Introduction
A diversity of streamlined drift features is found in the glacier foreland of Vestari-Hagafellsjökull, an outlet glacier of the Langajokull ice cap, central Iceland (Fig. 1). These include, drumlins defined as large (>0.5 m high) elliptical bed forms, flutes as small (>0.2 m high) elongated bed forms and lineations as very small (< 0.2 m high) very elongated features (often only one clast high). Vestari-Hagafellsjöis a relatively clean glacier with a low surface slope moving over outwash sand and very fine-grained till. This unconsolidated mixture is water-saturated at the glacier margm. Beneath the imconsol-sediments there are tumuli (lava blisters) of pahoehoe lava.
Fig. 1. Location of the site. Map based on the 1937-38 topographic map.
The southern Langajökull area is less well studied relative to other Icelandic ice caps, such as Vatnajökull, because of problems of accessibility and fresh-water shortage. The proglacial drift features have not been Lichens are rare and there is no accurate lichen curve. However, there are four sets of air photographs of this area, taken in 1945, 1960, 1979 and 1991, as well as a 1937-38 topographic map. Additionally, there are data on marginal positions at three times (Reference SigurðssonSigurðsson, 1988).
Details of the area, the three known ice limits and one other older limit of unknown age, are shown in a geomorphological map (Fig. 2). The furthermost limit is assumed to represent the Little Ice Age maximum, which was thought to have occurred around 1890 in Iceland (Reference ThoórarinssonThórarinsson, 1943). Since then the glacier retreated steadily from the early 1900s until 1961. Reference SigurðssonSigurðsson (1988) has recorded that between 1934 and 1961 the retreat was 1306 m. Between 1961 and the mid-1980s it re-advanced 249 m In 1979, it surged 700 m (Reference TheódórssonTheódórs-son, 1980) hut since 1986 it has retreated again.
Fig. 2. (a) Air photograph of the area in 1991. (Reproduced with permission from Landnaltngar Islands.) (b) Geomorphological map of the area. Short heavy lines with dates 1945,1968 and 1979 are ice-marginal positions taken from air photographs.
The till in this area is very sandy as it is a mixture of ash and outwash sand. There were so few clasts within the matrix that till-fabric analysis was possible only during the study of the lineations, which occur associated with a pavement.
This paper discusses the geomorphology of the glacial foreland, followed by the presentation of a model for the subglacial streamlined bed-form development at a range of scales.
The Different Geomorphological Areas
The glacier foreland of Vestari-Hagafellsjökull consists of four geomorphological zones that lie parallel to the glacier margin (Fig. 2) These are discussed in turn below.
Tumuli zone
The outermost zone consists of tumuli of pahoehoe lava (Fig. 3a). Tumuli are formed either by pressure from still-liquid lava flowing beneath the solidifying surface, or pressure associated with volatilization of ground water (Reference Cas and WrightCas and Wright, 1986). There tumuli are 1-2 m high and they show no evidence of having been glaciated.
Fig. 3. (a) Photograph of a tumulus, (b) Photograph of a drumlinized tumulus. Ice-flow direction..was left to right
Lake bed Zone
Closer to the glacier, there is large flat clay plain that was formerly covered by water. This lake, Sandvatn, is shown on the 1937-38 topographical map and 1945 air photographs but by 1960 it had drained. The southernmost limit of the lake a approximately coincident with the maximum ice extent Within the lake bed are morainic features up to 5 m high, some of which have steep ice-contact slopes. These probably represent end moraines and erosional remnants of a sandur deposit developed during retreat between 1890 and 1945.
Outer drumlin zone
The outer drumlin zone consists of drumlinized tumuli (Figs 3b and 4). These have a classical drumlin shape and a rock core although they are only 1-2 m high, 20-30 m long and 10-20 m wide The upper surfaces of the rock cores are commonly striated in the direction of ice flow. Since drumlins are formed subglacially, they were exposed between 1890 and 1960. The till in this zone has a complete moss cover and consists of sand and ash with only a few clasts.
Fig. 4. Gross section of a drumlin in Ike outer drumlin zone.
