Interlobate ice-sheet dynamics during the Last Glacial Maximum at Whitburn Bay, County Durham, England

Authors


  • Bethan J. Davies (e-mail: Bethan.davies@durham.ac.uk), David H. Roberts, David R. Bridgland and Colm Ó Cofaigh, Department of Geography, Durham University, Science Laboratories, South Road, Durham DH1 3LE, UK; James B. Riding, British Geological Survey, Kingsley Dunham Centre, Keyworth, Nottingham NG12 5GG, UK; Emrys R. Phillips, British Geological Survey, Murchison House, West Mains Road, Edinburgh EH9 3LA, UK; Derek A. Teasdale, Durham University, Science Laboratories, South Road, Durham DH1 3LE, UK

Abstract

This research reconstructs ice-sheet processes operating during the Late Devensian in northeast England. The article assesses the lithostratigraphy of the Devensian glacial tills of Whitburn Bay, eastern County Durham, and presents the first detailed analysis of petrological, geochemical and biostratigraphical data to reconstruct lithostratigraphy, provenance and iceflow pathways. Two Devensian tractions tills (the Blackhall and Horden tills) are separated by a boulder pavement, pointing to a switch in ice-bed conditions and the production of a melt-out lag prior to deposition of the upper traction till, the Horden Till. The Blackhall Till contains Magnesian Limestone, Carboniferous Limestone, Whin Sill dolerite and Old Red Sandstone, suggesting a northwesterly source, probably from the Midland Valley and the Southern Uplands. The Horden Till contains erratics and heavy minerals derived from crystalline bedrock sources in the Cheviot Hills and northeast Scotland. Within the Horden Till are numerous sand, clay and gravel-filled canals incised downwards into the diamicton which are attributed to a low-energy, distributed, subglacial canal drainage system. Coupled with hydro-fractures and the boulder pavement, this suggests that a partially decoupled, fast-flowing ice stream deposited the Horden Till. The uphill, landward direction of ice movement indicates that the ice stream was confined in the North Sea Basin, possibly by the presence of Scandinavian Ice.

The behaviour of the British–Irish Ice Sheet (BIIS) along the eastern coast of Britain during the Last Glacial Maximum (LGM) is poorly understood (Catt 1991b; Carr et al. 2006). Competing ice lobes from both northwest Britain and Scotland overran the area (Lunn 1995), but the flow phasing of these lobes and their dynamic interaction have only been partly reconstructed. Whitburn Bay (Fig. 1) in County Durham, northeast England, is located in an area of coalescence of several competing ice lobes (Lunn 1995). Furthermore, there is little chronostratigraphic control on the glacial sediments associated with both the advance and retreat phases of the BIIS in this area.

Figure 1.

 Map showing location of sections investigated at Whitburn Bay.

It has been suggested that ice from northwest Britain crossed the northeast coast first, before ice sourced from the Cheviot/Tweed area flowed north to south down the coast (Lunn 1995; Teasdale & Hughes 1999). The sediments and landforms related to the latter phase of ice flow suggest that a surge lobe may have operated along this coast (Eyles et al. 1982, 1994; Catt 1991a, b; Douglas 1991; Evans et al. 1995; Teasdale & Hughes 1999; Boulton & Hagdorn 2006); this late phase re-advance of ice has been linked to Heinrich Event 1 forcing (McCabe et al. 2005). The coast-parallel flow direction of the ice has also been linked to deflection of the BIIS by Fennoscandian ice in the central North Sea (Boulton et al. 1991), but this remains contentious (Carr et al. 2006).

Trechmann (1915) was the first to propose the existence of two tills in County Durham (now called the Blackhall and Horden tills), where they cap the local Magnesian Limestone bedrock (Smith & Francis 1967; Francis 1972; Smith 1994; Bridgland 1999; Thomas 1999). However, there has been little quantitative description of these tills and no detailed analysis of their provenance, depositional processes or type and style of deformation.

Francis (1972) correlated the lower Blackhall Till with the Skipsea Till in the Holderness cliffs on the basis of stratigraphic position and lithology (Madgett & Catt 1978), the type locality of the LGM in Britain (Rose 1985), where it overlies the Dimlington Silts, dated to 21 475–22 140 cal. yr BP (Penny et al. 1969; Bateman et al. 2008). These correlations are not quantitative and are based on little empirical evidence beyond matrix colour, stratigraphic position and erratic content. The glaciofluvial Peterlee Sands (tabular sands and gravels) overlie the Blackhall Till and are in turn overlain by the Horden Till (Francis 1972). Catt & Penny (1966) proposed that the Horden and Blackhall tills were deposited from the dirty basal parts of a contemporaneous, two-tiered (stacked) ice sheet, producing two separate tills. However, the Horden Till is more likely to be a basal till from an ice sheet that overrode the Blackhall Till after the recession of Lake District ice (Beaumont 1967).

Detailed process-based research on Devensian glacigenic sediments further north in Northumberland has identified till sequences interpreted as lodgement till complexes with shoestring channels cut subglacially into the till bed, and often containing irregular debris masses. These were inferred to have originated from the glacier sole, dropping from the roof of the channel. Eyles et al. (1982) argued that variable palaeo-discharges were the result of ponding episodes, possibly resulting from the closure of the channel downstream by ice or deformation of the till bed. The tills and glaciofluvial sediments were all deposited during a single, complex, episode of wet-based subglacial sedimentation (Eyles et al. 1982).

It is also apparent that, during deglaciation, separation of westerly ice from the North Sea lobe resulted in the formation of regionally extensive lakes, such as glacial lakes Wear, Tees, Pickering and Humber (Raistrick 1931; Smith 1994; Bateman et al. 2008), but the ice-marginal controls on their formation are poorly constrained.

Hence, there is substantial scope for new research to contribute to the understanding of ice-sheet dynamics in the region during the Late Devensian. This work provides new empirical, quantitative data and detailed sedimentological analysis, critically testing the proposed genesis of the tills, provenance models, regional stratigraphic correlations and interaction between the British and Fennoscandian ice sheets during the Late Devensian. It reconstructs the glacial processes operating during the Late Devensian at the northeast margin of the BIIS, as typified in glacigenic sediments exposed at Whitburn Bay. First, macroscale and microscale lithofacies analysis is used to reconstruct the subglacial and ice-marginal processes, as well as patterns and phasing of flow occurring during ice-sheet advance and recession. Second, the article seeks to determine the ice-flow pathways in the region, using lithological, palynological, heavy-mineral and heavy-metal provenancing techniques. Third, the article considers the glacial land system operating along the east coast during the LGM and evaluates the evidence for a surging ice lobe in the North Sea. Finally, the implications of this research for regional Quaternary stratigraphic correlations are considered.

Methodology

Sediment analysis and lithofacies reconstruction followed standard procedures outlined by Evans & Benn (2004). Detailed vertical profiles of individual sections comprise a number of ‘units’, and lithofacies associations consist of a number of associated units exposed in multiple sections (shown in Fig. 1). Clast fabric and eigenvector analyses followed Benn (2004). Micromorphological sampling, preparation and analysis followed procedures outlined by Murphy (1986), van der Meer (1993), Menzies (2000), Carr (2004) and Hiemstra (2007).

Detailed quantitative analysis of the geochemical and lithological properties of the sediments enabled the reconstruction of iceflow pathways and source regions, and correlation and differentiation between lithofacies. Particle-size analysis and clast lithology followed procedures suggested by Gale and Hoare (1991) and Walden (2004). Particle-size analysis was conducted using a laser granulometer for particles less than 2000 μm and hand sieving for particles greater than 2000 μm. Stones were identified using a low-powered binocular microscope (model ‘Motic SMZ-168’), with reference to a standard reference collection, and using identification guidelines outlined by Walden (2004) and Gale & Hoare (1991).

Allochthonous palynomorph analysis was also used to establish provenance and help in the iceflow pathway reconstruction (Lee et al. 2002; Riding et al. 2003), based on known and established geological and geographical distributions. Samples were prepared using the sodium hexametaphosphate method outlined by Riding & Kyffin-Hughes (2006).

Heavy minerals were obtained by density separation using full freezing in accordance with techniques outlined by Gale & Hoare (1991), Mange & Maurer (1992), Walden (2004) and Mange & Otvos (2005). Between 200 and 300 non-opaque minerals in the 63–125 μm fraction were point-counted per sample. Opaque minerals are defined here following the pedological approach (rather than mineralogical), in that they are defined as minerals that are black in plane polarized light (PPL) and black or very dark brown in cross-polarized light (XPL). This was in order to enable comparison with other regional data sets (cf. Madgett & Catt 1978).

