Subglacial bed deformation and dynamics of the Apriķi glacial tongue, W Latvia

Subglacial bed deformation and dynamics of the Aprini glacial tongue,W Latvia

TOMAS SAKS, ANDIS KALVANS AND VIT�ALIJS ZELCS

BOREAS Saks, T., Kalvans, A. & Zelcs, V. 2011: Subglacial bed deformation and dynamics of the Aprini glacial tongue,W Latvia. Boreas, 10.1111/j.1502-3885.2011.00222.x. ISSN 0300-9483.

We evaluate the glacial dynamics and subglacial processes of the Aprini glacial tongue in western Latvia during theNorthern Lithuanian (Linkuva) oscillation of the last Scandinavian glaciation. The spatial arrangement of glacialbedforms and deformation structures are used to reconstruct the ice dynamics in the study area. The relationshipbetween geological structures at the glacier bed and the spatial distribution of drumlins and glacigenic diapirs, onthe one hand, and the permeability of sediment and bedrock, on the other, is ascertained. Drumlins are found in theupper part of the Aprini glacial tongue area and are composed of soft deformable sediments overlying highlypermeable Devonian dolomite. The soft deformable clayey silty bed with low hydraulic conductivity is conduciveto the development of diapirs. The occurrence of diapirs and drumlins is controlled by the fluctuation of pore-water pressure at the glacier bed and is considered to be an indicator of fast ice flow of the Aprini glacial tongueduring its reactivation at the end of the Oldest Dryas (18–15kaBP).

Tomas Saks (e-mail: [email protected]), Andis Kalvans and Vit�alijs Zelcs, University of Latvia, Faculty of Geographyand Efarth Sciences, Alberta Street 10, Riga LV-1010, Latvia; received 25th October 2010, accepted 9th June 2011.

An association between fast ice flow and subglaciallydeforming sediments has been recognized for morethan 20 years (Boulton 1986; Alley 1991; Clark 1995).The understanding that many glaciers are underlain bysoft, deforming sediments, weaker than the ice itself,that can contribute significantly to fast ice flow hasbeen heralded as ‘paradigm shift in glacial geology’(Boulton 1986; Murray 1997; Maltman et al. 2000; vander Meer et al. 2003). Since then, the significance ofsubglacially deforming sediments for local and globalice-sheet dynamics (Boulton & Hindmarsh 1987), forlarge-scale sediment redistribution patterns (Alley1991; Piotrowski et al. 2004) and even as a global cli-mate factor (Clark 1994, 1995) has been recognized.Special attention has been paid to the subglacial dy-namics that govern ice streams and to the specific lim-itations or prerequisites for fast ice flow (Alley 1993;Stokes et al. 2006, 2007). It has been stressed recentlythat ‘sticky spots’ are crucial in the dynamics of icestreams (Christoffersen & Tulaczyk 2003a, b; Stokeset al. 2006, 2007).

Clayton et al. (1989) and Piotrowski et al. (2001)challenged the existence of widespread deforming bedconditions beneath the lobes of the southern Lauren-tide Ice Sheet and in general. It has been recognizedthat under a warm-based glacier resting on uncon-solidated sediments, movement can also occur by slid-ing along the ice–bed interface (Piotrowski et al. 2001).Sliding of the glacier on a thin water film or upon sedi-ment slurry of limited thickness, which is mainly con-trolled through subglacial water-pressure fluctuations(Alley 1993) and, to some extent, by sediment grainsizes, mineralogy, bed roughness and ice velocity(Knight 2002), has been proposed.

In this paper we evaluate the glacial dynamicsand subglacial processes of the Aprini glacial tongue(AGT) in western Latvia during the deglaciationphase of the last Scandinavian Ice Sheet (SIS). Ouranalysis of glaciotectonics and glacial landformspotentially provides important insights into subglacialdynamics and ice–bed interactions during deglacia-tion stages in areas underlain by deformable sedimentsin general.

Location of study area and geologicalbackground

During the Weichselian glaciations, western Latvia wascovered by the Baltic Ice Stream (BIS) of the Scandina-vian Ice Sheet (SIS) (Ehlers 1996; Boulton et al. 2001).During the course of deglaciation, the BIS in westernLatvia split into the Kurshian and Usma ice lobes, whichterminated in local glacier tongues (�Abolti’s et al. 1972,1977; Veinbergs 1972; Zelcs & Markots 2004), and sev-eral ice oscillations have been suggested for this area(Veinbergs 1964; Zelcs & Markots 2004).

These ice tongues are marked by marginal ridgesand associated landforms, which have been correlatedwith other ice-marginal formations in the EasternBaltic (Raukas et al. 1995). The Aprini glacial tongue(AGT) emerged from the Kurshian lobe during theNorth Lithuanian (Linkuva) phase of the deglaciation(Veinbergs 1972; Meirons et al. 1976; Meirons &Straume 1979; Zelcs & Markots 2004). Ice extentduring this phase is marked by lateral and marginalmoraines on the western slope of the Western KursaUpland (Veinbergs 1972; Meirons & Straume 1979)

DOI 10.1111/j.1502-3885.2011.00222.x r 2011 The AuthorsBoreasr 2011 The Boreas Collegium

Saks, T., Kalvans, A. & Zelcs, V. 2012 (January): Subglacial bed deformation and dynamics of the Aprini glacial tongue, W Latvia. Boreas, Vol. 41, pp. 124–140. 10.1111/j.1502-3885.2011.00222.x. ISSN 0300-9483.

(Fig. 1B). This relatively small extension of theKurshian ice lobe was about 35 km long, and down-glacier widened from 6 km to 15 km, providing adivergent ice-flow pattern.

The body of the AGT is situated on the Aprini andPiemare Plains (Figs. 1B, 10). The Aprini Plain is a de-pression along the western edge of the Western KursaUpland and corresponds to a local bedrock depression(Juskevics et al. 1998). The Baltic Ice Lake coastline, stageBII (Veinbergs 1964), marks the geomorphological bor-der between the Aprini and Piemare Plains (Fig. 1B).

The Piemare Plain is characterized by a complex se-quence of Pleistocene deposits up to 70m thick (Fig. 2).According to the established stratigraphy of the area, athin layer of Elsterian till at the base of the Pleistocenesediment sequence is covered by marine Holsteiniandark grey clayey silt, and Early Saalian sandy and siltysediments (Segli’s 1987; Kalni’a et al. 2000; Kalni’a2001). This marine sediment sequence is in turn coveredby thin layer of patchy, re-washed till supposedly ofMiddle Weichselian age (Saks et al. 2007, in press). Atthe top of the sequence, freshwater basin sediments morethan 40m thick are present, including dark grey clay-richsilt and fine-grained, sometimes silty, sand sediments(Segli’s 1987; Kalni’a et al. 2000). The Late Weichseliantill overlies the stratified basin sediments. It forms a dis-continuous cover of variable thickness (Saks et al. 2007,in press). Along the Aprini Plain, freshwater basin sedi-ments below the upper till are mostly replaced by theLate Weichselian glacifluvial deposits.

