glacial sedimentary processes and products (hambrey/glacial sedimentary processes and products) ||...

Anatomy and facies association of a drumlin in Co. Down,Northern Ireland, from seismic and electrical resistivity surveys

BERND KULESSA*1, GORDON CLARKE†, DAVID A. B. HUGHES† and S. LEE BARBOUR‡

*School of the Environment and Society, University of Wales Swansea, Singleton Park, Swansea, SA2 8PP, UK †School of Planning, Architecture, and Civil Engineering, Queen’s University Belfast, Belfast, BT9 5AG, UK

(e-mail: [email protected])‡Department of Civil and Geological Engineering, University of Saskatchewan, Saskatoon, Saskatchewan, SK S7N 5A9, Canada

ABSTRACT

Seismic refraction and electrical resistivity geophysical techniques were used to reconstruct theinternal architecture of a drumlin in Co. Down, Northern Ireland. Geophysical results were bothvalidated and complemented by borehole drilling, flow modelling of ground water, and geologicalmapping. The geophysical anatomy of the drumlin consists of five successive layers with depth includ-ing; topsoil, partially saturated and saturated glacial tills, and weathered and more competent greywackebedrock. There are numerous, often extensive inclusions of clay, sand, gravel, cobbles, and boulderswithin the topsoil and the till units. Together geophysical and geotechnical findings imply that thedrumlin is part of the subglacial lodgement, melt-out, debris flow, sheet flow facies described by pre-vious authors, and formed by re-sedimentation and streamlining of pre-existing sediments duringdeglaciation of the Late Devensian ice sheet. Seismic refraction imaging is particularly well suitedto delineating layering within the drumlin, and is able to reconstruct depths to interfaces to within±0.5 m accuracy. Refraction imaging ascertained that the weathered bedrock layer is continuous andof substantial thickness, so that it acts as a basal aquifer which underdrains the bulk of the drumlin.Electrical resistivity imaging was found to be capable of delineating relative spatial changes in themoisture content of the till units, as well as mapping sedimentary inclusions within the till. The mois-ture content appeared to be elevated near the margins of the drumlin, which may infer a weakeningof the drumlin slopes. Our findings advocate the use of seismic refraction and electrical resistivitymethods in future sedimentological and geotechnical studies of internal drumlin architecture anddrumlin formation, owing particularly to the superior, 3-D spatial coverage of these methods.

Keywords Drumlin, stratigraphy, formation, geophysics, seismic refraction, electricalresistivity.

INTRODUCTION

During the Late Devensian glaciation in North-ern Ireland, ice flowed offshore through coastalembayments away from the ice divide in theLough Neagh region (Fig. 1), terminating at marinemargins (e.g. McCabe, 1987). Following the LastGlacial Maximum, drumlinisation in NorthernIreland occurred on strong ice-moulded bedrock(e.g. Eyles & McCabe, 1989) during the BeldergStadial (~18–16.6 14C kyr BP) (e.g. Knight, 2002).Drumlins formed by mass and debris flow together

with lodgement processes in a hydrologicallyactive subglacial environment beneath warm ice,and can be classified into five facies associationswhich include (e.g. McCabe & Dardis, 1989):

Facies Association 1: ‘Forms with a core of older drift’(predating 30 ka BP); occur near major divides of icedispersion in the Late Pleistocene;Facies Association 2: ‘Overridden ice-marginal sub-aqueous facies’; occurs along the central and northerncoast of western Ireland and is characterised by inter-bedded diamictons, muds, sands, and gravels;

GSP_4_C11.qxd 11/5/07 13:33 65

Glacial Sedimentary Processes and Products Edited by Michael J. Hambrey, Poul Christoffersen, Neil F. Glasser and Bryn Hubbard © 2007 International Association of Sedimentologists. ISBN: 978-1-405-18300-0

166 B. Kulessa et al.

Facies Association 3: ‘Subglacial lodgement, melt-out,debris flow, sheet flow facies’; particularly commonin Ireland and characterised by drumlins composedof till and commonly inter-till sands beds and/orstratified diamictons;Facies Association 4: ‘Subglacial channel stratifiedfacies’; occur in subglacial tunnel-type valleys and aregenerally cored by sequences of sands and gravels;Facies Association 5: ‘Lee-side stratified facies’; occursthroughout the Irish drumlin belt and is ‘. . . charac-terised by proximal to distal sediment transforma-tions from massive to stratified diamicts which areinterbedded with a wide range of sand and gravellithofacies . . .’ (McCabe & Dardis, 1989).