ICP-MS (mass spectrometry) (total metals extraction and atomic absorption) was used to determine the elemental composition of sediments (McClenaghan et al. 2000). ICP-MS gives data on abundant and trace elements within the samples, so the suite is split into two groups: high abundance metals (e.g. aluminium) and low abundance metals (e.g. bismuth). This method can be used to determine between-sample and within-sample heterogeneity, and to critically test stratigraphic correlations.

Multivariate statistical analysis was applied to the heavy mineral, metals and clast-lithology data in the form of principal components analysis (PCA), an eigen-analysis of a covariance or a correlation matrix (Kovach 1995). PCA based on a covariance matrix relates to the squared standard deviation within the variables, whereas PCA based on a correlation matrix relates to the skew of the variables; both were used to reduce the complexity of the Whitburn data set, as PCA axes simplify and represent the variation in the data (Davis 1986; see below).

Sedimentology, stratigraphy and lithological and geochemical data

Site location and general description

At Whitburn Bay (Fig. 1), glacigenic sediments are well exposed above the Roker Formation Magnesian Limestone bedrock (Figs 2, 3). Two lithofacies associations (LFAs) are identified as superimposed lower (LFA 1) and upper (LFA 2) diamictons separated by a boulder pavement. Incised into LFA 2 are units of laminated sand and clay, rippled sand and bedded coarse sand and gravel. The presence of two distinct tills supports interpretations by Francis (1972), but the stratigraphy crucially differs in relation to the Peterlee Sands; the fluvioglacial sediments are considerably more variable and complex than previously presented.

Figure 2.

 Simplified geological map. Derived from BGS Digimap GB-625. British Geological Survey ©NERC. All rights reserved.

Figure 3.

 Facies architecture at Whitburn Bay.

Lithofacies Association 1: A massive, matrix-supported diamicton

Macroscale sedimentology. – LFA 1, the lower diamicton, is inconsistently exposed at Whitburn Bay (Figs 3, 4). It is a yellowish brown (10YR 4/4), matrix-supported, over-consolidated diamicton with abundant fine to coarse gravel and cobbles (Table 1). The upper contact is sharp, unconformable and sheared. The diamicton contains faceted and striated pebbles of both local and far-travelled provenance (see below; Table 2). In places, it is fissile and faintly stratified. Clast fabrics show a moderately strong clustering around the S1 axis of 0.6 from northwest to southeast, with a low dip angle to the southeast. A boulder pavement is embedded in the diamicton.

Figure 4.

 Photograph of Lithofacies Associations 1 and 2 and the boulder pavement in Section 1, Whitburn Bay. The trowel is 20 cm long. LFA 1 is a matrix-supported diamicton with a boulder pavement at the top. The boulders are striated. The diamicton is stone-rich and is 60 cm thick. It is dark yellowish brown (10YR 4/4), it is fissile and it is faintly stratified. It has an unconformable upper contact. LFA 2 is a stone-poor diamicton with coal, Carboniferous Limestone and sandstone gravel. It has a sandy texture and is a dark greyish brown (10YR 4/3).

Table 1.   Average particle size distribution.
 LFA 1LFA 2LFA 3a
% Gravel11.8211.760.00
% Sand18.3617.3717.60
% Silt39.1038.8872.50
% Clay30.7033.999.90
Table 2.   Average clast lithology from Whitburn Bay, 8–32 mm. Sandstones are distinguished on their quartz, feldspar and arenite content.
 Clast
lithology
LFA 1
average %
LFA 2
average %
IgneousDiorite0.000.20
Granite0.000.21
Gabbro0.000.20
Rhyolite0.000.82
Andesite0.510.72
Porphyry0.170.83
Felsite0.000.17
MetamorphicSlate0.000.49
Schist0.000.75
Sandstone and sedimentarySandstone10.5912.09
Arenite Sandstone0.681.55
Quartzitic Sandstone8.056.11
Siltstone0.160.70
Breccia0.160.00
JurassicIronstone1.571.76
Mudstone0.802.11
TriassicBrown Orthoquartzite1.050.61
Red Orthoquartzite0.170.74
White Orthoquartzite0.350.37
White Vein Quartz0.000.43
PermianMagnesian Limestone48.4633.10
Yellow Sands1.542.79
Whin Sill Dolerite0.831.97
CarboniferousCarboniferous Limestone12.5518.12
Coal1.631.72
DevonianChert0.190.57
Old Red Sandstone0.670.66
Shale0.000.17
Ordovician and SilurianArkose Sandstone2.252.36
Greywacke7.615.63
Total 5121073

The boulder pavement. – The boulder pavement occurs consistently at the top of LFA 1 and is laterally extensive across Whitburn Bay. It consists of well-faceted, striated boulders up to 50 cm diameter (Fig. 4), predominantly of Carboniferous Limestone, Magnesian Limestone and sandstone. The spacing between the boulders here ranges from a few centimetres up to 60 cm. In between and above the boulders there are channelized, massive, poorly sorted gravels with a sharp, undulating lower contact. The boulders are planated and polished on their upper surfaces and are either horizontally orientated or dip to the north–northwest. The mean dip direction of the boulders is 141° and the mean dip is 8.5°. Multiple measurements on many of the boulders indicate a consistent and strong striae orientation from north–northeast to south–southwest. These overprint fainter and less numerous striae orientated from northwest to southeast. Boulders lodged below the pavement into LFA 1 have numerous and well-clustered striae orientated from northwest to southeast (Fig. 3).

Micromorphology. – In thin section, LFA 1 is a light brown, matrix-supported, massive diamicton of even density (Table 3). Rotational structures and Type II pebbles are common, and there is a strong skelsepic plasmic fabric. The skeleton grains are poorly sorted. Larger skeleton grains exhibit edge-rounding; smaller grains are generally angular. Elongated grains are commonly lineated and aligned. The plasmic fabric is localized and patchy, with a strong skelsepic plasmic fabric around skeleton grains (Table 3).

Table 3.   Summary diagram of micromorphological analysis. One dot indicates that a feature is present. Two or three dots indicate that it is more common. Voids: L=laboratory-induced; P=packing-induced. Shape: SR=subrounded; SA=subangular; R=rounded; A=angular. Texture: C=coarse; M=medium; F=fine. Section elements: Be=bedding; Ba=banding, S=shear; Bo=boudinage.
Sample Skeleton grainsTextureVoidsDeformation structuresFluvial/marine featuresPlasmic fabric
Shape<500 μmShape>500 μmTextureVoidsVoid typeSection elementsRotationPressure shadowCrushed grainsPebble IPebble IIPebble IIIWater escapeLineationsFoldsDropstonesMicrofossilsLaminaeLattisepicSkelsepicOmnisepicMasepicKinking
LFA 11_1SRSAFL ••• •• •••        •••   
1_2SRSAFL ••• ••  ••     ••• •• 
Pipes2_1SRSRF/M/C••LS; Be      ••••••••  ••• •••  
LFA 210SR/SASR/SA    •••  ••       ••• ••• 
1_3SRSAF/CL/PBa••• •• •••  ••    ••• ••• 
1_4SRSAFP ••  ••  ••     ••• •• 
LFA 3b2_2SR/SAAFL/PF•••      •••    •• ••• 
9a_1 SA/AM  Be               
9a_2 SA/AM  Be       ••        
9a_3SR/SASR/SAM/CLBe; Bo      •••• •••   

Sample 1–3, which crosses the boundary between LFA 1 and LFA 2 in Section 1 (Fig. 5), is a brown, silty, matrix-supported diamicton divided by a deformed, folded bed of clast-supported skeleton grains with sharp contacts dissected by microscale water-escape structures. There are rare microfossils, possibly originating from the limestone clasts present. The sample contains well-developed rotational structures with associated pressure shadows, lineations of aligned, elongate skeleton grains and rounded Type II and Type III pebbles (van der Meer 1993). These intraclasts are coherent, well-rounded and display no intergranular disintegration or stringers. Small numbers of crushed grains and necking structures between grains are also apparent (cf. Menzies 2000) (Table 3). Microfabric analysis indicates a strong degree of alignment of skeleton grains. There is a strongly developed bimasepic and skelsepic plasmic fabric, where clay-sized matrix material is abundant.

Figure 5.

 Photograph, photomicrographs and sketch of thin section Sample 1a-3 taken from the boundary between LFAs 1 and 2 in Section 1. LFA 1 and LFA 2 in thin section are separated by a bed of clast-supported sand. Extensive deformation and mixing can be seen between the two diamictons. On a microscopic scale, Area 4 is folded and strongly deformed, and shows stringer initiation into Area 3. Boudinage structures are associated with this.