The glacial landforms of the Aprini Plain are re-presented by a small drumlin field surrounded by achain of marginal moraines (Figs 1B, 10). The drumlinshave to some extent been altered by a local, proglacialice-dammed lake (Veinbergs 1972; Straume 1979).Therefore drumlins are visible only in a limited areaand are mostly buried by a cover of glaciolacustrinesediments up to 5–9m thick (Figs 1B, 10).

Material and methods

This study is based on several fieldwork campaigns inwestern Latvia. During the fieldwork, attention wasfocused on the internal composition and deformationstructures of subglacial sediments. Detailed studies in-cluded the determination of spatial three-dimensionalarrangements of sediment units and structural fabricmeasurements. The fabric data were processed andplotted using STEREONET.

Numerous boreholes logged in the course of1:50 000-scale mapping were used as regional geologicalbackground information, and for compilation andreinterpretation of the geological cross-section.

Glacial landforms within the study area were studiedusing a digital terrain model of Latvia with grid stepof 20�20m, constructed by the Latvian GeospatialInformation Agency (LGIA). Topographic maps at thescale of 1:10 000 and aerial photographs were used fordetailed studies of the drumlin area.

Glacial sediments from the Strante site werestudied in thin sections prepared using a methodo-logy modified in accordance with Camuti & McGuire(1999) and Carr & Lee (1998) that involves sampleimpregnation with epoxy resin and, after resin hard-ening, proceeding with thin-section preparation simi-larly to in the case for hard-rock samples. There isno evidence of possible sand-grain reorientation duringthe thin-section preparation, as significant sampledeformation or disruption is needed to reorient rigidparticles locked in compact diamicton. The direc-tion, strength and spatial distribution of the appa-rent orientation of elongated sand grains were studiedwith the method of Kalv�ans & Saks (2008) usingthe data density plot as suggested by Fisher et al.(1985). Preferred orientation was calculated accordingto the eigenvalues method as used by Thomason &Iverson (2006).

Fig. 1. A. Location of the study area. B. Studyarea and its principal geomorphological setting.The white dashed line indicates the ice extentduring the Northern Lithuania (Oldest Dryas)deglaciation phase. The white dotted lineindicates the border between the Piemare andAprini Plains.

BOREAS Subglacial bed deformation and ice dynamics, W Latvia 125

Results

The AGT bed morphology was interpreted usingsatellite images and topographic data sets. Three keysites were selected for detailed geological study:Gudenieki, Ulmale and Strante (Figs 1B, 10B). Thesesites represent margins (Gudenieki and Strante) and amiddle portion (Ulmale) of the AGT.

The Gudenieki site

This is the northernmost of the outcrop sections dis-cussed (Figs 1B, 10B). The site is situated in the north-ern marginal zone of the AGT. The Pleistocenesediments at this site are typical for the Pleistocenesequence of the study area. Sand and silty sand sedi-ments are found at the bottom of the outcrop and be-low the upper till layer on the top of the section (Fig. 3).

Glaciotectonic structures at the site include a seriesof overthrusted sediment sheets, displaced in the NEdirection (Fig. 3). This is almost perpendicular to an

overall ice-flow direction that was from NNW to SSE(Gaigalas et al. 1967; Zelcs & Markots 2004). In-dividual thrust sheets are composed of a till layer at thebase capped by fluvial sandy sediments.

The thrust series starts with the youngest thrust sheeton the right-hand side of Fig. 3. Thrusting occurredtowards the margin of the ice tongue. All thrust sheetsdip in the same direction and form imbricated stacks ofoverthrust sheets. This kind of thrusting indicates asubglacial rather than proglacial origin. Proglacialthrusting is expected to produce an overthrust stack,given that the glacier is advancing such that youngerthrusts might be overlying older ones. Layering of thesandy strata is preserved, and brittle deformation sug-gests that sediments were frozen during thrusting.

At the southern end (Fig. 3, in the right part of theoutcrop sketch, and Fig. 4B) the thrust sequence iscross-cut by a diapir composed of diamicton, possiblyoriginating from the lower till unit. Along the diapirvery little deflection of strata is present, indicating thatthe upper part of the sandy sediments was in a frozen

Fig. 2. Geological cross-section of the Pleistocene sequence in western Latvia. The section is oriented approximately parallel to the moderncoastline. Figure modified after Kalni’a et al. (2000).

126 Tomas Saks et al. BOREAS

state during the formation of the diapirs. The upperpart of the diapir can be interpreted as a dyke, whichprotruded through a hydrofracture towards the glacierbase. Rijsdijk et al. (1999) emphasize that hydro-fractures originate at the glacier base, when waterpressures in the basal sediments exceed the overburdenpressure and tensile shear strength of the capping till. Itshould be noted that hydrofracturing in our study sitecould have been enhanced by the diapiric flow-inducedtensile stresses in the overlying sediments and cappingtill. Uplift of the diapir in the middle of the thrust se-quence caused subsidence of the sediments on eitherside, accompanied by normal faulting (Fig. 3).

The thrust sequence comprises part of a 4-m-highridge (Fig. 10B). The ridge can be traced for some 2 kmstretching in a WNW–ESE direction into the mainland.The long axis of the ridge is at right-angles to the mainstress direction of the thrust sequence. A similar land-

form has been suggested at the Sensala site, which issituated to the north of the Gudenieki site – at thesouthern margin of the Usma ice lobe in Sensala (Sakset al. 2007). A unidirectional deformation sequence ofthe ridge is exposed for at least 4 km.

The Ulmale site

This site is situated in the central part of the AGT. Thesediment sequence at the Ulmale site comprises mostlyshallow-basin fine-grained and silty sand sediments,overlain by a thick till layer (Fig. 5). The whole sectionis penetrated by diapirs composed of clayey silt (Figs4E, 5). The capping upper till over the topmost part ofthe diapirs is truncated to some extent by glacier ero-sion. The surface of the till was eroded by water of theBaltic Ice Lake.

Fig. 3. Geological cross-section showing the internal structure of the Pleistocene sequence at the Gudenieki site.


The upper till is heavily compressed and possessesfoliation typical for a basal till. Till thickness changessystematically and is controlled by the distribution ofthe diapirs (Fig. 5). It reaches up to 7m, but on averageis 2 to 3m, and is thickest in inter-diapir spaces (Fig. 5).The till contains a higher proportion of silt and claythan at the Gudenieki site, probably owing to enrich-ment from the underlying silt and clay.

We argue that the great thickness in the central partof the inter-diapir spaces is caused by till deformationand that, before the erosion, areas above the diapirswere covered by a very thin till only. Typically, tillthickness in the lowland areas is on average 2–3m

(Zelcs & Dreimanis 1997; Juskevics et al. 1998),and greater thicknesses typically occur in drumlins,morainic hills and other positive glacigenic landforms,whereas in the Ulmale site greater till thicknessescorrespond to negative topographic features (Fig. 5).