The internal architecture of drumlins, as repre-sented by these facies, is of interest because it is a common record of past glaciological conditionsand processes operating at the base of ice masses.From an applied engineering perspective, know-ledge of drumlin architecture is pre-requisite to thesuccessful design of geotechnical cuttings, e.g. forroad construction. Unfortunately, reconstructionof drumlin stratigraphy often suffers from poor spatial coverage because suitably large exposuresare commonly not available (e.g. Menzies, 1987),and borehole logging is limited to one spatialdimension. Geophysical investigations of internaldrumlin architecture are very sparsely documented.Nonetheless, existing work is encouraging, sug-gesting that particularly seismic and electricalresistivity surveys are well suited for identifica-tion of tills (e.g. Birch, 1989) and reconstruction ofinternal layering (e.g. Sutinen, 1985; Sharpe et al.,2004) in drumlins and other glacial deposits.

In 2004–05 a substantial cutting was excavatedthrough a large drumlin in Co. Down, Northern

Fig. 1 (A) Location of field site, marked by ‘X’, inNorthern Ireland (NI). The open and full circles in NIrespectively mark the locations of Lough Neagh andBelfast City. (B) Photograph of drumlin, view is to thenorth. The area of proposed excavation prior to roadconstruction is marked by stripped topsoil, as indicatedby lighter grey colours on the eastern side of thephotograph than on the western side. (C) Schematiclayout of seismic refraction (SWE, SSN) and electricalresistivity (RWE, RSN) profiles, together with borehole (BH)locations. The centre of the drumlin coincidesapproximately with the location of BH1, and the stossand lee sides of the drumlin are labelled.

BH1

BH2

BH3

BH4

N

Stoss

Lee100 m

RWE

SWE

SSN

RSN

NI

A

C

X

Dublin

100 km

N

B

GSP_4_C11.qxd 11/5/07 13:33 Page 166

Anatomy and facies association of a drumlin in Co. Down, Northern Ireland, from seismic and electrical resistivity surveys 167

Ireland (Fig. 1). Extensive seismic refraction and electrical resistivity geophysical data were collectedprior to excavation in order to reconstruct theinternal architecture of the drumlin. These data arereported here, together with direct field observa-tions and geotechnical borehole logs recordedduring subsequent drilling. These studies aimed toanswer four particular questions:

1 What value can geophysical surveying add tomore traditional methods of geological field mappingand geotechnical site investigation?2 How accurate are geophysical reconstructions ofinternal drumlin architecture, as compared with bore-hole logs and direct field observation?3 How well suited are seismic and electrical resistivitysurveys for reconstructing internal layering within thedrumlin and the physical properties of the sedimentsit is composed of?4 What are the implications of the geophysical andgeotechnical findings for existing models of drumlinformation?

It is hoped that the present study will increaseinterest in the use of geophysical methods in theinvestigation of glacial sediments.

FIELD SITE AND METHODS

Field site

The field site is a large drumlin located approx-imately 1 km south of Loughbrickland, Co. Down (Fig. 1A). The drumlin covers an area of approx-imately 0.75 km2 and is part of a swarm alignedapproximately north-south, consistent with iceflow away from the Late Devensian ice divide at Lough Neagh towards the Carlingford Loughcoastal embayment (e.g. McCabe, 1987; Eyles &McCabe, 1989; Knight, 2002). The present study was prompted by plans to construct a dual carri-ageway along the main road between Belfast andDublin (A1) in the Loughbrickland area. Prior toborehole drilling and excavation as part of the pro-posed cutting (Fig. 1B), several seismic refractionand electrical resistivity geophysical profiles wereacquired to reconstruct the spatial distribution of vadose zone thickness and depth to bedrockacross the drumlin (O’Loughlin, 2003; Sexton, 2003).

The locations of these profiles were chosen so thatthey coincided with the area of the proposed cut-ting and could be used to assist with the siting offour boreholes drilled subsequent to the geophys-ical surveys. The principles of both geophysicalmethods are well established and field equipmentand interpretation software is widely available.Only a brief introduction to either technique istherefore given below, and the interested reader is referred to popular textbooks for further details(e.g. Reynolds, 1997; Sharma, 1997).

Seismic refraction surveys

The seismic waves were generated using a sledge-hammer impacting on a metal plate coupled to theground surface. This is standard practice in near-surface applications and indeed compares favour-ably with alternative seismic sources (e.g. van derVeen et al., 2000). Seismic waves were expected totravel through the body of the drumlin, refractingat major interfaces, such as the phreatic surfacewithin the drumlin and the till-bedrock interfaceat its base.