Lithological and geochemical analysis. – Three bulk samples (from Sections 1, 2 and the pipes in 2) and two gravel samples (from Sections 1 and 2) were analysed from LFA 1. The samples reacted vigorously to 10% hydrochloric acid. Magnesian Limestone dominates LFA 1 (Table 2) with 48.5% of the clasts, and Carboniferous Limestone is present in smaller amounts (12.6%). There are also relatively low amounts of Whin Sill dolerite erratics (0.8%) and very low amounts of quartzite (1.6%), andesite (0.5%) and porphyries (0.2%). Other erratics within LFA 1 include Old Red Sandstone (0.7%), greywackes (7.6%) and coal (1.6%) (Table 2). Notably, there is a lack of distinctive Lake District erratics and granites (cf. Francis 1972).

Metals analysis of the matrix by ICP-MS indicates that LFA 1 is high in silicon, sodium, magnesium, aluminium, potassium, calcium, iron, manganese and titanium (Table 4). The most prominent heavy mineral is dolomite (excluded from Table 5 because of strong skew). The far-travelled suite contains an abundance of medium-grade metamorphic minerals such as clinozoisite (14.6%), andalusite (9.7%), kyanite (9.6%), garnet (16.6%) and olivine (5.6%). Biotite (1.8%) and tourmaline (3.5%), common in many igneous rocks, are also present.

Table 4.   Average heavy metal data from Whitburn Bay.
Average concentration (mg/kg)
LithofaciesLi7Be9B11Si14Na23Mg24Al27K39
 LFA 158.751.550.52786883822.51707.52960011550
 LFA 267.75267.25301196437515253247511725
 LFA 3b4114130.67365034001890013100
LithofaciesCa44Ti48V51Cr52Fe57Mn55Co59Ni60
 LFA 1174753762.576.25753177538913.7539.25
 LFA 215800436588.2587.25352005291544.75
 LFA 3b3060030705350248005951028
LithofaciesCu63Zn66As75Mo98Ag107Sb123Ba137Pb206
 LFA 12262.258.254.751033719.25
 LFA 225.5518.257.251.50.25350.522.75
 LFA 3b1744821040423
Table 5.   Average heavy minerals (percent non-opaques) in glacial deposits at Whitburn Bay (excluding carbonates due to strong skew). Combined results of multiple samples.
 SampleLFA 1LFA 2LFA 3b
Silicate groupn39026691528
% Opaques76.2269.9184.26
% Non-opaques23.7830.0915.74
% Heavy minerals2.378.2813.82
Olivine group5.543.775.02
Zircon3.725.699.62
Sphene1.561.621.67
Garnet group16.5718.5925.52
Sillimanite0.453.240.00
Andalusite9.709.713.35
Kyanite9.618.784.18
Staurolite1.832.770.42
Chloritoid0.220.620.00
Tourmaline group3.502.908.37
Epidote groupZoisite/clinozoisite14.617.805.86
Epidote2.352.580.84
Pyroxene groupEnstatite2.471.012.51
Hypersthene0.210.091.26
Diopsidic clinopyroxene1.853.144.18
Augitic clinopyroxene0.871.541.67
Amphibole groupTremolite0.540.000.00
Actinolite0.310.000.00
Hornblende0.570.772.93
Mica groupMuscovite9.1611.635.44
Biotite1.772.370.84
Chlorite group5.682.875.02
OxidesRutile0.632.302.93
Brookite0.561.021.67
Spinel group0.230.610.00
Anatase0.170.350.00
SulphatesBaryte0.310.190.00
SulphidesSphalerite0.000.000.42
PhosphatesApatite5.014.046.28

Lithofacies Association 2: Massive, matrix-supported diamicton with coarse sands and gravels and laminated sand and clay

LFA 2 comprises several different associations of lithofacies, including a dark brown, matrix-supported diamicton with abundant fine gravel but rare coarse gravel (Fig. 4; Table 1), pipe-injection structures, several different facies of laminated sand and clay bodies nested within the diamicton, and coarse sand and gravel facies.

LFA 2: Diamicton facies. – The general characteristics of the diamicton are the rare coarse gravel clasts, ranging from subangular to angular, which are faceted and striated, and clast fabrics indicating a moderate to strong clustering of a-axes from north–northeast to south–southwest and northeast to southwest. LFA 2 is a dark brown colour (10YR 3/3) and contains far-travelled erratics (see below) (Table 2). In some places the two diamictons (LFA 1 and LFA 2) are exposed in superposition (Fig. 3), but in others LFA 2 overlies either bedrock or sands and gravels.

LFA 2 is structurally highly variable. For instance, it exhibits a number of subvertical, slightly overturned, pipe-injection structures (Figs 3, 6), and contains planar-bedded, well-sorted, fine sand incised unconformably downwards into the diamicton. The injection structures are similar in colour to LFA 1; they occur repeatedly along Whitburn Bay and appear to be associated in particular with the laminated sands (Fig. 3). In other places, such as in Section 10 (Fig. 3), a coarse, poorly sorted sand and gravel rests directly on soft, dolomitized bedrock (Unit 1; LFA 2) (Fig. 7A). This is unconformably overlain by a stone-rich, massive, dark yellowish brown (10YR 3/3) diamicton (Unit 2; LFA 2). The diamicton is truncated by a coarse, poorly sorted sand and gravel, with a convex, undulating base (Unit 3; LFA 2) (Fig. 7A). Above this unit, coal grains pick out planar-laminated fine sands and sand-silt couplets (Unit 4; LFA 2). Units 5 and 6 are laminated sands and silts deformed by dewatering, loading-style soft-sediment deformation such as flame and ball and pillow structures. Unit 7 above contains extended climbing Type A ripples. Within these sands is a small, lenticular, detached unit of diamicton, surrounded by flame, ball and pillow and other dewatering structures. This is overlain by planar-laminated sand and clay (Unit 8; LFA 2) (Fig. 7A) truncated by the overlying, dark brown diamicton (LFA 2; Unit 9). This diamicton is stone-poor, dark brown and massive.

Figure 6.

 Photograph of pipe structures in Section 2a. Unit 1 (LFA 1) is a massive diamicton, coloured 10YR 5/2. It is gravel-rich and has a sharp, erosive upper contact. It rests directly on Magnesian Limestone bedrock. Unit 2 (LFA 2) is a massive, gravel-poor, silty diamicton coloured 10YR 3/3. Granite, Carboniferous Limestone and coal are common. Pebbles are striated and faceted. Within Unit 2, pipe structures have been intruded upwards from Unit 1. These have a sandy texture and are overturned to the south (180°). They are coloured 10YR 5/3. Unit 3 contains thinly cross-bedded, well-sorted, fine sands.

Figure 7.

 Photograph of Section 10 with thin section and photomicrographs of Sample 10, taken from the basal diamicton of Section 10. The slide shows characteristic, strongly birefringent plasmic fabric, Type II Pebbles, lineations of clasts, rotational structures and grain stacking indicative of ductile deformation in a high strain environment.

LFA 2: Rippled and laminated sand and clay facies. – The nested lenses of sands within LFA 2 include sequences of laterally variable, laminated, draped-rippled and deformed sands and clays (Fig. 3). Although each channel fill has a unique sedimentary signature, they consistently have closed edges and flat tops, and generally consist of discrete units of interlaminated sand and clay that have undergone varying degrees of soft sediment loading-style and dewatering deformation (cf. Mills 1983). In a number of locations, channel walls have partly disintegrated, with diamicton fragments dropping into the channels.

Typical characteristics include Type A climbing ripples overlain by alternating sand and clay planar laminations often deformed by soft-sediment deformation. In Section 9a, there is a diamicton (LFA 1) with a boulder pavement at its top, resting on Magnesian Limestone bedrock (Fig. 8). A rippled sand unit (Type A ripples) (Allen 1963) with a scoured lower contact and a pebble lag near the base overlies this unit. The sand is heavily deformed with convoluted bedding and recumbent folds overturned towards the south. Type A climbing ripples above this deformed unit grade conformably into a massive sand. The interbedded sand and clay unit above this is strongly deformed: the clay beds have been folded and the sands show extensive evidence of dewatering with flames and ball and pillow structures. This is overlain by a planar laminated sand and clay unit capped by a sandy, massive diamicton (LFA 2) with an erosive, sharp lower contact. Adjacent sections exhibit repeated changes between planar-laminated bedded sands, interlaminated sand and clay couplets and Type B climbing ripples (Allen 1963) with ripple drapes.

Figure 8.