In the thickest parts of the till layer, shear zonestypically occur 2–3m above the till base and stretchlaterally for 20–30m (Fig. 6). The subglacial shear zoneis tilted, dipping into the inter-diapir space, indicatingthat the downward movement and the diapir upliftwere still present during the formation of the shear zone(Fig. 6). The till macrofabric maxima measured aboveand below the shear zone are dipping at approximately

Fig. 4. A. Part of the section at Jurklane,�5 kmnorth of the Gudenieki site. In the coastal bluffsoutside the Aprini glacial tongue area noglaciotectonic deformation is present, andsandy sediments are covered by water-lain till.B. Till diapir at the Gudenieki site, at160–175m. C. Inner core of the diapir, showingentrained sand material from beneath the siltyclay layer at 250–270m (see Fig. 5). D. Silt andclay diapir �2 km along the coastline to the SWof the Gudenieki section. E. Silt and clay diapirat the Ulmale site at �680–720m (see Fig. 5).


201–301, suggesting reorientation resulting from theoverall sinking of the inter-diapir spaces (Fig. 6). Weargue that the greater upper-till thickness in the inter-diapir spaces indicates that till accretion was controlledby the diapirism and was contemporaneous with theformation of the diapirs.

According to the borehole data (Juskevics et al.1998), the top of silt and clay sediments forming diapirslies at a depth of �30m, so presumably the verticaldisplacement of the diapirs is at least 30m (Fig. 2).

The diapirs in general are not only simple anticlinefolds, but rather comprise several small domes and

Fig. 5. Geological cross-section showing the internal structure of the Pleistocene sequence at the Ulmale site. The table at the bottom of thefigure denotes the fabric sampling number (Nr), number of measurements (No), mean orientation (V1) and pole to girdle (V3).


dissecting dykes associated with vertical shear zones ofmaterial transport (Fig. 5). These vertical shear zonespossibly represent hydrofractures, as described byRijsdijk et al. (1999), but fracturing resulting from dia-pirism may have also played a role in their formation.In places, the core of the diapir is composed of sand,suggesting entrainment of the sandy material from be-neath the clayey silt layer (Fig. 4C). Upward movementof the diapir may have led to flexing of the overlyingbeds, leading to their fracturing. Subsequently, a flui-dized sediment of the diapir may have flowed into thefractures and formed dykes. The topmost parts of dia-pirs are deflected down-glacier, suggesting that diapirswere affected by the drag force of the overriding glacier.The wavelength of diapirs at Ulmale site varies between150 and 300m. The width of individual structures var-ies, and does not seem to be correlated with the ampli-tude. Individual diapirs change from rather narrow andsteep to rather wide (up to 150m) (Fig. 5). The ampli-tude and wavelength of the diapirs seem to be con-trolled by the thickness of the sediment layer giving riseto them. At the Ulmale site, the thickness of clayey siltsediments that constitute the diapirs is the greatest ofthe whole study area (Fig. 2). Accordingly, the diapirsat the Ulmale site are the highest and widest recorded inthe study area.

The diapir profiles in the sections perpendicular tothe ice-flow direction are symmetrical, with few excep-tions. However, in the sections parallel to the ice-flowdirection, the down-glacier slope is much steeper thanthe up-glacier slope (Fig. 4D), indicating the influenceof glacial drag and that the diapirism occurred sub-glacially.

The basal contact of the till is sharp, in places markedby a shear zone up to 10 cm thick. It has been shownthat shear zones of similar thickness can develop in un-consolidated sandy sediments below an active warm-based glacier (Boulton 1996), as probably was the caseat this site.

It can be concluded that subglacial deformation atthis site was controlled mainly by the diapiric flow ofthe soft silty sediments. The geological structure of theUlmale site is typical for a distance of �12 km betweenGudenieki and Strante in the central part of the AGT.

The Strante site

The Strante site represents the southernmost section ofthe outcrops, where sediments subglacially disturbed byAGT are exposed (Fig. 7). The exposed part of theUpper Pleistocene sequence reveals dark grey silt

Fig. 6. Part of the shear zone in the upper till unit. Location given in Fig. 4, at 350m. Shear direction is oriented into the picture. Note the shovelfor scale. Ulmale_02 till fabric was measured just to the right of the picture.


diapirs similar to those observed at Ulmale. Fine, inplaces silty, sand and consolidated, very sandy dia-micton are present along with diapirs, which are allcovered by the discontinuous upper till unit.

The upper diamicton at Strante is up to 6m thick andforms an almost 120-m-wide span of the outcrops. It

consists mainly of poorly sorted, cemented fine sandand silt with occasional pebbles and soft sediment gla-ciotectonic rafts (Fig. 7). Planar foliation within thediamicton bends upwards near the diapir structures.The intrapore cement consists of silty particles and,probably, precipitated carbonates. Below the till, fine

Fig. 7. Geological cross-section showing the internal structure of the Pleistocene sequence at the Strante site. The table at the bottom of thefigure denotes the fabric sampling number (Nr), number of measurements (No), mean orientation (V1) and pole to girdle (V3).


sand and coarse silt sediments are deformed into trac-tion folds and rotation structures with dextral (top tothe left) shear sense. This unit is interpreted as a gla-ciotectonite, as described in Benn & Evans (1996). Itpossesses subtle planar, near-horizontal foliation androunded soft sediment raft inclusions (Fig. 7). Theoverall deformation of the sediment rafts is similar tothe deformation of tectonic inclusions in ductile shearzones, with characteristic rotational deformation fea-tures (Fig. 8). Below the sandy sediments are truncated

in the direction of the shearing (Fig. 7), indicating adextral shear sense. Small traction folds along the min-or shear surfaces within the diamicton have a similarsense of shearing. The unit is bounded by the diapirs.Deformation around diapirs does not interfere with theshearing of the silty sand. Macrofabric within the dia-micton as well as other shear-sense indicators suggest aNNE to SSW shearing direction, which is in goodagreement with the regional glacier movement direction(Gaigalas et al. 1967; Zelcs & Markots 2004).

The glaciotectonite was studied in thin sections (seeKalv�ans & Saks 2008 for methodology). In some thinsections, fine, probably tectonic, foliation was ob-served, including attenuated folds, mirrored by micro-fabric distribution (Fig. 9). In vertical thin sections(Fig. 9D), a distinctly bimodal summary microfabricdistribution was observed, with a stronger horizontaland a weaker vertical mode. In horizontal sections,poorly developed preferred microfabric was found,with the dominant orientation spread in a sector be-tween ENE and SSE (Fig. 9C).

In summary, the microfabric is characterized by awell-expressed subhorizontal planar summary orienta-tion with fabric strengths S1 calculated according to thetwo-dimensional eigenvalue method (see Thomason &Iverson 2006: S1=0.5 for a random distribution, andS1=1.0 for a unidirectional distribution) ranging from0.55 to 0.65 in vertical sections and from 0.53 to 0.60 in

Fig. 8. Fine sand sedimentary raft inclusion in the glaciotectonite atthe Strante site. Picture is taken at 365m, �5m above sea level (Fig.7). Rotational tails suggest dextral shear.