A series of end moraines, formed as the glacier retreated, traverses this zone. These moraines are probably push moraines which are common in Iceland e.g. Fjallsjókull, Breiðamerkurjökull and Skalafellsjökull in southeast Iceland (Reference PricePrice, 1970; Reference SharpSharp, 1984; Reference BoultonBoulton, 1986). They are probably formed by longitudinal compression by the glacier, squeezing out due to the weight of ice and also by the down-glacier movement of sediment (advection) within the deforming layer (Reference HartHart, 1990).
In order to study their composition, the drumlins in a small area (Fig. 2) were examined and a number of excavations were made. Some drumlins were examined in more detail and the results are shown in Fig. 4. The drumlins were divided into four classes and the number in each class was recorded (Table 1), The four classes were defined as: (a) rock drumlin — a tumulus which is composed of rock but shows evidence of glacial striae and streamlining; (b) till drumlin — drumlins that excavations suggested were probably composed entirely of till; (c) rock-cored drumlin — a tumulus core with a till carapace (shown in Fig. 4); (d) rock core with a trail. The majority of the drumlins were rock-cored (46%) or had a rock core with a tail (31%).Only a small number were composed entirely of either rock or till.
Table 1. Different types of drumlins
Inner drumlin zone
The inner drumlin zone is far more complex. Ice re-advanced over this area between 1961 and the mid-1980s as noted and thus the ground surface is unvegetated. The zone’s southern end is marked by a 2 m high end moraine, is probably a push moraine The rest of the area is covered by small drumlins, similar in dimensions to those in the outer drumlin zone. However, many of these drumlins have flutes superimposed on their surfaces. The of bouldcr on the up-glacier end with a long lee-side accumulation of till (Fig. 5).
Fig. 5. Photograph of a flute in the inner drumlin Zone Approximate dimensions of the large clast: 0.3 m by 0.25 m by 0.1 m.
Close to the ice margin there are many flutes and also a series of lineated surfaces The lineated surfaces consist of lines of clasts, 0.02-0.05 m wide, that are oriented in the same direction as the flutes and drumlins, and spaced approximately 0.1 m apart. They usually contain no matrix but consist entirely of clasts, which sometimes coarsen upwards. Beneath them there is a fine sandy till which has very few clasts. Similar lineations are also superimposed on the drumlins. There are four different types of lineation.
-
(a) The first group (type a) consists of a large clast (approximately 0.5 m long) which has a lee-side till prow and a stoss-side tail of clasts (Fig. 6a). The prow in Fig. 6a consists of a 0.05 m thick dome of till pushed up by the large clast. A till-fabric analysis of the clasts in the tail produced S1 and S3 values of 0.660 and 0.028, respectively,Footnote * orientated parallel with the lineation.
-
(b) The second group (type b) shows a similar pattern but with a longer lee-side prow forming a till ndge that often coalesces with those from other clasts. Only the largest clasts have a stoss-side clast tail (Fig. 6b). These larger clasts are, however, only 0.2 m wide and are oriented transverse to ice flow.
-
(c) The third group (type c) is slightly more complex. In these, relatively clast-free till is raised in ridges above surfaces with abundant clasts The till ridges appear to emanate from large clasts (most are approximately 0.2 m long) that are orientated in the direcuon of ice tlow (Fig 6c). The till ridges also coalesce and at some of the junctions there is a large clast that appears to have been ploughed into the till. A till-fabric analysis of the clasts in the central clast-rich furrow (shown in Fig. 6c) produced S1 and S3 values of 0.825 and 0.008, respectively.
-
(d) The fourth group (type d) consists of alternating lines of high and low clast content. Zones oflower clast content are topographically higher. These zones appear in the lee of large clasts hut these large clasts are not ploughed into the surface and many are oriented transverse to ice flow (Fig. 6d) The largest clast is about 0.4 m across. A till-fabric analysis of the clasts in the clast-rich areas produced S1 and S3 values of 0.548, 0.002, respectively.
Fig. 6. Photographs and interpretative sketch maps of the four types of lineated surface in the inner drumlin zone.