 Vertical profiles of Section 9. Section 9a begins with 20 cm of coarse, poorly sorted sand (10YR 4/3) resting on bedrock, which is overlain by 75 cm of planar laminated sand and clay (10YR 3/4) with a sharp lower contact. These conformably grade into Type A climbing ripples (Unit 3). The ripples are unconformably overlain by a matrix-supported, massive diamicton (10YR 4/4; Unit 4). This is overlain by Type B clay draped-ripples and then by planar-bedded sand and clay (Unit 5; LFA 3). Section 9b has a well-sorted, fine, laminated sand (Unit 1) at the base, which is overlain by a well-sorted sand with ball and pillow and flame structures. Embedded within this is a diamicton with sharp, scoured contacts. It is overlain by ripple and planar-laminated fine sand and clay. Section 9c starts with a strongly deformed, fine, well-sorted sand, overlain by a massive fine sand and then by a clast-rich diamicton with sharp, erosive contacts. This is overlain by 0.5 m of laminated sand and clay. Section 9d starts with a massive, over-consolidated gravel-rich diamicton (Unit 1). This contains stringers emanating from the Magnesian Limestone bedrock. It is directly and unconformably overlain by heavily deformed sand with a sharp, scoured lower contact, and features recumbent folds, overfolds and convolute lamination (Unit 2). This is overlain by Type A climbing ripples and then by more strongly deformed sand, and grades into Unit 3, planar-laminated sand. In Section 9d, Unit 4, clay and sand laminations are strongly folded and anastomosed into overfolds and recumbent folds. Unit 5 consists of planar-laminated sand and clay, with a sharp, erosive lower contact. Unit 6 (LFA 2), a dark brown, massive, gravel-poor diamicton caps the sequence.

LFA 2: Coarse sand and gravel facies. – Numerous medium to coarse, clast-supported gravel dykes with sharp, erosive contacts vertically dissect the diamicton (Unit 1, LFA 2) in Section 5 (Fig. 9). Smaller dykes occur as branched offshoots from the larger dykes (Unit 2). The clastic dykes have clear-cut, sharply erosive boundaries. Juxtaposed to the coarse gravels is an infill of fine, calcreted, thinly bedded sands (Unit 3, LFA 2), which are strongly deformed and contorted. These dykes transform laterally into subvertically orientated crudely bedded gravel strata (Unit 4) that fan out into the overlying diamicton unit (LFA 2). The lithologies of the gravel are similar to those of the underlying diamicton (LFA 2), with Magnesian Limestone the most abundant lithology. Intraclasts of diamicton occur within the gravels. There are also channels exposed in Sections 6, 7 and 10, where a channelized, coarse, poorly sorted, bedded sand and gravel is well exposed (Fig. 3). For example, the incised gravel bed in Section 7 exposes a well-sorted, planar-bedded, cobble gravel with a sharp, erosive, undulating bottom contact and a flat top contact.

Figure 9.

 Photographs of Section 5. Unit 1: Massive, matrix-supported diamicton, 10YR 3/3, dark brown. Gravel-poor. Unconformable, sharp contacts with Unit 2. Unit 2: Coarse, massive, poorly sorted, clast-supported gravel dyke. Coarsest gravel in the centre. Gravel is subangular to angular. Sharp, erosive, unconformable contacts. Unit 3: Horizontally bedded, calcreted, strongly deformed and contorted fine sand. Yellowish-brown colour (10YR 5/4). Unit 4: Planar-bedded, coarse sand and gravel, matrix-supported. Unit 5: Planar-bedded, well-sorted, coarse sand. Fines upwards into well-sorted fine sand and clay. Pebble lag at base.

Micromorphology. – LFA 2 is a dense, dark brown diamict with a small number of subrounded to rounded, matrix-supported skeleton grains, predominantly in the fine sand to silt size fraction (Table 3). Sample 10 (Fig. 7) shows a highly variable structure. The limestone bedrock has been drawn up into the diamicton, forming a stringer structure, and matrix material has been squeezed downwards into the limestone. Numerous smaller angular flakes and coal grains are evenly distributed across the slide. Aligned clay particles along the boundary of the diamicton and bedrock indicate shearing (Table 3; Fig. 8). There is a strong presence of rotational structures, Type II pebbles and a strong skelsepic and masepic plasmic fabric.

Thin-section analysis of the interlaminated channel fills shows both primary sedimentary structures and secondary deformation. Samples 9a-1 and 9a-2 (from Units 1 and 2 of section 9a) (Fig. 8) show a predominance of thinly bedded sands. The well-sorted fine sand skeleton grains, with rare larger coal grains, are mostly clast supported, and are predominantly subangular to angular. Matrix material is largely absent apart from in small, localized, patches distributed across the slides (Table 3). Elongate grains are aligned subparallel with each other.

Sample 9a-4, from Unit 4, has macroscopic, upward-fining sand and silt-clay laminations, is normally faulted and has recumbent folds (Fig. 10). The moderately sorted sand laminations are clast-supported. Skeleton grains are subrounded to subangular, with rare larger grains including sandstone, limestone and fossiliferous coal. The inversely graded coarse laminations have sharp contacts and contain thin lenses of masepic plasmic fabric. In contrast, the normally graded fine sand and silt laminations are matrix-supported, with graded basal contacts. Within the fine clay laminations, there are rip-up clasts and numerous water-escape structures (Fig. 10).

Figure 10.

 Photomicrograph of deformed bedding in thin section sample 9a-5, demonstrating strong masepic plasmic fabric development, which is highlighted on the sketch.

In Section 2 (Fig. 3), a channel incises into the underlying diamicton (LFA 1). The interlaminated sands and silts within the channel are heavily deformed, and a series of recumbent folds are overturned to the south. A thin section sample taken from one recumbent fold (axis c. 70°) demonstrates polyphase brittle and ductile deformation (Fig. 11). The primary sedimentary lamination is normally graded and consists of planar and ripple cross-lamination. Some masepic plasmic fabric development is apparent subparallel to the bedding. The sequence is compressively folded from a northerly direction and then cross-cut by two phases of normal faults, which have triggered water escape structures that have pierced the clay laminae and caused fluidization of the coarser sand laminae (Fig. 11).

Figure 11.

 Microfaults and water escape structures relating to overfolding in Section 2b. Primary sedimentary lamination (graded bedding with planar and ripple cross-lamination). Compaction causes masepic plasmic fabric development subparallel to bedding. Compression from the north caused overfolding, with the offset nature of the faults being conjugate and indicating vertical compression. Faulting has been accompanied by liquidization of sands and water escape.

Geochemical and lithological analysis. – Four bulk samples from sections 1, 2, 4 and 10, and three hand-picked pebble samples from sections 1, 2 and 10 were taken from LFA 2.

Both LFA 1 and LFA 2 diamicts have similar particle-size distributions (Table 1). LFA 2 has a weak to moderate reaction to hydrochloric acid. It is poorer in Magnesian Limestone (33.1%) and richer in Carboniferous Limestone (18.1%) than LFA 1 (Table 2). There are slightly higher percentages of igneous clasts present, including pink porphyries (0.8%), rhyolites (0.8%) and granites (0.2%) typical of the Cheviots region (Fig. 2), schist typical of the Scottish Highlands and Whin Sill dolerite (2.0%), a local Permian igneous lithology.

ICP-MS (Table 4) reveals that LFA 1 and LFA 2 have similar elemental compositions. The heavy-mineral suite has significant amounts of clinopyroxenes (4.68%) and kyanite (8.8%) and small amounts of rutile (2.3%) and baryte (0.2%), differentiating LFA 2 from LFA 1 (Table 5). Detrital rutile is sourced from high grade, regionally metamorphosed terranes or sediments, while amphiboles are likely to be sourced from various crystalline bedrock types.

Palynological analysis indicates that abundant wood fragments and well-preserved palynomorphs are present, with lower proportions of non-woody plant tissue (Riding 2007). The palynomorphs are dominated by the long-ranging Carboniferous spores Densosporites and Lycospora pusilla. Lower numbers of Endosporites globiformis, Florinites spp. and Radiizonates spp. were also present, and are indicative of the Westphalian (Smith & Butterworth 1967). Tripartites trilinguis and Tripartites vestustus suggest some Namurian input, possibly from the Newcastle coalfield.

The fine laminated sand facies was sampled in Section 9a. It is a yellowish-brown, well-sorted sand (Table 1) that reacts vigorously to HCl. The metals suite (Table 4) has noticeably higher proportions of magnesium and calcium than the diamicton samples. The sand facies is comparatively rich in igneous or high-grade metamorphic minerals, such as clinopyroxenes (5.9%), rutile (2.9%) and amphiboles (2.9%) (Table 5).