A commercially available 24-channel seismicsystem was used to record seismic arrivals at theground surface. The system consisted of 24 vert-ical 40 Hz geophones connected to a GeometricsGeode seismograph, which in turn was linked to a laptop computer. The OYO SeisImager software,version 2.20, was used for picking of refractedarrivals in the recorded seismograms, estimation ofvelocity models using delay times, and for tomo-graphic reconstruction of layer seismic velocities and thicknesses. The SeisImager software is widelyused in near-surface applications of the seismicrefraction method (e.g. Sheehan et al., 2005). Thetomographic reconstruction iteratively updates auser-specified initial model, consisting of layervelocities and thicknesses, until a minimum, non-linear least-squares misfit is obtained betweenobserved and calculated seismic velocities.

Several seismic refraction profiles with a geo-phone spacing of 3 m were acquired using nine forward or reverse shots per profile. Shot pointswere located at successively greater in-line dis-tances from the geophone array. This yielded totalprofile lengths of approximately 170 m. Profileswere aligned either south-north parallel to the longaxis of the drumlin, or west-east parallel to the short

GSP_4_C11.qxd 11/5/07 13:33 Page 167


axis of the drumlin. These refraction profiles wereacquired for use in reconstructing the internalarchitecture of the drumlin. Shorter seismic refrac-tion profiles with a geophone spacing of 1 m werealso acquired, again using several, laterally offsetforward or reverse shots. These shorter profiles (total length: ~85 m) served for accurate estima-tion of very near-surface seismic velocities, and forimproving the user-specified initial model for tomo-graphic inversion of the longer seismic refractionprofiles (above). Respectively one long refractionprofile in the south-north direction (profile SSN) andin the west-east direction (profile SWE) are reportedhere (Fig. 1C), together with a short refraction pro-file centred on the midpoint of the long refractionprofile SWE.

Electrical resistivity surveys

Electrical resistivity surveys create an electricalcurrent in the ground using two dedicated elec-trodes, and measure the resulting electrical potentialfield using a further two electrodes. By combininga large number of such four-pole measurements atregular intervals along a profile, the bulk electricalresistivity distribution (inverse of bulk electrical conductivity) in the subsurface along the profile can be reconstructed. It was expected that theinterfaces between (i) the partially saturated andthe saturated tills, and (ii) the overlying till and theunderlying greywacke bedrock could be detectedbased on noticeable contrasts in bulk resistivitybetween these layers.

A commercially available electrical resistivityimaging system, the IRIS Syscal R1 with 48 nodes(electrodes), was used in the present study. Thestainless steel electrodes were inserted into theground, and connected to an imaging cable, atregular spacing along a straight line. The cable wasconnected to the Syscal R1 resistivity meter, whichwas programmed by the user to switch between alarge number of electrical four-poles. Data acquisi-tion was rapid and automated, with the data beinginitially stored in the meter and later downloadedto a PC once acquisition completed. The inversioncode DCIP2D (e.g. Li & Oldenburg, 2000) was usedto tomographically reconstruct the bulk subsurfaceresistivity distribution. The reconstruction processinvolves iterative updating of a current model, calculated from the field data together with the

measurement errors, until the least-squares misfitbetween observed and calculated data is minimised(e.g. Li & Oldenburg, 2000). The current model isthen assumed to correspond as closely to the actualbulk resistivity distribution in the subsurface as can feasibly be reconstructed from the field data.

Several electrical resistivity profiles with an elec-trode spacing of 5 m, yielding total profile lengthsof 235 m, were acquired in the same area as the seismic refraction surveys. Profiles were alignedeither south-north parallel to the long axis of thedrumlin, or west-east parallel to the short axis ofthe drumlin. These four resistivity profiles wereacquired for use in reconstructing the internalarchitecture of the drumlin. Each of the long pro-files was repeated with a smaller electrode spacingof 1 m, yielding shorter profile lengths of 47 m. The shorter profiles were used to estimate thevery near-surface electrical resistivity distribution,which allowed estimation of the thickness of thepartially saturated zone, as well as improvementof the tomographic reconstruction of the longprofiles. Respectively one long resistivity profile inthe south-north direction (profile RSN) and in thewest-east direction (profile RSN) are reported here(Fig. 1C), together with a short resistivity profilecentred on the midpoint of the long profile RSN.

Borehole logs and other relevant field observations

During borehole drilling, geological materialsencountered with depth were logged by the on-sitecrew. Upon completion of drilling, falling-head per-meability tests were conducted to measure bulkhydraulic conductivity. Vibrating-wire sensors wereinstalled in all four boreholes to continuouslymonitor the long-term response of pore waterpressure to rainfall events and stress relief duringexcavation. Excavation of a south-north section ofthe drumlin as part of road construction (Fig. 1B)provided an exposure along which the bedrock elevation could be established in the direction ofthe roadway.