At the margin, there is a 10 m wide zone of water-saturated subglacial material (Fig. 7). Here, the on-going glacier retreat is exposing composite drumlins up to 3 m high. One particular feature, 2 m high, has a tumulus core which is highly plucked on its lee face, indicating that it was a roche moutonnée before the drumlinizing event. The main body of the feature is in the shape of the drumlin but it has a very long (over 100 m) tail (flute) down-glacier from it. The tail is composed both of the very sandy till and of angular blocks which have probably been plucked from the tumulus.
Fig. 7. Photograph of the ice margin
Discussion
Vestari-Hagafellsjökull represents a glacier under which there is a very high probability that the subglacial sediment is deforming. The evidence for this is that the glacier surface has a low slope, the glacier is moving over a potentially weak unconsolidated sandy ash-rich till and at the margin the till is highly saturated. These conditions are those that Reference BoultonBoulton (1979), Reference Alley, Blankenship, Bentley and RooneyAlley and others (1986) and Reference Boulton and HindmarshBoulton and Hindmarsh (1987) suggested are conducive to deformation of the bed. Several of the explanations, given below, for the origin of the observed streamlined features require a deforming bed and, conversely, they would be difficult to explain if the bed were not deforming.
At Vestari-Hagafellsjökull, there are three scales of streamlined bed forms being exposed on the glacier foreland: drumlins, flutes and lineated surfaces. Studies of flutes associated with contemporary glaciers are common (e.g. Blomstrandbreen, Spitshergen-Reference Paul and EvansPaul and Evans, 1974; Breiðamcrkurjökull, Iceland-Boulton, 1976; Is-fallsgläciaren, Sweden — Reference ÅmarkÅmark, 1980; Austre Okstindbre, Norway-Rose, 1989; Lyngsdalen, Norway — Reference Gardon, Whalley, Gellatly and VereGordon and others, 1992) but studies of recently exposed drumlins are less common (e.g. Reference Krüger and ThomsenKrüger and Thomsen, 1984). Reference Rose, Menzies and RoseRose (1987) suggested that there is a continuum of streamlined land forms from small highly elongated flutes to larger less elongated drumlins, formed by similar processes. At Vestari-Hagafellsjökull,there is an opportunity to test model as there was such a range in scale of subglacial streamlined bed forms. Reference Rose, Menzies and RoseRose (1987) divided up the streamlined bed forms on the basis of scale, as I have in this paper.
The Vestari-Hagafellsjökull flutes
Flutes tend to consist of a stoss-side clast, which acts as a core, with a long lee-side till tail (Reference DysonDyson, 1952; Reference BoultonBoulton, 1976; Reference RoseRose, 1989; Reference BennBenn, 1994) Most modern authors have suggested that flutes form as a large clast is dragged through the subglacial sediment until it becomes lodged, whereupon saturated till flows into the low-pressure zone or cavity m the lee of the clast. Thus, flutes may be indicative of deforming bed conditions.
The pattern of till movement within the deforming layer has been well demonstrated by till-fabric studies by the authors mentioned above. Additionally, Reference RoseRose (1989) showed that the flow then diverges at the down-glacier end of the flute (Fig. 8). Although Reference RoseRose (1989) suggested that this represents ice flow, it seems more likely that it represents sediment flow within the deforming layer as suggested by Reference BoultonBoulton (1976) and Reference BennBenn (1994). Reference Gardon, Whalley, Gellatly and VereGordon and others (1992) have suggested other processes of flute formation but they failed to produce substantial evidence to disprove the more widely accepted model outlined above. Similarly, Reference DreimanisDreimanis (1993) has suggested that flutes with crescentic scours at their proximal ends may have a subglacial fluvial origin. However, this type of proximal erosion was not observed at Vestari-Hagafellsjökull.
Fig. 8. Flow of sediment in flute (after Reference RoseRose, 1989). Closely spaced dot represent initiating rock. Widely spaced dots are die flute. Dots with bars show the site cf till-fabric analysis. Lines it ilh arrows show inferred direction of flow based en till fabrics.