Interpretation

Lithofacies Association 1

Subglacial traction till. – LFA 1 has the macro-scale characteristics of a subglacial traction till (Evans et al. 2006). These include striated, far-travelled, faceted clasts, far-travelled heavy minerals, an over-consolidated, matrix-supported structure and well-orientated, clustered clast fabrics (Benn & Evans 1998; Evans et al. 2006). The micromorphological analysis also supports an interpretation of LFA 1 as a subglacial traction till. Associations of rotational structures, rounded Type II and III pebbles, grain lineations, masepic plasmic fabrics and skelsepic plasmic fabrics denote ductile deformation (van der Meer 1997; Hiemstra & Rijsdijk 2003) and grain rotation in a till matrix (van der Meer 1997; Nelson et al. 2005). The rotation of skeleton grains has caused preferential alignment of clay particles (Roberts & Hart 2005) through the transmission of stress perpendicular to the particle edges. Crushed grains imply high stress, low strain conditions in a brittle environment (Menzies 2000). The combination of both brittle and ductile deformation structures suggests polyphase deformation. The common rotational structures and the lack of flow, marbling and tile structures preclude a genesis by mass movement or debris flow (Lachniet et al. 2001; Menzies & Zaniewski 2003). Sample 1–3 (Section 1; Fig. 5) shows where LFA 2 was emplaced over the LFA 1. The two varieties of masepic (bimasepic) plasmic fabric indicate strong shear in two directions. The numerous soft sediment clasts indicate cannibalization of pre-existing sediments.

LFA 1 is thus interpreted as a terrestrial subglacial traction till with evidence of both lodgement and deformation, which varied spatially throughout Whitburn Bay. The striated clasts with aligned long axes and the boulder pavement in Section 1 are indicative of lodgement (Menzies et al. 2006). In contrast, LFA 1 in other sections (e.g. 2a, 3 and 10) at Whitburn Bay demonstrates structures indicative of extensional deformation, such as stringer initiation, weak clast fabrics, massive, homogenous diamictons and micromorphological evidence of extensive ductile deformation.

The boulder pavement. – The boulder pavement occurs at the top of LFA 1. Striations and clast fabrics are northeasterly, consistent with fabric evidence from LFA 2, indicating that the second phase of ice flow directly overrode the boulders. There are several competing theories for boulder pavement formation, ranging from lodgement (Boulton & Paul 1976), deformation with formation by the sinking of clasts through the deforming layer (Clark 1991; van der Wateren 1999; Glasser et al. 2001) and a continuum between deformation, melt-out and lodgement with subsequent extensive abrasion, truncating and striating the upper surfaces (Hicock 1991). Boulton (1996) maintained that during erosion of the till down to the A/B interface, dense clasts, on being exposed, resist being drawn into glacier flow and remain immediately above the A/B interface, concentrating larger clasts from the mobilized till. Jørgensen & Piotrowski (2003) argue that a boulder pavement is an erosional surface, indicating truncation and removal of underlying sediments. Lodgement and extensive abrasion at the sole of the ice sheet truncates, orientates and striates the upper surfaces of boulder pavements.

A simple erosional mechanism best explains the boulder pavement at Whitburn Bay (Fig. 12). As the boulders are orientated, planated and striated strongly in one direction, it is unlikely that they have sunk through the deforming layer, as this would have led to multiple striae directions and to rotation of clasts (Jørgensen & Piotrowski 2003). The clustering of the boulders at one level suggests that they are some sort of lag deposit. This may be the result of the down-ice removal of till matrix by ice, leaving a boulder lag, or the result of meltwater activity at the ice/bed interface, which removed the matrix to leave a lag of coarser material. Crudely stratified and sorted sand and gravel lenses and channels within the pavement point to a surplus of meltwater at the ice/bed interface, which would support this latter hypothesis. The boulders were lodged into the deforming substrate as ice began to flow north–northeast to south–southwest, causing planation and striation before eventual burial by further till accretion (LFA 2). The strong unimodal orientation of the clast long axes supports this. This model suggests abundant meltwater at the ice/bed interface, perhaps during the quiescence of the first flow phase, and before the second phase of ice flowing northeast to southwest, which planated and striated the boulder pavement and deposited the upper diamict.

Figure 12.

 Cartoon depicting formation of the boulder pavement and hydrofracture at Whitburn Bay. Ice overburden pressure is denoted by ‘P’. A. First ice lobe advances and deposits LFA 1. Ice lobe quiesces. Boulders melt out of ice. Second ice lobe advances, lodges boulders and imposes striations on them. Boulder pavement represents an erosional surface. LFA 2 is deposited over the top of the boulders. High amounts of meltwater are inefficiently evacuated as porewater flow under high pressure. B. Hydrofracture is initiated. Bedrock fractures, and groundwater is piped under pressure towards ice-bed interface. Channels form and the water flows away under the ice-sheet. C. Once the main force of water is dissipated, course gravels and bedded sands are deposited. They are subsequently glaciotectonised. A later-phase nye channel forms above.

Lithofacies Association 2: A subglacial traction till with hydro-fractures, pipes and nested Nye channels

The subglacial traction till. – Faceted, striated, far-travelled clasts, and the over-consolidated nature of the diamicton, indicate that LFA 2 is also a subglacial traction till (Evans et al. 2006). The variable localized incorporation of LFA 1 indicates that extensive deformation has homogenized and mixed the two tills. PCA of the heavy mineral content shows strong inter-sample heterogeneity, but no dichotomy between the diamictons (Fig. 13). The failure of this method to discriminate between the lithofacies could be a result of the small sample numbers, as this can create an artificially high skew.

Figure 13.

 Principle components analysis (PCA) and ternary diagram of heavy mineral analysis, clast lithological analysis and metals analysis. The ternary diagram shows Lithofacies Association 1 (squares) and Lithofacies Association 2 (circles) stone counts. Lithofacies Association 3b is denoted by triangles. The first component of the metals PCA analysis explains 47% of the data, and is explained mostly by the proportions of lithium, boron, aluminium, calcium, titanium, vanadium, beryllium, cobalt, nickel, iron, chromium and copper. Component 2 explains 19% of the data, and is explained by potassium, manganese, arsenic, molybdenum, silver, antimony and barium. For the heavy mineral PCA correlation, sulphides were excluded from the analysis due to low numbers; carbonates were excluded due to strong skew. The strong first component, with 52% of the variation, is explained by the proportion of the silicate group, pyroxene and micas in the sample. The second component, with 19% of the variance, is explained by oxides and then by pyroxene. In this analysis, samples WH07 (LFA 1) and WH13 (LFA 3b) do not stand alone, but correlate with the other samples.

Clast fabrics and striations are poorly clustered, but indicate general ice movement from the north–northeast. The low and variable S1 values from the clast fabrics could suggest extensive deformation (cf. Evans et al. 2006), possibly driven by fluctuating subglacial conditions and water pressure. Hart (2007) argues that weak clast fabric strengths reflect clast interaction and a dominance of deformation and rotation, and furthermore that large grain size and low sorting emphasize rotation, leading to low clast fabric strength in subglacial diamictons. Others argue that soft, water-saturated sediments such as LFA 2 behave as Coulomb plastic materials (Larsen & Piotrowski 2003). Clasts become aligned parallel to the shear direction when exposed to high strain rates, giving a high fabric strength. Weak fabrics are therefore associated with low shear strains (Hooyer & Iverson 2000; Larsen & Piotrowski 2003). Larsen & Piotrowski argue that strong clast fabrics only develop under conditions of strain homogeneity and uniform local conditions. The weak and variable clast fabrics at Whitburn Bay reflect the heterogeneity in strain, water content and till rheology.

In Section 2a, the overturned pipe structures are evidence of the squeezing of the soft, saturated LFA 1 into LFA 2 under a high ice overburden pressure (Fig. 14). They are overturned towards the southeast, supporting this direction of ice flow. These features have a low preservation potential and are unlikely to have survived the process of lodgement (Nelson et al. 2005), suggesting that they post-date the deposition of LFA 2. In all, the highly variable and often weak clast fabric strengths, as well as the preservation of the pipe structures and canal fills, suggest a low strain deformation signature (cf. Evans et al. 2006) within LFA 2. They may have formed contemporaneously with till accretion of LFA 2. As LFA 1 became buried by an increasingly thick layer of sediment, sealing the lower hydraulic pathways and increasing overburden pressure, and the deforming layer moved upwards (the ‘A Horizon’), the lower non-dilated ‘B Horizon’ would have undergone little deformation, despite ongoing ice movement and till deformation in the ‘A Horizon’ above (Boulton & Hindmarsh 1987; Evans et al. 2006). The pipers would have formed in saturated, confined conditions but could have experienced little deformation.

Figure 14.