Fig. 9. Microfabric distribution as observed inthin sections of glaciotectonite at the Strantesite. A. Composite image in cross-polarizedlight of vertical thin section from the basal partof glaciotectonite. Light spots represent sandgrains, and the matrix is dark. B. Microfabricdistribution showing the fold-like pattern of abimodal signature. C. Examples of microfabricdistribution in horizontal thin sections showingweak summary microfabric. Image top isNorth. D. Examples of summary microfabricsin vertical thin sections showing a characteristicbimodal distribution. The diagrams representthe apparent orientation of elongated fine sandgrains (microfabric) presented as data densityplots (Fisher et al. 1985). Dark lines show themaximum clustering direction. The darker col-our of these lines indicates stronger fabric cor-responding to larger S1 eigenvalues calculatedaccording to Thomason & Iverson (2006).


horizontal sections. Thomason & Iverson (2006) foundthat the steady-state two-dimensional microfabricstrength S1 in vertical sections is �0.7 when extensionowing to shear strain reaches values of 7 to 39. Ac-cording to this indicator, sediments were subjected onlyto moderate simple shear (less than 7 to 39). However,the bimodal nature of the microfabric indicates that theformation of glaciotectonite was a more complicatedprocess than simple shear.

The Strante site possesses a similar deformationstructure assemblage to the Ulmale site. Similarly,the subglacial deformation is expressed by a diapiricflow. Deformation could have been enhanced by thefact that the Strante site is situated near to the lateralmargin of the glacial tongue. The glacier velocity is ex-pected to be lower near the lateral margin than in themiddle part (Ulmale site) of the tongue, which wouldlead to lower pore-water pressures and a higher degreeof glacier coupling, resulting in a deeper-rooted sub-glacial deformation.

Geomorphological imprint

The AGT configuration was retrieved from the mar-ginal landform assemblage based on digital elevationdata and a geological map of Quaternary surface sedi-ments. The flow path of the AGT is marked by elon-gated ridges on both sides of the glacier tongue thatembrace drumlin-like lineations and end with an arcu-ate end-moraine ridge in the western part of the Wes-tern Kursa Upland.

The largest end-moraine ridge is the Alm�ale-V�ardupemarginal ridge, clearly visible on the hill-shade imageas an arc transverse to the former ice-flow direction(Fig. 10A, B). It rises more than 30m above the ApriniPlain. The Alm�ale-V�ardupe marginal ridge has been

segmented by glacifluvial gullies that terminated in fansand deltas deposited into the Venta ice-dammed lake,which existed on the eastern side of the Western KursaUpland during the AGT advance. The second largestsegment of the marginal moraine chain is a lateralmoraine on the SW side of the AGT (Fig. 10A, B). Itis not situated on the slopes of the Western KursaUpland, and therefore its position may have beencontrolled by the presence of dead ice in the area.

The subglacial geomorphological imprint of theAGT is largely obscured as a result of erosion duringthe regression of the Baltic Ice Lake in the PiemarePlain and glaciolacustrine sedimentation of the Apriniice-dammed lake in the lowest part of the Aprini Plain(Fig. 10B). In the northeastern part of the study area,some flat-topped drumlins rise above the glaciolacus-trine plain (Fig. 10A). These streamlined landforms arewell pronounced in the hillshade image. The drumlinsare oriented in a WNW–ESE direction, marking theice-flow direction. A total of 21 drumlin or drumlin-likefeatures can be distinguished in the area. The lengthof these bedforms is 1.5–5 km, which is somewhatlonger than the average drumlin length in some classicaldrumlin fields (e.g. Zelcs & Dreimanis 1997; Clark& Stokes 2001; Stokes & Clark 2002b), and thuscomposite origin of these features cannot be excluded(Fig. 10A).

The onset of the drumlins at the AGT bed coincideswith the abrupt change in glacier bed geology (Figs 10,13), where a soft, deformable bed of loose sediments isreplaced by Upper Devonian dolomite. Along this line,the thickness of unconsolidated Pleistocene sedimentsdrops sharply from 40–50m to 10–20m.

Elongated, gentle ridges near Gudenieki and Strante,identified as shear margin moraines as described byKleman & Borgstrom (1994), Stokes & Clark (2002a)and Hindmarsh & Stokes (2008), are composed mostly

Fig. 10. A. Hillshade image of the Aprini glacialtongue advance area; 3� vertical exaggeration.1=margin of the occurrence of the dolomitenear the surface; 2=marginal moraines;3=drumlins. Note the gullies stretching fromthe marginal moraine chain at the right-handside of the picture. B. Geomorphological sketchof the study area.


of till (Fig. 10B). We use the term shear margin mor-aines in a wider sense – as subglacial ridges formed onthe border between fast and slow (or stagnant) ice,supporting the definition by Kleman & Borgstrom(1994), who described similar landforms on the borderbetween ‘glacially lineated and a relict (formerly cold-based) surface’. They suggested that such ridges arediagnostic thermal boundary landforms (Kleman &Borgstrom 1994). The orientation of these ridgescoincides with the orientation of drumlins stretchingparallel to the ice-flow direction of the AGT. The re-lative height of these ridges does not exceed 5–10m.The Gudenieki ridge is 200–400m wide and can betraced for a distance of about 2 km inland from themodern coastal bluffs. The outcrops at the Gudeniekisite display its internal structure in the section trans-verse to the flow direction, showing unidirectionalshortening of the subglacial sediments directed out-wards from the AGT bed. The Strante ridge is a dis-tinct, slightly sinusoidal ridge 400–800m wide, situatedhigher than the bedforms within the AGT bed, whichdoes not follow the topography. In parts, these ridgesare disrupted into segments, suggesting lateral migra-tion of the ice-stream bed.

The subglacial origin of the shear margin moraineshas been emphasized by Stokes & Clark (2002a). Theyreport ridges much longer than drumlins on the sides ofthe M’Clintock Channel palaeo-ice-stream bed. How-ever, they also support the idea that the shear marginmoraines could form on the sides of smaller and slowerice masses, such as an ice lobe or glacial tongue,surrounded by ice frozen to the bed (Fig. 4A). We arguethat the formation of the Gudenieki and Strante ridgeswas triggered by differential movement of ice masses ofthe tongue on one side and areas of dead ice on theother side (Fig. 4A).

Outside the pathway of the AGT, some lineamentstransverse to the ice-movement direction can be identi-fied (Fig. 10A, B). These could be interpreted as drum-linized features within the terrain created by previousglacial advances, which were directed from NNW toSSE rather than from W to E as with the AGT(�Abolti’s et al. 1972, 1977; Straume 1979; Raukas et al.1995). In places, these older lineaments are truncatedby younger ones. Figure 10 shows a linear feature thatcontinues in the NNE–SSW direction until it comesabruptly to an end at the margin of the AGT.

Discussion

The way in which the generation mechanics of glacialbedforms is understood has important implications forreconstructions of the history of glacial dynamics (VanLandeghem et al. 2009). Piotrowski et al. (2004) pro-posed the mosaic ice-bed deformation model to explainthe patchy warm-based glacier bed deformation. It was

proposed that the deformation of glacier bedsediments occurs when the shear stress exerted bythe glacier exceeds the shear strength of the sediments,resulting in a deformation spot. In subglacial systems,the most variable factor controlling sediment strengthis pore-water pressure. When pore-water pressureapproaches the floatation point of the overlying ice, se-diment strength is dramatically reduced, and deforma-tion can occur. However, when pore-water pressurereaches the floatation point, glacier ice can decouplefrom its bed, thereby precluding any bed deformation(e.g. Evans et al. 2006).

