydraulic system that accompanies a surge (e.g. Reference KambKamb and others, 1985). Distributed drainage promotes basal water lubrication, for which there is much evidence as a facilitator of fast glacier flow (e.g. Reference RaymondRaymond, 1987). Well-lubricated basal glide beneath surging glaciers in Iceland is suggested by a general lack of push-moraine formation in front of an advancing terminus. An alternative explanation for this is that surges proceed dominantly by deformation, rather than sliding. However, this notion remains to be reconciled with the very rapid rates of glacier advance combined with low ice surface and bed slopes.

During a surge, long-term dislocation of ice and water divides can take place high on the glacier. Surge-induced changes in the glacier surface structure alter the subglacial hydraulic catchment structure. A natural experiment demonstrated this in 1994 and 1995, when jökulhlaups from central western Vatnajökull partly drained to the river Hverfisfljόt instead of the river Skafta. Surface lowering of Síðujökull (feeding the river Hverfisfljot) during its 1994 surge extended its subglacial water catchment northward, causing this temporary change in flood routing. After Tungnaárjökull stopped surging in 1996, Skaftá jökulhlaups resumed their original course to the river Skafta (Reference Zόphόníasson and PálssonZόphόnísson and Pálsson, 1996). Although outburst floods from subglacial geothermal areas have occurred during surges (e.g. Reference BjörnssonBjörnsson, 1998), surge timing is not related to flood incidence nor has a flood ever been known to trigger a glacier surge in Iceland.

It remains an open question whether surge termination is related to release of stored water. In runoff records during 20 surges, both small and large, no indication can be found of a flood occurring at the termination of a surge. River gauges from which these data were collected tap into large and sometimes multiple catchment basins, which are usually only partially glacierized, so one might expect some surge-related variations to be masked. Floods have been known to occur without interrupting surges, such as took place in 1998 at Drangajökull. Fluvial erosion of deep new watercourses in front of western Hagafellsjokull in Langjökull (Fig. 3b) in autumn 1998 invites speculation as to whether a sudden flood terminated an incipient surge. Whether or not a flood occurs, melt-water production typically increases for several years following a surge, due to the increased surface area of the ablation zone and new crevasses exposed to turbulent heat fluxes and radiation. On Tungnaárjökull and Dyngjujökull, a 30% increase in runoff the first year after the surge can be ascribed to this effect, of which 5% was due to the altered hypsometry.

Muddy discharge appears in the wake of a propagating surface bulge, which acts as a dam to basal water, when it intersects the glacier margin. While the total glacier runoff does not change appreciably during a surge, the sediment load (and hence sediment concentration) increases substantially, especially in the finest grain-sizes (Reference SigurðssonSigurðsson, 1995; Reference Pálsson and VigfussonPálsson and Vigfusson, 1996). During surges in Vatnajökull, the sediment concentration in affected outlet rivers is generally 7–10 kg m^–3 for 1–2 years (National Energy Authority database, see Reference BjörnssonBjörnsson, 1980; Reference Pálsson and VigfussonPálsson and Vigfusson, 1996). These concentrations are comparable to those during glacier outburst floods. During the 1963–64 surge of Brúarjökull, the river Jökulsá á Brú had an average suspended-sediment concentration of 6.5 kg m^–3. In 1964 the total sediment load was 2.5×10¹⁰ kg (river discharge of 120 m³ s^–1), equivalent to a denudation rate of 12 mma^–1 over the 1500 km² affected by the surge. The aftermath of the surge was seen in decreased annual suspended load up to 1980 (Reference Tomasson, Pálsson and VigfússonTόmasson and others, 1996; Reference Pálsson, Harðardόttir, Vigfússon and SnorrasonPálsson and others, 2000). During surges of Hagafellsjökull in southern Langjökull, sediment concentrations in the outlet rivers increase on average from 0.2 to 1.0 kg m^–3.

Surges in adjacent glaciers

Most surge-type glaciers in Iceland are outlets of ice caps that during surges have sharply defined boundaries with respect to their neighbours, in both the reservoir and receiving areas. Observations from neighbouring Tungnaárjökull and Síðujökull demonstrate that surges can cause migration of ice and water divides in the ice cap. The dynamic relation between these adjacent glaciers is complex. Sometimes they surge separately, as Tungnaárjökull did in 1945 and Síðujökull did in 1963, and sometimes in tandem as in the early 1990s. During their independent surges, Tungnaárjökull and Síðujökull scavenge ice from one another’s drainage basins, resulting in a 200 km² area that has participated in surges of both glaciers. The same applies for water, as exemplified by the previously mentioned drainage of Skaftá jökulhlaups in 1994 and 1995 to the river Hverfisfljόt after the surge of Síðujökull.

Measured surface velocities on western Vatnajökull illustrate the acceleration of Tungnaárjökull and Síðujökull in the early 1990s (Fig. 8). Crevasse formation on upper Síðujökull (personal communication from M. T. Guðmundsson, 1990) and anomalous low-frequency seismic activity in summer 1990 were the first indicators of a surge (personal communication from B. Brandsdottir, 1990). In January 1994 a 70 m high crescent-shaped bulge was observed on the glacier surface. The river Djupa, draining the eastern margin of Síðujökull, discharged anomalously turbid water after the passage of the bulge. The bulge propagated down-glacier at a rate of 20 m d^–1 to the terminus, and turbid water was eventually discharged from all of the major outlet rivers draining Síðujökull. This surge lasted for approximately 4 months after the bulge was first observed, affecting a 500 km² area and resulting in a 1150m advance of the glacier terminus.

Fig. 8.

Surface velocity vectors along profiles on adjacent outlets Síðujökull and Tungnaárjökull in western Vatnajökull, capturing the initiation of surges on both glaciers.

Surge activity continued in western Vatnajökull through most of the 1990s. Velocity increases were first detected on Tungnaárjökull in 1992, and in 1993 the measured velocity of central Tungnaárjökull (Figs 6–8) was more than double its 1986 value. By the summer of 1994, velocities had increased significantly over most of the length of the glacier. Figure 8 shows this dramatic increase in speed as well as the southward deflection of velocity vectors toward the depressed basin of Síðujökull. By the autumn of 1995, Tungnaárjökull had advanced a total of 1200 m. In 1994 it became obvious that Sylgjujökull (north of Tungnaárjökull) was surging, and this surge may have continued until 2000 in its northernmost region.

Hagafellsjokull, in southern Langjökull (Fig. 3b), demonstrates similar surge-pattern complexity. Western and eastern Hagafellsjokull started surging separately in 1971 and 1975, respectively. In 1980, both branches of Hagafellsjökull surged together. From summer 1997 to autumn 1998 the surface velocity of western Hagafellsjokull increased from 50 to 250 ma^–1 in a zone 4–12km above the terminus, but then suddenly dropped down again. It appears as though an incipient surge in the western branch aborted, while a surge of eastern Hagafellsjokull that had begun around the same time continued.