 Land-system development at Whitburn Bay. A. First ice lobe advances and deposits LFA 1. Ice lobe stagnates prior to retreat and boulders melt out. As the second ice lobe advances, it striates and lodges the boulder pavement as an erosional surface. B. The deposition of LFA 2 buries the boulder pavement. A well-developed discrete ice-marginal, subglacial drainage system drains the glacier subaerially. Pipe structures and hydro-fracture are formed. The sediments are extensively tectonized. C. The discrete drainage system may evolve seasonally into a distributed, braided, subglacial canal system. Low flow periods, possibly seasonal or diurnal, result in quiescence in the channels. As the ice lobes retreat, Glacial Lake Wear was formed in between them. The depth of the lake varied between several highstands (C1 and C2). The lake may have periodically back-filled the canals, resulting in quiescence and ponding.

Micromorphological analysis of LFA 2 reveals numerous micro-scale deformational structures, such as circular structures with associated skelsepic fabrics, associated with aligned grains and grain lineations (Table 3). Elongate grains near the shear plane have rotated until they are aligned plane-parallel. Crushed grains and strong masepic plasmic fabrics indicate pervasive strain and high pressure. These features are indicative of subglacial deposition and deformation (van der Meer 1997; Carr 2001). Thin section 10 shows in detail the junction of the LFA 2 diamicton with the underlying bedrock, and further supports a subglacial till interpretation, based on the strong presence of circular structures, crushed grains, Type II pebbles and a strong plasmic fabric (Fig. 7). The bedrock has been entrained into the till with evidence of stringer formation and cannibalization of the lower soft, dolomitic limestone.

Hydro-fracture. – The subvertical clastic dykes in Section 5 (Fig. 9) are the result of the escape of high-pressure groundwater beneath the ice sheet, i.e. as hydro-fracture fills (Evans et al. 2006). Tensional cracks are infilled by sediment fluidized by the escaping water (Fig. 12). Hydro-fractures occur where fluid pressure exceeds the tensile strength of the sediment and the smallest component of the ice overburden pressure (Rijsdijk et al. 1999). Juxtaposed thinly bedded, calcreted sands and coarse, well-sorted gravels indicate variable flow regimes.

At Whitburn Bay, subglacial meltwater would have discharged into the Magnesian Limestone aquifer. The overlying tills acted as aquicludes, hydraulically confining the bedrock (as also observed by Rijsdijk et al. 1999 at Killiney Bay, Ireland). When the supply of meltwater exceeded the capacity of the bedrock aquifer, water pressures rose beneath the overlying till. When the water pressure exceeded the tensile strength of the till, it caused tensile fracturing, which enabled hydro-fracturing (Fig. 12). The discharge of water was sufficient to fluidize the sands and gravels, transport them through the hydro-fracture and inject them into the overlying sediments. A high percentage of calcium carbonate in the ground water, derived from the bedrock, resulted in calcretion of the gravels. Hydro-fractures and high porewater pressures may be related to the coarse-grained channels operating at the ice/bed interface. These discrete, hydraulically efficient channels are highly erosive and incise into the deforming layer at the ice/bed interface, suggesting effective evacuation of subglacial meltwater. They were later buried by continued till accretion. Hydro-fracturing and channel formation was therefore associated with increasing basal water pressures during the accretion of LFA 2 from ice overriding from the north.

Nested Nye channels. – The isolated channelized forms, with concave-up bases and flat tops, and the strong variation in height of these channels, preclude a proglacial origin. Distal proglacial sandur systems are characterized by trough cross-bedded, cyclic fining-upward sequences of gravels, sands and silts with slip face migration of longitudinal and linguoid braid bars. This gives rise to planar cross-beds, with abundant ripple-drift and cross-lamination (Maizels 1995). Distal proglacial outwash sediments typically exhibit mega-ripple migration on point-bar surfaces, producing large-scale trough cross-beds. The lack of bars, dunes and tabular bedding in the Whitburn channel systems suggests that they are subglacial glaciofluvial sediments (Collinson 1996). They are a low-energy subglacial braided canal network, as defined by Walden & Fowler (1994), which was active beneath the ice sheet that formed the traction tills. The channels formed at the ice/bed interface and were later buried by till accretion. This interpretation is in agreement with previous process-based research further north in Northumberland (Eyles et al. 1982).

This type of distributed subglacial drainage system is inefficient, and can result in low energy flows and ponding beneath the ice sheet. These systems operate when pore-water flow cannot evacuate all the excess water in the system (Benn & Evans 1996). Such systems have long been recognized in the geological record as broad lenses of sorted sediment within the tills with concave-up lower contacts and nearly planar upper contacts (e.g. Dreimanis et al. 1986; Shaw 1987; Lunkka 1994). Englacial conduits rarely extend further than 200 m into the ice sheet (Fountain & Walder 1998), indicating that the sequence at Whitburn Bay formed in a submarginal environment. Ductile deformation, despite the high drainage capacity of the bed, would therefore have occurred close to the ice margin under low cryostatic pressures.

Swift et al. (2002) argued that seasonal reorganization of subglacial drainage can occur beneath many temperate and polythermal glaciers, resulting in distributed and channelized configurations, with the development of a hydraulically efficient, channelized subglacial drainage system during the ablation season. This would involve higher-pressure, channelized, faster-flowing canals (such as Section 7) and slower-moving, lower-energy, distributed drainage systems (such as Section 9) evolving (perhaps seasonally) throughout the year. These lower-energy channels also demonstrate flow variability at smaller scales, with evidence of periodic quiescence and ponding, as well as periods of faster flow, demonstrated by the juxtaposition of Type A and B ripples, planar lamination and clay drapes, which represent alterations of fast flow and suspension settling under low-flow or no-flow conditions. This indicates periodic quiescence of the channel with little sediment input (cf. Ashley et al. 1982, 1985). Repeated changes between planar laminated sands and Type B climbing ripples in Section 9d (Allen 1963) represent fluctuating flow, indicating slowly migrating ripples with high vertical aggradation rates (cf. Ashley 1995), with ripple drapes indicating water ponding. Interlaminated sand and clay units indicate repeated quiescence and fast flow in the channels. A pebble lag near the basal contact in Section 9d points to traction current activity with bedload saltation.

Lithofacies Association 2 therefore demonstrates fluctuations of meltwater discharge at multiple scales and perhaps seasonally. An alternative explanation for the laminated clay and sand facies at Whitburn Bay could be backfilling of subglacial canals during times of low or no water flow related to ice-contact lake level fluctuations. There is extensive evidence of such proglacial lakes in County Durham and Yorkshire (Smith 1994). During periods of low lake levels, high submarginal discharge in these channels would have resulted in the deposition of rippled and bedded sands, but during periods of high ice marginal lake levels, backfilling of the channels could have caused ponding and clay drape lamination. The canal fills could thus be related to the activity of local lakes such as Glacial Lake Wear. Indeed, episodically changing lake levels, related to both seasonal variations in meltwater and to the movement of ice lobes, have been suggested in Glacial Lake Wear (Smith 1994). The geometry of these sand infills is therefore significantly different to the tabular, widespread, subaerial glaciofluvial sands and gravels presented by Francis (1972).

Within the channels, the laminated sediments are locally strongly deformed; their load and water escape structures indicate deformation and remobilization of the water-saturated beds, suggesting rapid deposition (Glasser et al. 2001) or a high overburden pressure. The consistent overturning of folds to the south also suggests glaciotectonic disturbance as ice flowed southwards. This is supported by the micromorphological analysis, which shows compression of the primary bedding structures and masepic plasmic fabric development subparallel to bedding, followed by conjugate fault development as folds have been compressed, overturned and extended (Fig. 11). Compression has elevated pore-water pressure within the sand beds, which has been released during faulting, causing fluidization of sands and water escape and injection structures.

Ultimately, LFA 1 and LFA 2 represent a mosaic of processes operating subglacially at the time of sediment deposition at both the macro- and the micro-scale (cf. Piotrowski et al. 2004). The hydro-fracture and infilled canals clearly show that basal water pressures fluctuated at Whitburn Bay during deposition of LFAs 1 and 2, as will have pore-water pressures, which have been influential in the degree of till deformation (cf. Boulton et al. 2001). Multiple switches in different modes of deposition and deformation resulted in the variable appearance of the two tills and the boulder pavement at Whitburn Bay (cf. Piotrowski et al. 2004, 2006).

Provenance of the Whitburn Bay tills

Principal components analysis (PCA)

A comparison of LFA 1 and LFA 2 (Table 2) shows that locally sourced, non-durable lithologies dominate LFA 1, whereas LFA 2 has a larger component of far-travelled and igneous erratics (Fig. 13). Most of the variation could be accounted for by variation between the crystalline, Permian and Carboniferous groups, which have strong internal correlation indices. On a ternary plot of these three variables (Fig. 13), the two diamicton lithofacies are differentiated, with sample WH14, from Section 10, in between. PCA on both the correlation and the covariance matrices replicates this result. The low percentages of crystalline rocks and high percentages of Permian rocks are the main factors differentiating LFA 1 from LFA 2. A correlation PCA (Fig. 13) on the total metals suite fails to distinguish the three lithofacies and indicates strong within-till heterogeneity. Sample WH13 from the sands and sample WH07 from LFA 1 show consistent differences to the other samples.