RESULTS

Inverted, long seismic refraction and electricalresistivity profiles in the south-north and west-eastdirections are illustrated in Figs 2 and 3. These

GSP_4_C11.qxd 11/5/07 13:33 Page 168


Fig. 2 Representative long seismicrefraction profiles. (A) Profile SWE. (B) Profile SSN. Approximate boreholeslocations are shown for reference,labelled 1, 2, 3, 4. Note that theseismic velocities of topsoil andpartially saturated till combine toproduce an averaged value of <1.3 km s−1. See Fig. 1 for a plan viewof profile and borehole locations.

Fig. 3 Representative long electricalresistivity profiles. (A) Profile RWE. (B) Profile RSN. Approximateboreholes (BH) locations are shownfor reference. See Fig. 1 for a planview of profile and boreholelocations.

Dep

th (

m)

0

20

40

60

80

1

0 40 80 120 160

2

A

W E

4 3

Dep

th (

m)

0

20

40

60

800 40 80

Distance (m)

km s–1

<1.3

120 160

B

S N4 3

1.7–2.3 <2.7 >3.2

Dep

th (

m)

0

20

40

BH 1

0 40 20016012080

BH 2A W E

BH 3

Dep

th (

m)

0

20

400 40 200

100806040

Ω m

160120

Distance (m)

80

B S NBH 3BH 4

GSP_4_C11.qxd 11/5/07 13:33 Page 169


profiles were representative and chosen based onthree criteria: (a) profile direction, guided by theneed to illustrate changes in seismic velocity andelectrical resistivity along both the long and shortaxes of the drumlin; (b) proximity to a maximumnumber of boreholes; and (c) avoidance of repeti-tion (including further profiles would add no extravalue). Similarly, the inverted, short seismic refrac-tion and electrical resistivity profiles illustrated inFig. 4 were respectively representative of all shortseismic and resistivity profiles acquired.

Seismic refraction profiles

West-east profiles were aligned parallel to the shortaxis of the drumlin, and therefore typically have anoticeably steeper surface slope (Figs 2A and 4A)than profiles aligned south-north and thus parallelto the long axis of the drumlin (Fig. 2B). Synthesis oflong and short refraction profiles reveals a sequenceof five horizontal layers, having seismic velocitiesof <0.6 km s−1 (shallowest layer), 1.0–1.3 km s−1

(layer 2), 1.7–2.3 km s−1 (layer 3), 2.7 km s−1 (layer4) and >3.2 m s−1 (deepest layer) (Figs 2 and 4A;Table 1). Velocity errors are analysed in Section 4.5,including a discussion of possible origins of thevelocity ranges given for layers 2 and 3 (Table 1).

Electrical resistivity profiles

The west-east resistivity profiles are aligned para-llel to the short axis of the drumlin, and thereforehave a noticeably steeper surface slope (Fig. 3A)than the profiles aligned south-north (Figs 3B and4B). Synthesis of long and short resistivity pro-files reveals the presence of three dominant layersincluding; a shallow layer of >100 Ω m, an inter-mediate layer of <50 Ω m, and a deep layer of>100 Ω m (Figs 3 and 4A, Table 1). As discussed in more detail in Section 4.3, comparison of layerthicknesses with those obtained from refractionimaging suggests that the shallow resistivity layercorresponds to a combination of seismic layers 1and 2, and the intermediate and deep resistivity layers correspond to seismic layers 3 and 5 (Table 1). Uncertainties in estimates of resistivity and layer thickness are analysed below. Notably,in all long west-east profiles resistivities of the inter-mediate layer were lower downslope (eastern endin Fig. 3A) than nearer the centre of the drumlin(western side in Fig. 3A). Many localised anomaliessituated within the shallow and the intermediatelayers were also found. These anomalies variablywere less or more resistive than the dominant layerresistivity.

Dep

th (

m)

0

10

20

0 20 806040

Distance (m)

A W E

km s–1

< 0.6 1.0–1.3 > 1.7

Dep

th (

m)

0

5

0 10 40

1007040

Ω m

3020

Distance (m)

B S N

Fig. 4 Representative short seismicrefraction and electrical resistivityprofiles. (A) Short seismic refractionprofile centred the midpoint of the long profile SWE (see Fig. 2A). (B) Short electrical resistivity profilecentred the midpoint of the longprofile RSN (see Fig. 3B). Note that in(B) the resistivities of topsoil andpartially saturated till combine toproduce an averaged value of >100 Ω m. See Fig. 1 for a plan viewof profile and borehole locations.