The flutes at Vestari-Hagafellsjökull show characteristics similar to those of the flutes discussed ahove and so it is most likely that they too were Conned by movement of saturated till into the Ice of a large clast.
The Vestari-Hagafellsjökull drumlins
Although there have been many models proposed for drumlin formation (see Reference MenziesMenzies, 1984), at the moment most researchers support one of two theories. In the first, based on the ideas of Reference Smalley and UnwinSmalley and Unwin (1968) and elaborated upon by Reference MenziesMenzies (1979) and Reference Boulton, Menzies and RoseBoulton (1987), it is suggested that drumlins form when material flows around more competent obstructions within the subglacial deforming layer. The core can be composed either of a clast of hard rock or a more competent mass of soft sediment, such as sand and gravel or till. The other theory, proposed by Reference ShawShaw (1983) suggests that drumlins form either as depositional bed forms from subglacial fluvial scour of the ice base or are erosional features from large catastrophic subglacial discharges.
The latter theory has little relevance to the present study, as the drumlins at Vestari-Hagafellsjökull were not formad of sand and gravel and there was no evidence of unusually large subglacial discharges. Instead, the drumlins at Vestari-Hagafellsjökull were composed of a tumulus core with a lee side (and sometimes a stoss side) accumulation of till. Thus, the drumlins appear to be a larger-scale example of the flutes and I suggest that they may well have been formed in a similar manner. That is, subglacial saturated till moved into the low-pressure area in the lee of the tumulus cores If this is true, then these arc a type of deforming bed drumlin.
The Vestari-Hagafellsjökull lineations
It has been shown above that there were four different, but related, types of lineation. These lineations consist of a clast-rich layer, of single-clast thickness, resting on a sandy, almost clast-free till. Reference KrügerKrüger (1993) described a similar clast-paved till from Myrdalsjökull, Iceland. Reference Krüger and HumlumKrüger and Humlum (1981) described these clasts as being in traction against the glacier bed. However, Reference KrügerKrüger (1993) noted that only 43% of the clasts in the upper clast unit were striated, although most striations were on the lower surface.
Based on the frictional lodgement theory of Reference Boulton, Wright and MoseleyBoulton (1975), Reference KrügerKrüger (1993)interpreted the clast-rich layer and the underlying till to he lodgement till because of grain-size composition, clast shape and roundness, striations, till and small-scale deformation structures. However, suhsequent work has shown that many of these parameters, particularly the latter, are indicative of subglacial deformation (e.g. Reference Hart and BoultonHart and Boulton, 1991; Reference Hicock and DreimanisHicock and Dreimanis, 1992) and strong till fabrics can be indicative of either lodgement till or deforming bed till associated with a thin deforming layer (Reference Hart and BoultonHart, 1994).
This problem of the origin of the clast-rich surface will be discussed later. However, given its existence, the various types of lineation are interpreted as follows.
The type a lineations consist of a large clast with a lee-side prow and a stoss-side tail. The large clast seems to have ploughed into the sediment as described theoretically by Reference Brown, Hallet and BoothBrown and others (1987) and from field evidence by Reference Clark and HanselClark and Hansel (1989). They suggested that large clasts that were held within the ice were dragged through the suhglacial sediment and, because it was saturated, the till ploughed up in front of the clast. This is probably what happened at Vestari-Hagafellsjökulls.
The tail of small stones behind the large clast may have formed because, once the large clast acted as an obstacle to flow, smaller clasts were stacked up on its stoss side. This is further indicated by the relatively high strength of the fabric in the tail. If this mechanism is correct, this would imply that clasts were moving at velocities within the deforming layer. That is, the sediment within the deforming layer must have behaved as a shear zone at the base of a slow-moving debris flow as suggested by Reference Alley, Blankenship, Bentley and RooneyAlley and others (1987) md Reference ClarkClark (1991).
The second type of lineation (type b) has a longer Leeside prow forming a till ridge, merging with those from other clasts, and a clast-rich tail. The sandy till ridge in Fig. 6b may have been, in part, ploughed up in front of the larger clasts but this would not account for its full length. It seems more likely that a low-pressure area developed in front of the clast and that till was forced into this pressure shadow. They were therefore formed as mini-flutes. The clast tails associated with some of the larger clasts were probably formed in the way suggested above for type a lineations.