Splitting of the BIS into glacial tongues and areas ofstagnant ice in western Latvia was emphasized byStraume (1979) and Zelcs & Markots (2004). Bitinaset al. (2004), in their study of the morphology and dis-tribution of glaciolacustrine kame terraces in Lithua-nia, concluded that during the deglaciation stages vastareas, including the study area, were covered by activeand dead ice fields.

Here we will elaborate on the ideas of Piotrowski et al.(2004) and Jørgensen & Piotrowski (2003), who sug-gested that patterns of glacial bed deformation that canbe identified within the sediments can be used to re-construct the dynamic evolution of a paleo-ice stream.

Glacier bed deformation

We suggest that the diapir formation occurred con-temporaneously with glacier bed erosion and till accre-tion during the active ice flow and was directly initiatedby the reactivation (surge) of the AGT. This is sup-ported by the bending of the shear markers in shearzones and the lens-like shape of till bodies situated inthe inter-diapir spaces.

Moving over the diapirs, the glacier sheared and as-similated sand material above the rising diapirs and de-posited it in the subsiding spaces between the diapirs. Atthe Strante site, till is composed almost entirely of localmaterial. The coeval operation of both processes – dia-pirism and glacier bed deformation – is marked bysynclinally dipping till within the inter-diapir spaces, in-cluding distinct shear zones within the till, reflected alsoin the tilt of till fabric (Fig. 6). Thus the till accumulationwas controlled by the geometry and dynamics of the dia-pirs. Similar short transport of subglacial material wassuggested by Waller et al. (2007) during the formation ofdrumlins around proglacial outwash fans as a result ofthe recent surge of a modern glacier in Iceland.

The mode of glacier movement shifted locally fromsliding across the bed on the stoss side to bed deforma-tion and till accretion on the lee side of diapirs (Fig. 11).Pore-water pressure at the ice–bed interface abovethe diapirs during the diapirism was locally close to orexceeding the ice overburden, resulting in additionalloss of sediment strength and therefore enhanced ice


velocities and erosion. In inter-diapiric spaces, owing tothe sinking of the glacial bed material and extra volumecreated, pore-water pressure decreased, leading to pat-ches of basally coupled glacier, reduced ice velocitiesand till accretion (Fig. 11). Jørgensen & Piotrowski(2003) described the sequence of subglacial processesbeneath the Baltic ice stream on Funen Island,Denmark. They argued that the glacial bed deforma-tion, glacial dynamics and glacial bedform formationhave been controlled by the basal water pressurechange in the following stages: (i) glacial deformationduring the build-up of the pressure, (ii) sliding ormovement primarly by bed deformation during icestreaming with basal water pressure at the floatationpoint, and (iii) the termination stage of the ice streammarked by channelized subglacial drainage andesker formation. In contrast to this model, the AGTtermination stage is not marked by the development ofan efficient subglacial drainage system, probably at-testing to the surge behaviour of the AGT. As sug-gested byMurray et al. (2000), propagation of the surgefront was possibly associated with a thermal boundaryat the bed.

The diamicton was gradually deposited as tectonicslices (Evans et al. 2006), and compaction occurredtime-transgressively. New material was constantly ad-ded to the till accretion area, resulting in the bufferingof already accreted sediments from glacial shear stress.On a micro-scale, shear deformation produced sub-horizontal lineations. Isolated diamicton was subject tovertical compaction as a result of water expulsion, and

all the elongated grains except those oriented close tovertical were rotated towards the horizontal plane,which generated the bimodal fabric strength.

There are no superimposed deformation structuresaround the diapir tops that would be expected if theactive ice margin had passed across such distinct struc-tural perturbations. This indicates that the subglacialdiapir formation occurred during a very late stage ofdeglaciation or that the glacier remained decoupledfrom its bed after the diapirism.

The diapirs found in the central part of the AGTconsist of low-permeability material, for example silt,clay or till, that has a low capacity for pore-water pres-sure dissipation compared with the surrounding sandysediments. These diapirs moved up close to or evenreached the ice–bed interface and could have generateddeforming spots, according to the Piotrowski et al.(2004) model. However, the diapirs at Ulmale andStrante do not show any evidence of deformation. In-stead, glacial erosion occurred around these structures(Fig. 11), although deformation prior to the erosioncannot be excluded.

Hindmarsh & Rijsdik (2000) pointed out that in-stabilities in the uniformly layered medium either canbe caused by the viscosity gradient between the layersor can be driven by changes in the effective pressure. Inthe case of subglacial materials, effective pressure-dependent rheology affects Rayleigh–Taylor in-stabilities (Hindmarsh & Rijsdik 2000).

We consider that diapirism was a short-lived phe-nomenon, facilitated by the build-up of pore-water

Fig. 11. Conceptual model showing sedimentredistribution at the glacier bed caused byvertical movement of diapirs. A. Map view ofthe material transport from diapir-rising areasinto inter-diapiric subsiding spaces. B. Cross-section parallel to the axis of the glacial flow.C. Graph showing changes in the glacier slidingvelocity.


pressure, which exceeded the normal stress exerted bythe overlying ice masses and sediments. Such a situationcan occur if the gradual pore-water pressure build-upunder stable or slowly thickening warm-based ice isfollowed by a rapid ice thickness reduction in a surgeevent, which would rapidly lower the effective pressurewithin the subglacial sediments. The excess pore waterin the sand at the glacier sole would be drained to theice margin in a water layer at the ice base, in channels,or as groundwater flow. In contrast, pore-water pres-sure in the silt–clay strata below would remain highowing to lower permeability. Silt sediments then couldbecome unstable as the pore-water pressures within thesilt approached the floatation point. High pore-waterpressures within clayey silt also would lower the visc-osity of this unit, allowing it to flow. It has been shownthat high pore-water pressure and low normal stressesfor effective pressure-dependent rheology such as thesediments in question can lead to viscous deformationsuch as diapirism (Hindmarsh & Rijsdik 2000).

We suggest that diapiric flow occurred during the iceadvance in a very late stage of deglaciation, as there arenone of the superimposed deformation structures aroundthe diapir tops that would be expected if the active icemargin had passed across such distinct structural pertur-bations. It is likely that during the maximum advanceof the Late Weichselian ice sheet, the bed was frozen,inhibiting subglacial deformation. This is supported bythe presence of unconsolidated sediment interclastswithin the shear zone at the Strante site. In addition, thetill macrofabric in the inter-diapiric spaces correspondsto the flow direction of the AGT (Fig. 6) but not to thatof the BIS (Zelcs & Markots 2004).

Initially the active ice masses were compressed againstcold, inactive ice, which led to a build-up of ice thickness.At some point the critical shear stress was reached, andthe glacier started to flow rapidly, leading to a rapidlowering of the ice surface. Along with the lowering ofthe glacier surface, subglacial pore-water pressures drop-ped at the glacier bed. Pore-water pressures in fine-grained sediments deeper in the bed remained high owingto low water conductivity. Rapidly thinning ice cover ledto a simultaneous reduction of normal stress and rising ofpore-water pressures to the floatation point within thesilt, which could have triggered diapir formation. Shear-ing of the diapir tops and the deposition of relativelydense till in the space between them would further facil-itate the rise of the diapirs, as long as the high waterpressure below the silt was maintained.