GSP_4_C11.qxd 11/5/07 13:33 Page 170


Borehole logs and other relevant field observations

The two western (boreholes 1 and 2) and eastern(boreholes 3 and 4) logs were characterised by sim-ilar lithologies respectively (Fig. 5). In the westernboreholes, topsoil approximately 0.5 m thick isunderlain by glacial tills of varying consistency andcolour, although differences in these properties aresmall (Fig. 5A). The total thickness of the till unitsvaried between 23.5 m in borehole 1 and 22.5 m inborehole 2. In both boreholes a thin (~0.5 m) bandof sandy gravel is sandwiched between the tilland the underlying greywacke bedrock (Fig. 5A).Since the boreholes only marginally penetratedbedrock, neither thickness nor spatial extent of theweathered greywacke can be ascertained from the borehole logs. Boreholes 3 and 4 were locatedeast of boreholes 1 and 2 (Fig. 1C), approximately50–80 m further away from the centroid of thedrumlin. The thickness of topsoil (~0.5 m) andsequence of glacial till units in the eastern boreholes(Fig. 5B) were similar to those in the western bore-holes (Fig. 5A). Till thickness totalled 18.5 m and20 m in boreholes 3 and 4 respectively, and wastherefore smaller than in the western boreholes,albeit not substantially. By comparison with thewestern boreholes, no layer of sandy gravel wasfound in the eastern boreholes, but the latter consistently terminated in weathered greywackebedrock (Fig. 5).

Further evidence of internal drumlin architec-ture comes from field observation conducted onceexcavation had begun (Fig. 1B), exposing topsoil,till profiles, and in some areas bedrock. Next to the

Table 1 Geophysical anatomy of drumlin. Seismic velocities are accurate to within ±0.05 km−1, and electricalresistivities to within ±4 Ω m (Section 4). The resistivities of layers 1 and 2 combine to produce an averaged valueof >100 Ω m (Fig. 4b). The layer of weathered greywacke could not be detected using the electrical resistivitysurveys, and is therefore marked by ‘?’.

Description Velocity (km s−1) Resistivity (X m)

Layer 1 Topsoil <0.6 # >100Layer 2 Partially saturated till 1.0–1.3 $Layer 3 Saturated till 1.7–2.3 <50Layer 4 Weathered greywacke ~2.7 ?Layer 5 Competent greywacke >3.2 >100

0 m

2.0 m

7.0 m

16.0 m

22.5 m23.5 m24.0 m

TopsoilA B0 m

2.0 m

4.5 m

11.0 m

20.0 m

Topsoil

Stiff grey brown gravelly mottled silt with some cobbles

Very stiff grey sandy gravelly silt with some cobbles

Very stiff grey sandy gravelly silt with some cobblesand boulders

Very stiff dark grey sandy gravelly silt with some cobbles andboulders

Band of dense grey and fine medium and coarse sandy gravelwith some cobbles

Completely to highly weathered grey shale rock

Highly to moderately weathered grey shale rock

Fig. 5 Representative boreholes logs as recorded during drilling by the on-site crew. (A) Borehole 2. (B) Borehole 4. See Fig. 1 for a plan view of boreholelocations.

GSP_4_C11.qxd 11/5/07 13:33 Page 171


dominant layering apparent from the four bore-hole profiles, several inclusions of sand, gravel orclay, in some cases mixed to varying proportions,were observed to be embedded within the tills. Such inclusions were often several metres in vert-ical and horizontal extent (Fig. 6). Several fielddrains, exposed as thin, long bands of gravel, werealso observed close to the surface of the drumlin.Detailed results from falling-head permeability tests,pore water pressure monitoring, and other geo-technical tests will be reported elsewhere. Insitu permeability tests revealed that the hydraulic conductivity of the till units is very small (~10−9–10−8 m s−1) compared to that of the underlyingweathered bedrock (hydraulic conductivity wastoo large to be measured reliably). Monitoredwater levels allowed the seasonal fluctuations of the thickness of the partially saturated zone to be tracked within the upper till units. The 2-D,south-north profile of bedrock elevation, recon-structed post-excavation, revealed that total thick-ness of topsoil and till units was several metres less

near the southern margin of the survey area thanat the northern margin. Groundwater discharge wasobserved to be common near the toe of the drumlin.

INTERPRETATION

Calibration of geophysical data

Geophysically-derived layering is convenientlycalibrated using the borehole logs (Fig. 5) togetherwith other field observations, as summarised inTable 1. Layers 1, 2, and 3 in the seismic refractionand electrical resistivity data correspond to topsoiland partially saturated and saturated glacial tills.Layer 4 corresponds to weathered greywackebedrock, and in contrast to the seismic refractiondata cannot be distinguished reliably in the elec-trical resistivity data (Table 1). In the absence of anyother plausible explanation, it is inferred that thatlayer 5 most likely corresponds to relatively com-petent bedrock. The interface between weatheredbedrock (lower seismic velocity of approximately2.7 km s−1) and more competent bedrock (higherseismic velocity of >3.2 km s−1) thus gives rise to astrong seismic refraction.