Heavy mineral sources

Both LFA 1 and LFA 2 are rich in clinozoisite, which is common in schists and is a product of low- to medium-grade metamorphism (Mange & Maurer 1992). Both tills are rich in olivine and pyroxenes (Table 5), which could be sourced from ultramafic to mafic igneous sources. The Permian Whin Sill Dolerite, which outcrops extensively to the north and northeast of Whitburn (Smith & Francis 1967), may have been a primary source of detrital pyroxene. However, this micro-gabbroic intrusion does not contain olivine, indicating that a separate basic igneous source was also supplying detritus. One such potential source of both olivine and clinopyroxene is the Carboniferous volcanic rocks (olivine and clinopyroxene phyric basalts) and high level intrusions of Northumbria, the Midland Valley of Scotland and locally within the Southern Uplands. As olivine is not very robust, it is likely to have been sourced from local areas such as Northumbria.

The metamorphic assemblage of heavy minerals within both LFAs 1 and 2 is consistent with a source terrane that includes a significant proportion of upper greenschist to upper amphibolite facies of regionally metamorphosed pelitic mudstones. The two key mineral assemblages are, first, garnet, staurolite and chloritoid, and, second, garnet, andalusite and kyanite. Each assemblage is found only in specific areas of polydeformed and metamorphosed orogenic belts. In particular, the Highlands and Islands of Scotland are the sole source for chloritoid (Mange et al. 2005). The garnet-staurolite-chloritoid assemblage is characteristic of Stonehavian-type metamorphism developed in a small area to the east of Stonehaven close to the Highland Boundary Fault, implying a possible source in northeast Scotland (Stephenson & Gould 1995; Trewin 2002). The garnet–andalusite–kyanite assemblage (possibly with sillimanite) is a higher grade assemblage developed in the Buchan-type metamorphism of northeast Scotland (Stephenson & Gould 1995).

Younger sedimentary rocks may be a source of reworked minerals. Tourmaline in particular is very resistant and is common in many sedimentary rock types. Further research should include identification of the heavy mineral suites in sedimentary rocks in the Southern Uplands and Midland Valley as well as in northern England. This does not explain the presence of non-durable mafic minerals, which would not have survived reworking.

Overall till provenance

The dominant lithologies in LFA 1 are principally locally sourced Magnesian Limestone, sandstones, Carboniferous Limestone, Whin Sill Dolerite and Old Red Sandstone. The Carboniferous Coal Measures immediately to the west and north of eastern Durham are sources for the coal and sandstones of LFA 1 (Fig. 2). Whin Sill Dolerite is a local Permian intrusive igneous rock with abundant outcrops to the north and west of the area. LFA 1 is largely a locally derived till. Old Red Sandstone outcrops to the northwest.

The heavy mineral assemblage suggests that ice entrained material sourced from northwestern England, the Southern Uplands, the Midland Valley and possibly further north from northeastern Scotland. Clast fabrics and striae orientations support a first phase of ice flow from west to east, which crossed the Pennines through the Tyne Gap and deposited the lower traction till. If the northwest to southeast ice-flow interpretation is correct, East Grampian minerals within LFA 1 are readily explained by the reworking of older northerly derived glacigenic sediments situated on the Durham coastal lowlands.

The palynomorph assemblages within LFA 2 indicate a likely derivation from the Newcastle coalfield to the north. A lack of Magnesian Limestone in LFA 2 compared with LFA 1 (Fig. 13) indicates that LFA 2 has been isolated from the local Permian bedrock by a mantle of earlier till (LFA 1), which is widespread in the Durham region (Beaumont 1967). The distinctive pink granites, rhyolites and porphyries within LFA 2 are from the Cheviots; its greywackes may come from the Southern Uplands, while the Grampian Highlands are sources of schist and slate.

The combination of dioritic/granitic lithics with the metamorphic mineral assemblage of LFA 2 suggests that material was sourced from the Dalradian Supergroup, exposed in northeastern Scotland. Sources of Old Red Sandstone and Carboniferous sandstone and limestones extend eastwards into the North Sea, where they terminate against the North Sea central graben. Any ice feeding down the coast from the Scottish Highlands would have had to cross this area, and may also have coalesced with ice from the Midland Valley. The lower percentages of olivine and pyroxenes within LFA 2 suggest that the Midland Valley had less influence on this till. These minerals could also have been derived from reworking of the lower LFA 1.

Collectively, this evidence indicates a northerly source as far north as the East Grampian Highlands before ice coalesced with the Tweed ice stream and entrained Cheviot erratics. A clast fabric and striae on the boulder pavement indicating a north–northwesterly flow direction supports this interpretation.

Discussion

Evolution of the glacigenic sequence at Whitburn Bay

Recent research has highlighted the fact that there is large spatial variability in basal friction below ice sheets, and that glacier beds are mosaics of sliding, deformation, lodgement and ploughing (Piotrowski et al. 2004; Nelson et al. 2005; Evans et al. 2006). Whitburn Bay demonstrates this well, with evidence of lodgement in the form of a boulder pavement adjacent to extensive evidence of deformation and hydro-fracturing.

Glacigenic sediments at Whitburn Bay can be understood through a multiphase model of development (Fig. 14). The first phase involved the arrival of an ice lobe from the west, which deposited LFA 1 (Fig. 15). This ice originated from Scotland and entered eastern England via the Tyne Gap, some time after 21 475–22 140 cal. yr BP based on correlation with the Dimlington type site (Cameron et al. 1992; Bateman et al. 2008). It was later superseded by ice flowing northeast to southwest, which lodged and planated the boulder pavement and deposited LFA 2 (the Horden Till in County Durham and the Bolders Bank Formation in the offshore area (Catt 2007)). The simplest interpretation of a secondary northeast to southwest ice flow is deflection by Fennoscandian ice in the North Sea. This, however, conflicts with recent research, which indicates decoupling by the Fennoscandian and British ice sheets by the Late Devensian (Carr et al. 2006).

Figure 15.

 Map showing inferred ice-flow directions, overlain onto the BRITICE data set for the northeast region (Clark et al. 2004). Ice flow around the Tweed area from Raistrick (1931).

East coast surging

The initiation of the boulder pavement may have been triggered by a period of basal ice melting as flow from the west waned, resulting in aqueous washing of matrix material. The boulders were later lodged and abraded by the ice flowing from the north that deposited LFA 2 (Fig. 15). A channelized subglacial drainage system of multiple braided canals developed beneath the ice, which may have changed, possibly seasonally, to a discrete, higher-pressure drainage system of more efficient Nye channels. The sedimentary signal within the canals was controlled by a fluctuating internal drainage system; alternating periods of current flow and ponding suggest an inefficient hydraulic regime, which is often typical of surging glaciers (Björnsson 1998; Evans et al. 2006). It is possible that fluctuating local lake levels also influenced the hydraulic efficiency in these submarginal canal environments, although this is difficult to substantiate.

Björnsson (1998) argued that glacial surges and fast ice flow are associated with subglacially distributed drainage systems, because an increase in pore-water pressure can lead to decoupling of the ice from its bed (Björnsson 1998; Evans et al. 2006) and cause enhanced sliding (Ng 2000; Boulton et al. 2001).

Modelling by Boulton & Hagdorn (2006) has shown that a powerful ice stream flowed down the eastern coast of Britain, its advance to the Wash embayment triggering the initiation of Glacial Lake Humber. Recent Optically Stimulated Luminescence dates on the highstand of this lake (Bateman et al. 2008) of 16.6±1.2 kyr suggest that the BIIS was flowing at the western edge of the North Sea Basin possibly during Heinrich Event 1, depositing the Skipsea Till and the Bolders Bank Formation. Eyles et al. (1994) argue that this east coast ice stream experienced recurrent onshore surging against the rising bedrock surface of Holderness and Lincolnshire. The deforming bed, mosaic deformation and subglacial canal evidence from Whitburn could therefore support the existence of such a surge lobe, which is supported by external evidence from modelling and the dates of 16.6 kyr from Lake Humber.

Implications for Quaternary stratigraphy of eastern England

The locally derived Blackhall Till (LFA 1) at Whitburn Bay was previously correlated with the Skipsea Till of Holderness (Francis 1972) due to its stratigraphical position, but this is disputed here because of the provenance data. We favour an origin for ice flowing from the Lake District and across the Pennines through the Tyne Gap. The Blackhall Till shows a distinct west-to-east movement and comprises mainly local clasts. Previous work has suggested that it is limited in extent, and there seems to be no equivalent further south than County Durham (Beaumont 1967). Therefore, the Blackhall Till does not appear to correlate with the Skipsea Till. Instead, the mineralogy, particle size, erratic content and characteristics of the Skipsea Till indicate a correlation with the Horden Till (Table 6).