It seems that the amplitude and wavelength of thediapirs are dependent on the thickness of the layer thatis prone to diapirism. At the central part of the AGT(Ulmale site), silt sediments reach their maximumthickness (up to 30m), which corresponds to the high-est and widest diapirs outcropped in the coastal bluffs.In the areas where weak sediments are thinner, diapirsare smaller and narrower. The diapir profiles suggest

formation under dynamic subglacial conditions. Dia-pirs formed as a result of geostatic pressure by over-lying ice and sediments at or near the ice margin wouldprobably have symmetrical profiles with respect to theglacier margin.

The configuration of the AGT

The configuration of the AGT during its last maximumextent can be deciphered from the assemblage of mar-ginal moraines, truncated older glacial landforms andlinear marginal ridges upstream interpreted as shearmargin moraines (Fig. 12B).

The shear margin moraines, as well as the older trun-cated glacial landforms, indicate that the AGT wasbounded by slow-moving or stagnant ice on its sides,suggesting that the dynamics of the AGT were controlledmostly by the side drag forces from the surrounding stag-nant ice. Side drag can support a significant amount of theice-stream driving force (Raymond et al. 2006). In themiddle part, where ice velocity is expected to be highest,ice flow occurred primarily by basal sliding (Fig. 12B).Side drag led to the formation of shear margin moraines.The sharp, abrupt changes of the subglacial deformationpattern below the AGT and the surrounding areas, andthe preservation of older glacial landforms next to theAGT, suggest that the BIS dynamics in the rather wide(up to 150km) marginal zone were changing abruptly.This is also reflected by the change of ice-flow direction,which was NNW to SSE during the earlier phase ofthe deglaciation and WNW to ESE for the AGT(Fig. 12A, B).

Shear margin moraines and the orientation of drum-lins suggest that the re-arrangement of BIS ice massesinto the AGT must have occurred as a short-lived eventrather than as a gradual collapse of the BIS into minorglacier tongues. Glacier ice in the inter-tongue areasbecame stagnant. It can be debated whether the BISstopped entirely prior to the next glacial re-advance.Christoffersen & Tulaczyk (2003b) argue that the BIScould have experienced stagnation phases during itsadvance, even in the mid-latitude regions. If this wasthe case, the BIS stopped at some point in westernLatvia and, during the next cold stage, reactivated inthe area as several glacial tongues.

Outcrops to the north of the Gudenieki site reveal thesame sandy–silt sediments topped in places by a till thathas been, to some extent, washed. These sediments pos-sess no signs of subglacial deformation. We argue that,in this area, the ice was cold-based during the AGTadvance. Clark & Stokes (2001) and Stokes & Clark(2003) suggest that the termination of an ice stream isaccompanied by the rapid formation of large-scale stickyspots, where ice becomes inactive and frozen to its bed.Accordingly, we propose that the termination of the BISin western Latvia was induced mainly by a change fromwarm-based to cold-based conditions.


Drumlins in the AGT area

A small drumlin field comprising 21 preserved land-forms occurs in the distal portion of the AGT area (Fig.10A, B). The onset of the drumlins in the area is fairlysharp, and occurs along one common line (Fig. 13).This line coincides with changes in bed geology. Alongthis line, soft Devonian sandstone with rather low wa-ter conductivity is replaced by hard, highly permeabledolomite (Kurss 1992; Juskevics et al. 1998).

We suggest that the change at the AGT bed from slid-ing to stick–slip behaviour and the formation of drumlinswere governed by changes in pore-water pressure, whichis partly a function of bed permeability (Fig. 13). Clark &Stokes (2001) in their description of the M’Clintockchannel ice stream note that the onset of drumlins coin-cides with changes in bedrock roughness at the ice-streambed. Rattas & Piotrowski (2003) emphasize that drumlinformation in the Saadjarve drumlin field was controlledmainly by pore-water discharge to the bedrock. TheAGT bed composition did not influence the glacier dy-namics, but changes in bedrock composition affected thepore-water pressure in the subglacial sediments. The do-lomite bedrock, owing to its high permeability, caused apore-water pressure reduction in the subglacial bed,which increased the glacier coupling to its bed and in-itiated the formation of drumlins (Fig. 13). It has beenshown that high basal water pressure, in the vicinity ofthe floatation point, is required to reduce the strength ofbasal coupling and initiate sliding (Jørgensen & Pio-trowski 2003), and therefore moderate changes in the

glacier bed water permeability can cause basal waterpressure to drop below the floatation point and initiatedrumliniziation. Enhanced drumlin formation in areas ofhigh bedrock permeability has been observed in theSaadjarve drumlin field, where the primary control onthe drumlin size would have been the subglacial pore-water pressure (Rattas & Piotrowski 2003).

Conclusions

The glacial dynamics of the AGT were governed mainlyby changes in pore-water pressure, which influenced gla-cial coupling to its bed. We argue that the highest glacierflow velocities during the AGT advance were reached bymeans of the glacier sliding over the surface of soft sedi-ments, while glacier movement due to the deformation ofthe subglacial sediments had secondary importance.Therefore, the onset of drumlins marks glacier re-cou-pling and deceleration of flow due to lowering of thepore-water pressure caused by a change of bed perme-ability as the glacier moved over different sediments. Ourstudy reconfirms that subglacial pore water strongly in-fluences the nature of glacial processes.

Diapir formation was a short-lived phenomenondriven by changes in the subglacial pore-water pressure,and probably reflects a surge-type glacial advance withrapid changes in glacial surface topography. The risingdiapirs and subsiding inter-diapiric areas created a mo-saic pattern of subglacial material redistribution.

Fig. 12. Palaeoglaciological reconstruction ofthe Aprini glacial tongue (AGT) flow area. A.At the beginning of deglaciation. B. During theAGT advance.


The assemblage of glaciotectonic structures suggeststhat shear margin moraines developed time-transgres-sively, and that the actual shear margin moraine ridgereflects a later stage of the ice advance.

The shear margin moraines of western Latvia marka transitional zone between fast-flowing ice and areasof slow or dead ice. Their internal compositionshows unidirectional deformation, typically achievedthrough thrusting. The main driving stress directionwas close to perpendicular to the overall glacial flowdirection.

Thin-section analysis of the glaciotectonite suggeststhat the material was in a viscous-plastic state, prob-ably induced by the high pore-water pressures.

The BIS experienced dramatic changes during degla-ciation. We argue that the BIS stagnated in the areaprior to the AGT advance, when the ice reactivated aslobes or tongues leaving dead ice in between.