Neither partially saturated (1.0–1.3 km s−1) nor saturated (1.7–2.3 km s−1) tills could be assigned atypical seismic velocity (Table 2). Instead, velocitywas found to increase progressively with depth in these layers, which is interpreted to be due toincreasing material bulk density with depth. A sim-ilar increase in seismic velocity in a high-resolution,cross-borehole seismic profile of a section of glacialtill was observed at a local test site (Mulholland,2004).

Topsoil and glacial till units

Following calibration, the seismic refraction datain particular allow multi-dimensional extrapolationof lithology across the drumlin, thus complement-ing results from more traditional geotechnical siteinvestigation well. Topsoil thickness is largely con-stant at approximately 0.5 m within the survey area(Fig. 2), and the overall thickness of the partiallysaturated zone (including topsoil) is typically 2–2.5 m (Fig. 4A). Noticeable spatial changes in thethickness of topsoil and the partially saturated zoneappear to occur only near the southern (Fig. 2B) and eastern (Fig. 4A) margins of the survey area.

Fig. 6 Example of a massive clay inclusion, showing oneof authors (~2 m tall) for scale.

GSP_4_C11.qxd 11/5/07 13:33 Page 172


These apparent changes must be considered withgreat caution since seismic data coverage near themargins of the survey area is poor compared tomore central portions. The changes may thus wellbe artefacts produced by mathematical extrapola-tion during data inversion. Any minor changes inthickness of topsoil and partially saturated zone(Figs 2 and 4A) must similarly be considered withcaution since they typically lie within the range ofuncertainty of depth estimates.

Seismic refraction (Fig. 2) and electrical resistivity(Fig. 3) data consistently suggest that the thicknessof the saturated zone is greatest near the centre of thedrumlin, thinning towards the southern and easternmargins of the survey area. These changes inthickness are generally consistent with the boreholelogs (Fig. 5) and the 2-D south-north profile ofbedrock elevation reconstructed post excavation.The direction of thinning of the saturated tills is,thus, aligned in the direction of former ice flow.

Importantly from a geotechnical perspective,electrical resistivities of saturated tills were foundto decrease away from the centre of the drumlin(Fig. 3A). As discussed in popular geophysicaltextbooks (e.g. Reynolds, 1997; Sharma, 1997), themost prominent factors that could generate adecrease in resistivity within a given material areincreases in moisture content, porosity, cementa-tion of grains, conductivity of the waters filling the pores, or sediment clay content. Consistentdownslope variations in till composition were not observed, and in particular neither porosity,cementation of grains, water conductivity, or claycontent changed consistently in the study area. Themost likely explanation for the observed, down-slope decrease in resistivity is therefore a gradualincrease in till moisture content.

The marked difference in hydraulic conductivitybetween the till units and the weathered bedrockwas modelled to produce nearly vertical ground-water flow down through the drumlin to theunderlying aquifer. Groundwater flow directionwas modelled to be reversed near the toe of thedrumlin, where strong upward gradients werepredicted to cause groundwater discharge at theground surface. Field observations confirm thatsuch discharge indeed occurs at the toe of thedrumlin. The upward hydraulic gradients are inter-preted to decrease the effective stress near the toeof the drumlin, resulting in the inferred increase in moisture content and decrease in resistivity inthis area. This is important since softening of tillsdue to a decrease in effective stress may decreasethe stability of the drumlin slopes.

Depth to bedrock

Seismic and electrical resistivity estimates of depthto bedrock are summarised in Table 2 togetherwith actual depths taken from the borehole logs.For convenient visual comparison, boreholes aresuperimposed on seismic and resistivity images inFigs 2 and 3.

For seismic profiles SWE and SSN the sudden jumpin velocity from a maximum of approximately 2.3 km s−1 in layer 3 to approximately 2.7 km s−1

in layer 4 (Table 1, Fig. 2) is inferred to indicate the transition from dense, fully saturated tills to weathered bedrock. Within the given ranges of uncertainty, depths to bedrock estimated for both seismic profiles (SWE and SSN) match actualdepths in all boreholes closely (Table 2). Notably,the thickness of weathered bedrock (2.7 km s−1)appears to suddenly increase near the southern

Table 2 Comparison of depths to bedrock (from ground surface using a topsoil thickness of 0.5 m), as taken from borehole (BH) logs and seismic refraction (SWE, SSN) and electrical resistivity profiles (RWE, RSN)(Sections 4.3 and 4.5).