The third type of lineation (type c) consists of coalescing till ridges with clast-rich furrows. These till ridges may have formed both as prows (due to ploughing) and mini-flutes, with clasts again collecting in the stoss areas.
The fourth type of lineation (type d) consists of more regular lines of till ridges and clast-rich furrows. These may be a further evolution from the third type. Specifically, flute-like ridges of till occur down-glacier from larger clasts, whilst the clasts fill the areas between ridges. The clasts associated with the lineations, although oriented, are not imbricated as would be expected of a lodgement till. The fabric strength, S1, ranges from 0.55 to 0.825, with an average value of 0.677, and thus is typical of that inferred for a deforming-bed till with a thin deforming layer (Reference HartHart, 1994).
These lineations may represent a continuum of increasing deformation, from a single large clast that produces a lee-side prow and stoss-side tail (type a) through to a collection of clasts that create a series of till with the lee-side flute and clast-rich stoss-rich-stoss-side tails (type d)
The clast-rich surface could be formed in a number of ways:
-
(a) A residual from the removal of fines from fluvial activity. This is unlikely because of the observations of the clast-rich layer beneath the ice at Fjallsjökull and Myrdalsjökull (Reference KrügerKrüger, 1993)
-
(b) Local particle-size sorting due to lodgement (Reference KrügerKrüger, 1993). This, too, is unlikely as there is little evidence for lodgement at this site and very strong evidence for subglacial deformation. Furthermore, it is not clear how the lodgement process could sort clasts.
-
(c) “Depositional” deforming-bed boulder pavement. Both Reference HicockHicock (1991) and Reference ClarkClark (1991) have suggested that boulder pavements form in deforming-bed tills because heavier clasts sink to the bottom. This is not the process that occurred at the present site because the boulder pavement is at the top of the sequence.
-
(d) “Erosional” deforming-bed boulder pavement — within the deforming layer, particles of different sizes can move at different velocities. The finer particles can be moved away down-glacier, whilst the coarser particles could remain as an erosional lag (see Fig. 9). In this way, the deforming bed has “winnowed” away the fines
Given the other evidence for deforming-bed conditions at this site, the latter explanation is the most likely. The lineations probably form under conditions where the deforming bed is thin due to its marginal situation (i.e. low basal shear stress) (Reference Boulton and HindmarshBoulton and Hindmarsh, 1987; Reference Hart, Hindmarsh and BoulronHart and others, 1990). However, there is also evidence that the deforming layer consists of two layers of contrasting competencD: an upper rapidly deforming layer of clasts and a lower more slowly deforming layer of sandy til I The presence of two deforming layers ifi very common in front of Icelandic glaciers travelling over deformable beds and was first suggested by Reference Boulton and DentBoulton and Dent (1974) but also observed by Reference DowdesWell and SharpDowdeswell and Sharp (1986). At this site, the upper deforming layer has produced a boulder pavement, whilst the lower, more slowly moving layer has formed the ridges and mini-flutes (Fig. 9).
Fig. 9. The formation of lineations. The upper rapidly deforming layer produces the boulder pavement and the lower more slowly deforming layer produces the ridges and mini-flutes.
The different scales and superimposition of streamlined bed forms
It was shown above that there are three scales of bed forms. drumlins, flutes and lineations, and it was argued that these are interlinked and formed by deformational processes. These are shown schematically in Fig. 10, All three bed forms have the following similarities:
-
(a) A rigid core, which for the lineations and flutes was a clast and for the drumlins was a tumulus.
-
(b) A lee-side till ridge which was postulated to have formed by the movement of saturated material into the lee of the core.
However, there were also some differences in the formational processes:
-
(i) The core clasts associated with the lineations and flutes could and probably did move, whilst the drumlin cores were fixed. Thus, the flutes and lineations are more mobile features and this may explain why the flutes were much longer than the drumlins. However, flutes may also be longer because they are formed under thinner ire which has a slower closure rate.