Acknowledgements. – We thank the Latvian Geospatial InformationAgency (LGIA) for supplying the necessary data sets. We are gratefulto the Latvian Council of Science for financial support of this

Fig. 13. Possible subglacial hydrological conditions under the Aprini glacial tongue (AGT). A. Cross-section parallel to the AGT flow direction.The AGT profile is schematic and not to scale. B. Detail showing the ice-movement mechanism. C. Changes in pore-water pressure (Pw) andsediment shear strength (t�) at the glacier bed along the AGT flow line.


research. This study was co-financed by the European Social Fundproject ‘Starpnozaru zin�atnieku grupas un modewu sistemas izveidepazemes ude’u petıjumiem’, projekta lıguma nr: 2009/0212/1DP/1.1.1.2.0/09/APIA/VIAA/060. We are grateful to Mark Dayton forreviewing and correcting the English, as well as for providing veryuseful suggestions on the overall organization and style of the paper.We thank the referees Jasper Knight and Kenneth F. Rijsdijk forthorough reviews and comments, as well as Jan A. Piotrowski for va-luable suggestions that sharpened and clarified the paper.

References

�Abolti’s, O., Vaitekunas, P., Danil�ans, I., Karukapp, R., Klive, G.,Raukas, A., Rothko, L., Chebotareva, N. S. & Yanke, V. 1977: TheBaltic ice stream. In Chebotareva, N. S. (ed.): The Structure andDynamics of the Last Ice Sheet of Europe, 17–44. Nauka, Moscow(in Russian).

�Abolti’s, O. P., Veinbergs, I., Danilans, I., Stelle, V., Straume, J.,Eberhards, G. & Yushkevichs, V. 1972: Main features of glacialmorphogenesis and peculiarities of last ice sheet deglaciation in theterritory of Latvia. In Danil�ans, I. & �Abolti’s, O. P. (eds): Guide-book of the Field Symposium of the Third All-Union JointMeeting onMarginal Formations of Continental Glaciation, 3–16. Latvian StateUniversity, Riga (in Russian).

Alley, R. B. 1991: Deforming-bed origin for southern Laurentide tillsheets. Journal of Glaciology 37, 67–76.

Alley, R. B. 1993: In search of ice stream sticky spots. Journal of Gla-ciology 39, 447–454.

Benn, D. I. & Evans, D. J. A. 1996: The interpretations and classifi-cation of subglacially deformed materials. Quaternary Science Re-views 15, 23–52.

Bitinas, A., Karmaziene, D. & Jusiene, A. 2004: Glaciolacustrinekame terraces as an indicator of conditions of deglaciation inLithuania during Last Glaciation. Sedimentary Geology 165,285–294.

Boulton, G. S. 1986: A paradigm shift in glaciology? Nature 322, 18.Boulton, G. S. 1996: Theory of glacial erosion, transport and deposi-tion as a consequence of subglacial sediment deformation. Journalof Glaciology 42, 43–62.

Boulton, G. S. & Hindmarsh, R. C. A. 1987: Sediment deformationbeneath glaciers: rheology and geological consequences. Journal ofGeophysical Research 92, 9059–9082.

Boulton, G. S., Dongelmans, P., Punkari, M. & Broadgate, M. 2001:Palaeoglaciology of an ice sheet through a glacial cycle: the Eur-opean ice sheet through the Weichselian. Quaternary Science Re-views 20, 591–625.

Camuti, K. S. & McGuire, P. T. 1999: Preparation of polished thinsections from poorly consolidated regolith and sediment materials.Sedimentary Geology 128, 171–178.

Carr, S. C. & Lee, J. A. 1998: Thin-section production of diamicts –problems and solutions. Journal of Sedimentary Research 68,217–220.

Christoffersen, P. & Tulaczyk, S. 2003a: Response of subglacial sedi-ments to basal freeze-on 1. Theory and comparision to observa-tions from beneath the West Antarctic Ice Sheet. Journal ofGeophysical Research 108(B4), 2222–2238.

Christoffersen, P. & Tulaczyk, S. 2003b: Signature of paleo-ice-stream stagnation: till consolidation induced by basal freeze-on.Boreas 32, 114–129.

Clark, C. D. & Stokes, C. R. 2001: Extent and basal characteristics ofthe M’Clintock channel ice stream. Quaternary International 86,81–101.

Clark, P. U. 1994: Unstable behaviour of the Laurentide Ice Sheetover deforming sediment and its implications for the climatechange. Quaternary Research 41, 19–25.

Clark, P. U. 1995: Fast glacier flow over soft beds. Science 267, 43–44.Clayton, L., Mickelson, D. M. & Attig, J. W. 1989: Evidence againstpervasively deformed bed material beneath rapidly moving lobes ofthe southern Laurentide Ice Sheet. Sedimentary Geology 62,203–208.

Ehlers, J. 1996: Quaternary and Glacial Geology. 579 pp. John Wiley& Sons Ltd, Chichester.

Evans, D. J. A., Phillips, E. R., Hiemstra, J. F. & Auton, C. A. 2006:Subglacial till: Formation, sedimentary characteristics and classifi-cation. Earth-Science Reviews 78, 115–176.

Fisher, N. I., Huntington, J. F., Jackett, D. R., Willcox, M. E. &Creasey, J. W. 1985: Spatial analysis of two-dimensional orienta-tion data.Mathematical Geology 17, 177–194.

Gaigalas, A., Gudelis, V., Spri’gis, K., Konshin, G., Savvaitov, A.,Veinbergs, I. & Raukas, A. 1967: Orientation of pebble long axes inthe tills of the last glaciation in the East Baltic region and its rela-tion to ice sheet shrinkage. Baltica 3, 215–233.

Hindmarsh, R. C. A. & Rijsdik, K. F. 2000: Use of a viscous model oftill rheology to describe gravitational loading instabilities in glacialsediments. InMaltman, A. J., Hubbard, B. &Hambrey,M. J. (eds):Deformation of Glacial Materials, 191–201. Geological SocietyPublishing House, Bath. Special Publication 176.

Hindmarsh, R. C. A. & Stokes, C. R. 2008: Formation mechanismsfor ice-stream lateral shear margin moraines. Earth Surface Pro-cesses and Landforms 33, 610–626.

Jørgensen, F. & Piotrowski, J. A. 2003: Signature of the Baltic IceStream on Funen Island, Denmark during the Weichselian glacia-tion. Boreas 32, 242–255.

Juskevics, V., Kondratjeva, S., Murnieks, A. & Murniece, S. 1998:Geological map of Latvia. Scale 1:200,000. In �Abolti’s, O. P. &Kurss, V. (eds): Explanatory Note andMap, Sheet 41 – Ventspils, 48pp. State Geological Survey, Riga (in Latvian).

Kalni’a, L. 2001: Middle and Late Pleistocene environmental chan-ges recorded in the Latvian part of the Baltic Sea basin.QuaternariaA9, 173 pp.

Kalni’a, L., Dreimanis, A. & Murniece, S. 2000: Palynology andlithostratigraphy of Late Elsterian to Early Saalian aquatic sedi-ments in the Ziemupe-Jurkalne area, western Latvia. QuaternaryInternational 68-71, 87–109.

Kalv�ans, A. & Saks, T. 2008: Two dimensional apparent microfabricof the basal Late Weichselian till and associated shear zone: casestudy from western Latvia. Estonian Journal of Earth Sciences 57,241–255.