Table 6.   Comparison of characteristics of Skipsea Till, Dimlington and Horden Till, Whitburn Bay (Penny & Catt 1967; Madgett & Catt 1978; Evans et al. 1995; Catt 2007).
 Skipsea TillBlackhall Till (LFA 1)Horden Till (LFA 2)
Type siteDimlington, YorkshireBlackhall's RocksBlackhall's Rocks
ColourVery dark greyish brown. 10YR 3/2Dark yellowish-brown. 10YR 4/4Dark brown. 10YR 3/3
Sedimentology5–9 m thick. Interbedded diamictons, discontinuous bodies of stratified sedimentsMassive, matrix-supported diamicton. Common cobbles of local origin. Boulder pavement at topMatrix-supported diamicton. Nested channels within diamicton
Particle size22–38% clay, 32–42% silt, 22–42% sand31% clay, 39% silt, 18% sand, 12% gravel33% clay, 38% silt, 17% sand, 11% gravel
Erratic contentChalk, shale, greywacke, Cheviot porphyries, granites, Whin Sill Dolerite, Carboniferous and Magnesian Limestone, coal, rhomb porphyryMagnesian Limestone, Carboniferous Limestone, Carboniferous sandstones, greywacke, Old Red SandstoneCheviot rhyolite and andesite, granite, Carboniferous Limestone, Magnesian Limestone, Old Red Sandstone, schist, greywacke
Heavy mineral contentEnriched in amphibole and epidote, poor in chlorite and biotiteEnriched in clinozoisite, micas, kyanite, amphibolesSignificant amounts of pyroxene, kyanite, epidote, limited chlorite and biotite
PalynologyNone availableNone availableCarboniferous spores: Westphalian and Namurian
Macro-fabricNNE–SSW direction; WNW–ESE fold axesModerately strong; NW to SEWeak macro-fabric, NE to SW
AgeOverlies Dimlington silts:>21 kyr BPNone availableNone available

The macro-fabrics of the Horden and Skipsea tills indicate deposition by ice moving inland from the North Sea basin. Ice is suggested to have originated from the Southern Uplands, streamed down the eastern coast of Britain (Eyles et al. 1994), invaded the North Sea Basin and deposited the Skipsea Till and the Bolders Bank Till (Cameron et al. 1992; Carr et al. 2006). The movement inland could have resulted from coalescence of British ice with Scandinavian ice offshore in the North Sea. Sejrup et al. (2000, 2005) and Carr et al. (2006) argue that the British and Fennoscandian ice sheets had decoupled by this stage, but the southerly extension of the North Sea Lobe during the latter phases of the LGM cannot have occurred without the continued presence of Scandinavian ice offshore. The presence of heavy minerals derived from the eastern Grampian Highlands indicates a distant northerly source. The presence of these minerals indicates that the ice has been deflected strongly southwards, suggesting that contact with Scandinavian ice offshore is needed to direct the ice flow southwards. Shap erratics within the Skipsea Till are derived from coalescence of the Durham and Tees ice lobes in the Tees Gap. Hence, this work suggests that the southward-flowing North Sea Lobe was the second ice lobe to transgress the Durham area, but the first ice body to reach the Holderness coast during the LGM.

Conclusions

Research at Whitburn Bay supports the existence of a complex, multi-lobate, late Devensian BIIS along the east coast of Britain. The traction tills LFA 1 and 2 represent ice flow from two different directions within the same glaciation. The Blackhall Till (LFA 1) originated in northwestern England, possibly sourced from the Midland Valley and western Southern Uplands. It may also have a component of ice from northeastern Scotland. It flowed south, before passing through the Tyne Gap. There is little evidence of a Lake District ice source. The Horden Till (LFA 2) was deposited by an ice lobe flowing down the eastern coast of Britain that may have originated as far north as the eastern Grampian Highlands, and which is dominated by Cheviot and Northumbrian erratics. The boulder pavement was deposited as a lag through a combination of subglacial erosion and aqueous winnowing as ice flow from the west waned and was later lodged and abraded by ice moving southwards.

The braided canal system preserved in the Horden Till suggests a fluctuating low-flow subglacial drainage system, juxtaposed with high-energy gravel channels, reflecting periodic changes in the subglacial drainage hydraulic regime. Hydro-fracturing supports the notion of extreme variations in subglacial drainage, which may have triggered the development of high-energy channels at the ice/bed interface. The episodic changes from rapid water flow to quiescence in these submarginal channels indicate periodic ponding events. Seasonal re-organizations of the subglacial drainage system could have resulted in the juxtaposition of sand and very high-energy gravel channels. It is also possible that fluctuating lake levels in the proglacial Glacial Lake Wear may have contributed to this variation, as high lake levels would have resulted in backfilling of the channels, quiescence and the formation of draped lamination. Evidence of basal decoupling, extreme fluctuations in subglacial pore-water pressure and the presence of a deforming bed supports the notion of a fast-flowing ice lobe in the vicinity of Whitburn Bay and further south, such as at Holderness (Eyles et al. 1994; Evans et al. 1995; Boulton & Hagdorn 2006). The mineralogy, lithologies and process interpretation of the Horden Till demonstrate that it is correlative with the Dimlington Stadial Skipsea Till and Bolders Bank Tills in Yorkshire and offshore, respectively. This may suggest that the North Sea ice lobe flowing south was the second ice lobe to transgress the Durham area, but the first ice mass to cross the Holderness coast during the LGM.

Acknowledgments

Acknowledgements. – This research was funded by the Department of Geography and Hatfield College at Durham University. B.J.D. thanks D.H.R., D.R.B. and C.O.C. for encouragement, guidance, support and help in many ways. We thank Dr. Ian Evans for advice on statistical analysis, Frank Davies and the laboratory staff at Durham University for help with ICP-MS and other laboratory work, and David Sales of Durham University Earth Sciences Department for thin section preparation. The staff at the Design and Imaging Unit at Durham University are thanked for help with the graphics. We gratefully acknowledge John Catt, Simon Carr and Jon Lee for constructive, informative reviews, which did much to clarify the content of the article. Emrys R. Phillips and James B. Riding publish with the approval of the Executive Director, British Geological Survey (NERC). The article was guest-edited by Jasper Knight, University of Exeter.

Glossary of micromorphological terms

  • Anisotropic: The anisotropic skeleton grains and plasmic matrix of the slide transmit plane-polarized light, but under cross-polarized light they extinguish (i.e. transmit no light) four times per complete rotation. Isotropic minerals remain black in all positions when viewed under cross-polarized light.
  • Birefringence: If the clay particles that form the plasma are strongly orientated by stress, they transmit cross-polarized light and extinguish partly or completely on rotation of the stage.
  • Domain: A localized zone displaying a characteristic plasmic fabric.
  • Necking structures: Related to galaxy structures. Two large juxtaposed skeleton grains are surrounded by a halo of aligned smaller skeleton grains. Where the two larger skeleton grains are closest, the smaller grains have flowed between them as a result of ductile deformation.
  • Pebble type I: Arrangement of brecciated sediment such that it appears to form a series of rounded intraclasts delineated by packing voids.
  • Pebble type II: Soft sediment intraclasts of material similar in nature to the surrounding material, but with a clearly defined discrete internal plasmic fabric.
  • Pebble type III: Soft sediment intraclasts of material different in nature from the surrounding sample, indicating reworking of pre-existing sediments.
  • Plasmic fabric: The arrangement of clay particles in a sample.
  • Lattisepic: Preferred orientation of plasmic fabric in two perpendicular directions: commonly associated with skelsepic plasmic fabric.
  • Masepic: Preferred orientation of plasmic fabric in diffuse domains of parallel orientation: indicative of pervasive shearing.
  • Skelsepic: Preferred orientation of plasmic fabric around the surfaces of larger grains: indicates rolling of larger grains.
  • Unistrial: Preferred orientation of plasmic fabric in discrete parallel domains; indicative of discrete shear.
  •   Pressure shadows: Symmetric or asymmetric tails of material on the stoss and lee sides of large grains; indicative of planar shearing (symmetric) or rotation (asymmetric).
  •   Rotational structures: Circular alignments of grains around cores of consolidated sediment or larger grains; indicative of rotation. May also occur without a core grain.
  •   Skeleton grain: Single sand or coarse silt grains which are larger than the thickness of a thin section (20 to 30 μm).

Ancillary