Log (m) SWE (m) SSN (m) RWE (m) RSN (m)

BH 1 24.5 24 ± 0.5 – 23 ± 1.5 –BH 2 23.5 23.5 ± 0.5 – 23 ± 1.5 –BH 3 19 19.5 ± 0.5 20.5 ± 0.5 – 20.5 ± 1.5BH 4 20.5 19.5 ± 0.5 21 ± 0.5 – 21.5 ± 1.5

GSP_4_C11.qxd 11/5/07 13:33 Page 173


margin of survey area (Fig. 2B). Since seismic datacoverage near the margins of the survey area is poor compared to more central portions), thisobserved increase is probably an artefact pro-duced by mathematical extrapolation during datainversion.

For resistivity profiles RWE and RSN depths were determined by calculating vertical resistivity gradients at the locations closest to the boreholes(Figs 1C and 3), and taking the depth correspond-ing to the largest increase in resistivity as beingequal to depth to bedrock. Within the given rangesof uncertainty, depths to bedrock estimated forboth resistivity profiles match actual boreholedepths closely (Table 2).

Drumlin facies and formation

In elucidating past glaciological conditions and pro-cesses operating at the base of the Late Devensianice sheet in Northern Ireland, it is interesting to classify the drumlin surveyed here within theframework of the five distinct facies associationsidentified by McCabe & Dardis (1989). The collect-ive evidence from borehole logs (Fig. 5), seismic(Figs 2 and 4A) and electrical resistivity (Figs 3 and 4B) surveys, and associated field observa-tions (Fig. 6) is entirely consistent with the criteriadefining Facies Association 3, but violates at least one criterion defining each of the other four facies associations so that the drumlin cannot bepart of:

• Facies Association 1 because it is not located neara major ice divide and apparently does not have a coreof older drift;• Facies Association 2 because it is located in the eastrather than the west of Ireland and apparently lacksevidence of interbedded diamictons, muds, sands, orgravels;• Facies Association 4 because it is not located in asubglacial tunnel-type valley and sequences of sandsand gravels are not apparent;• Facies Association 5 because there is no apparentevidence for interbedded sequences of massive orstratified diamicts and sand or gravel lithofacies.

We therefore infer that the drumlin is part of thesubglacial lodgement, melt-out, debris flow, sheet flowfacies (Facies Association 3, McCabe & Dardis,1989). Synthesis of the findings from the present

study with the detailed facies model of McCabe &Dardis (1989) reveals that:

• Several stratified units of poorly sorted till/diamicton are present, thinning and generally dippingtowards the south/south-east, and thus in the generaldirection of former ice flow in this area. These unitsare believed to have formed by re-sedimentation ofmaterial originally deposited subglacially by debrisflow.• Near the centre of the drumlin, a horizontallyextensive (>20 m in diameter), thin (~0.5 m) inclusionof sand and gravel is sandwiched between till andweathered bedrock. Such basal deposits are typicalof sheet flow events during transient, enhanced flowof subglacial melt water.• Numerous, often extensive sediment inclusions ofvarying geometries are present within the till units,consisting of various mixtures of clay, sand, gravel,cobbles, and boulders. These inclusions probablyoriginated at different times by transient subglacialdebris and sheet flow events.• Weathered, hydraulically highly permeable bedrockis present beneath the saturated till units, forming acontinuous layer that is typically >5 m thick. Allowingfor temporal changes in the thermal regime at the baseof the former ice sheet, weathering processes likelyincluded bed fracture and formation of Kamb-typecavities and/or Nye-type meltwater channels at dif-ferent times (e.g. Knight, 2002).

The main phase of original sedimentation isbelieved to have occurred prior to drumlinisationin this area. Slow or negligible ice sheet motion is pre-requisite to sedimentation of several tens ofmetres of subglacial debris (e.g. McCabe & Dardis,1989), which likely occurred particularly during theGlenavy Stadial (~25–18 14C kyr BP; e.g. Knight,2002). It has been speculated that drumlins werelater streamlined by fast ice flow during the BeldergStadial (~18–16.6 14C kyr BP; e.g. Knight, 2002),and further that fast ice flow eventually lead to the rapid disintegration of the Late Devensian icesheet (e.g. Eyles & McCabe, 1989).

Uncertainty in geophysical estimates

Two main lines of evidence can be used to assessthe accuracy of geophysical estimates of subsur-face layer geometries and physical properties (i.e.seismic velocity and electrical resistivity). First, aninitial assessment of uncertainty can be obtained

GSP_4_C11.qxd 11/5/07 13:33 Page 174


by comparing geophysically derived layer thick-nesses or depths to bedrock with actual thick-nesses and depths taken from the borehole logs, asimplied e.g. by Table 2. However, this shows onlyan approximate indication of geophysical accuracysince boreholes were often located several metres to>10 m away from the geophysical profiles (Fig. 1C).Second, estimates of uncertainty can be obtainedby systematically comparing geophysically derivedlayer geometries and physical properties at thecross-over points between seismic profiles or be-tween electrical resistivity profiles (Fig. 1C). This hasthe advantage that geophysical estimates shouldmatch between any two profiles at their cross-overpoint so that quantitative estimation of relativeerrors is possible. However, it has the disadvant-age that absolute errors cannot be estimated sincea ground-truthing data set (as derived e.g. fromborehole logs) is not considered.