Kleman, J. & Borgstrom, I. 1994: Glacial landforms indicative of apartly frozen bed. Journal of Glaciology 40, 255–264.

Knight, J. 2002: Glacial sedimentary evidence supporting stick-slipbasal ice flow. Quaternary Science Reviews 21, 975–983.

Kurss, V. 1992: Devonian Terrigenous Deposition on the Main Devo-nian Field. 208 pp. Zin�atne, Riga (in Russian).

Maltman, A. J., Hubbard, B. & Hambrey, M. J. 2000: Deformationof glacial materials: introduction and overview. In Maltman, A. J.,Hubbard, B. & Hambrey, M. J. (eds): Deformation of GlacialMaterials, 1–9. Geological Society Publishing House, Bath. SpecialPublication 176.

van der Meer, J. J. M., Menzies, J. & Rose, J. 2003: Subglacial till: thedeforming glacier bed. Quaternary Science Reviews 22, 1659–1685.

Meirons, Z. & Straume, J. 1979: Cenozoic group. In Misans, J.,Brangulis, A., Danilans, I. & Kurss, V. (eds): Geological Structureand Mineral Resources of Latvia, 176–268. Zin�atne, Riga(in Russian).

Meirons, Z., Straume, J. & Juskevics, V. 1976: Main varieties of themarginal formations and deglaciation of the last glaciation in theterritory of Latvian SSR. In Danil�ans, I. (ed.): Problems of Qua-ternary Geology, 9, 50–73. Zin�atne, Riga (in Russian).

Murray, T. 1997: Assessing the paradigm shift: deformable glacierbeds. Quaternary Science Reviews 16, 995–1016.

Murray, T., Stuart, G. W., Miller, P. J., Woodward, J., Smith, A. M.,Porter, P. R. & Jiskoot, H. 2000: Glacier surge propagation bythermal evolution of the bed. Journal of Geophysical Research105(B6), 13491–13507.

Piotrowski, J. A., Larsen, N. J. & Junge, F. W. 2004: Reflections onsoft subglacial beds as a mosaic of deforming and stable spots.Quaternary Science Reviews 23, 993–1000.

Piotrowski, J. A., Mickelson, D. A., Tulaczyk, S., Krzyszkowski, D.& Junge, F. W. 2001: Were deforming subglacial beds beneath pastice sheets really widespread? Quaternary International 86, 139–150.

Rattas, M. & Piotrowski, J. A. 2003: Influence of bedrock perme-ability and till grain size on the formation of the Saadjarve drumlin


field, Estonia, under an east-Baltic Weichselian ice stream. Boreas32, 167–177.

Raukas, A., �Abolti’s, O. & Gaigalas, A. 1995: Deglaciation of theterritory. In Schirmer, W. (ed.): Quaternary Field Trips in CentralEurope, 1, 148–151. Verlag Friedrich Pfeil, Munchen.

Raymond, C. F., Catania, G. A., Nereson, N. & van der Veen, C. J.2006: Bed reflectivity across the north margin of Whillans IceStream, West Antarctica, and implications for margin migration.Journal of Glaciology 52, 3–10.

Rijsdijk, K. F., Geraint, O., Warren, P. W., McCarrol, D. & van derMeer, J. J. M. 1999: Clastic dykes in over-consolidated tills: evi-dence for subglacial hydrofracturing at Killiney Bay, eastern Ire-land. Sedimentary Geology 129, 111–126.

Saks, T., Kalv�ans, A. & Zelcs, V. 2007: Structure and micro-morphology of glacial and non-glacial deposits in coastal bluffs atSensala, Western Latvia. Baltica 20, 19–27.

Saks, T., Kalv�ans, A. & Zelcs, V. In press: OSL dating evidence ofMiddle Weichselian age of shallow basin sediments in WesternLatvia, Eastern Baltic. Quaternary Science Reviews.

Segli’s, V. 1987: Pleistocene Stratigraphy of Western Latvia. Com-pendium of Partial Fulfilment of Requirements of the Candidate ofGeological and Mineralogical Sciences. PhD Thesis, Academy ofSciences of Estonia, Tallin, 14 pp. (in Russian).

Stokes, C. R. & Clark, C. D. 2002a: Ice stream shear margin mor-aines. Earth Surface Processes and Landforms 27, 547–558.

Stokes, C. R. & Clark, C. D. 2002b: Are long subglacial bedformsindicative of fast ice flow? Boreas 31, 239–249.

Stokes, C. R. & Clark, C. D. 2003: The Dubawnt Lake palaeo-icestream: evidence for dynamic ice sheet behaviour on the CanadianShield and insights regarding the controls on ice-stream locationand vigour. Boreas 32, 263–279.

Stokes, C. R., Clark, C. D., Lian, O. & Tulaczyk, S. 2006: Geomor-phological map of ribbed moraine on the Dubawnt Lake Ice

Stream bed: a signature of ice stream shut-down? Journal of Maps2006, 1–9.

Stokes, C. R., Clark, C. D., Lian, O. & Tulaczyk, S. 2007: Ice streamsticky spots: a review of their identification and influence beneathcontemporary and palaeo-ice streams. Earth Science Reviews 81,217–249.

Straume, J. 1979: Geomorphology. In Mis�ans, J., Brangulis, A.,Danil�ans, I. & Kurss, V. (eds): Geological Structure and MineralResources of Latvia, 297–439. Zin�atne, Riga (in Russian).

Thomason, J. F. & Iverson, N. R. 2006: Microfabric and microshearevolution in deformed till.Quaternary Science Reviews 25, 1027–1038.

Van Landeghem, K. J. J., Wheeler, A. J. & Mitchell, N. C. 2009:Seafloor evidence for palaeo-ice streaming and calving of thegrounded Irish Sea Ice Stream: Implications for the interpretationof its final deglaciation phase. Boreas 38, 119–131.

Veinbergs, I. 1964: Coastal morphology and dynamics of the BalticIce Lake on the territory of the Latvian SSR. In Danil�ans, I. (ed.):Questions of Quaternary Geology III, 331–369. Zin�atne, Riga (inRussian).

Veinbergs, I. 1972: Marginal landforms (Western Latvia is used to il-lustrate). InGoretskiy, G. I., Pogulayev, D. I. & Shick, S. M. (eds):Marginal Formations of Continental Glaciations, 37–45. Nauka,Moscow (in Russian).

Waller, R., Van Dijk, T. A. G. P. & Knudsen, O. 2007: Subglacialbedforms and conditions associated with the 1991 surge of Skei-jararjokull, Iceland. Boreas 37, 179–194.

Zelcs, V. & Dreimanis, A. 1997: Morphology, internal structure andgenesis of the Burtnieks drumlin field, Northern Vidzeme, Latvia.Journal of Sedimentary Geology 111, 73–90.

Zelcs, V. & Markots, A. 2004: Deglaciation history of Latvia. In Eh-lers, J. & Gibbard, P. L. (eds): Quaternary Glaciations – Extent andChronology of Glaciations. Part I: Europe, 225–244Developments inQuaternary Science 2. Elsevier, Rotterdam.


Subglacial bed deformation and dynamics of the Apriķi glacial tongue, W Latvia

Documents