-
(ii) The flutes and drumlins have a relatively high relief (> 0.2 m) and are composed of sandy till with a few clasts. The lineations consist of very low clast ridges surrounded by clast-rich areas. Flutes and drumlins seem to be associated with a thick (> 0.2 m) deforming layer, whilst the lineations may be associated with a thin deforming layer( < 0.2 m).
-
(iii) Associated with the lineations there was stoss-side stacking behind the core of the bed form. Although this was not observed in the flutes and drumlins at this site, stoss-side stacking on a larger scale has been observed in drumlins (Reference HartHart, 1995) On a single-clast scale, this stoss-side stacking has also been observed in Quaternary till sequences (clast clusters) (e.g. Reference Stea and BrownStea and Brown, 1989) and may not always be an indication of lodgement till as suggested by Reference Boulton and PaulBoulton and Paul (1976) but may rather be indicative of deforming-bed conditions
Fig. 10. Schematic diagram of the streamlined bed forms al Vestari-Hagafelhjökull.
It seems plausible that the size of the bed form scales with the thickness of the deforming layer. Thus, the drumlins may have been formed when the deforming layer is approximately 0.5-2 m thick, the flutes when it is 0.2-0.5 thick and the lineations at a thickness of < 0.2m.
Within this site there are many instances in which flutes and lineations are superimposed on the drumlins. It has been shown by Reference Rose and LetzarRose and Letzer (1977), Reference Meer, Evenson, Schlüchter and Rabassavan der Meer (1983), Reference Krüger and ThomsenKrüger and Thomsen (1984) and (Reference Rose, Menzies and RoseRose 1987, Reference Rose1989) that there is often superimposition of one streamlined drift form over another. The thickness of the deforming layer has been shown to relate to the effective pressure, basal shear stress and the nature of the underlying sediment (Reference Boulton and HindmarshBoulton and Hindmarsh, 1987; Reference Hart, Hindmarsh and BoulronHart and others, 1990). If the streamlined bed forms at different scales are related to different thicknesses of the deforming layer, then, assuming the above parameters remain constant, as the glacier retreats so the deforming layer will thin and smaller streamlined bed forms will be superimposed on the larger ones.
Conclusions
There are three distinct scales of streamlined hed forms at Vestari-Hagafellsjöfull and, although smaller ones are superimposed upon larger ones, all are postulated to have been formed by similar processes.
Flutes are the intermcdiate-size bed forms and these have a mobile core (clast). These are thought to have been formed by material moving into a lee-side pressure-shadow area. Drumlins art the largest feature; they have a fixed core (tumulus). These are interpreted to have formed in a similar way to flutes, with subglacial material collecting in the pressure-shadow areas in the lee of the core. These features were both formed of sandy till, implying one thick deforming layer.
At the smallest scale are the lineations, of which there were four types, all of which appear to have had a mobile core (clast).These form in a sandy till with an upper, one-clast thick, clast-rich layer. This is interpreted to be a boulder pavement resulting from net subglacial deforming-bed erosion. The simplest lineations were probahly by large clasts which were held by the glacier and thus ploughed through the deforming layer (type a). This produced a lee-side prow of sandy till and a stoss-side tail of clasts. In others, the frontal prow becomes elongated because material from the deforming layer moves into the pressure-shadow area, forming a mini-flute (type b). These mini-flutes sometimes formed till ridges, some of which coalesced (type c),whilst others remained parallel (type d).
It seems probable that the bed forms were formed during subglacial deformation but in association with different thicknesses of the deforming layer. As the thickness of the deforming layer changed, perhaps as the glacier retreated, one style of subglacial bed form was superimposed upon another. In this way a continuum of subglacial bed forms was formed.
Acknowledgements
I should like to thank K. Martinez for his help in the field, J. Rose for drawing my attention to the site and the Icelandic Research Council for permit 92037. I should also like to thank A. Burns and his colleagues in the Cartographic Unit for figure reproduction and Dr R. LeB. Hook and M. Hendy for useful comments on the text. This field work was funded by a University of Southampton small grant.