Application of standard statistical techniques oferror estimation to all available data for cross-overpoints (including all additional geophysical data not presented here) reveals that seismic velocitiesare accurate to within ±0.05 km s−1, and depths tointerfaces derived from seismic refraction data areaccurate to within ±0.5 m. Estimates of electricalresistivity were found to be accurate to within ±4 Ω m, and depths to interfaces derived fromelectrical resistivity data are accurate to within±1.5 m. Within the given ranges of uncertainty,comparison of geophysically-derived depths tobedrock with actual depths taken from boreholeslogs (Table 2) was found to be satisfactory in allcases.

SYNTHESIS AND CONCLUSIONS

Synthesis of results

The geophysical anatomy of the drumlin consistsof five distinct layers successively including; top-soil (~0.5 m thick), partially saturated glacial tills(~1.5–2 m thick), fully saturated glacial tills (up to >20 m thick), weathered greywacke bedrock (>5 m thick), and more competent greywackebedrock. This succession of layers is confirmed byborehole logs and direct field observations. Thereare numerous, often extensive, inclusions of clay,sand, gravel, cobbles, and boulders included withinthe topsoil and the till units.

Novelty value of the present study

It is now possible to answer the three key questionsthat motivated the present study:

1 The key additional value of geophysical surveyingis its superior spatial coverage, often supporting re-construction of subsurface properties in three spatialdimensions at metre-scale spatial resolution. This is particularly desirable in sedimentological recon-structions of drumlins since suitably large outcropsfor field mapping are commonly rare, constituting the drumlin problem (e.g. Menzies, 1987; McCabe &Dardis, 1989). The present study presents a particu-larly desirable scenario: initially geophysical dataare calibrated at point locations in terms of drumlinstratigraphy using the borehole logs and other fieldobservations, and subsequently these data are usedto extrapolate changes in stratigraphic layering acrossthe drumlin. Here, additional value is also added tothe information from borehole logs since:

(a) Drilling through bedrock is logistically andfinancially particularly expensive, such that bore-holes were terminated when weathered bedrock was reached. In contrast, relatively cheaply acquiredseismic refraction data could map both thicknessand spatial extent of the weathered layer, which isgeotechnically particularly relevant.(b) Low electrical resistivities are diagnostic ofincreased till moisture content near the toe of thedrumlin, indicating strong upward gradients inwater flow which potentially increase the risk oflandslides. Detecting and delineating such zones of enhanced moisture content using traditional,invasive techniques of geotechnical site investiga-tion is logistically and financially more expensive.

2 Geophysical estimates of depths to interfaces werefound to be accurate to within ±0.5 m and ±1.5 m forseismic refraction and electrical resistivity data, respect-ively. Within these ranges of uncertainty, geophysicallyderived depths to bedrock matched those taken fromborehole logs very well (Table 2). Further case studiesmust confirm whether such encouraging results area fortunate exception in the present case, or indeedtypically the norm in seismic refraction and electricalresistivity reconstructions of drumlin anatomy.3 Since the seismic refraction method is subject to con-siderably less depth uncertainty than the electricalresistivity method (±0.5 m vs. ±1.5 m), it is relat-ively well suited for mapping of layering within thedrumlin. This inference is confirmed by the fact thatelectrical resistivity data could neither detect nordelineate the layer of weathered bedrock, and could

GSP_4_C11.qxd 11/5/07 13:33 Page 175


not distinguish between topsoil and partially saturatedtills. In contrast, electrical resistivity is much bettersuited to delineating lateral changes in till moisturecontent and mapping inclusions within topsoil andtill than seismic refraction imaging.4 Together geophysical and geotechnical findingsimply that the drumlin is part of the subglacial lodge-ment, melt-out, debris flow, sheet flow facies as describedby McCabe & Dardis (1989). The drumlin is inferredto have formed by re-sedimentation and streamlin-ing of pre-existing sediments, controlled primarily by debris and sheet water flows at the base of fast-flowing ice during deglaciation of the Late Devensianice sheet (e.g. McCabe & Dardis, 1989; Knight, 2002).

Future work

The findings presented here generally advocatethe use of seismic refraction and electrical resistivitymethods in reconstructing the internal architectureof glacial sediments. This has triggered profoundinterest in conducting further extensive studies at a number of other sites in Northern Ireland.Motivated particularly by the success of the presentstudy, it is planned to use seismic and electrical geophysical methods as a standard tool of siteinvestigation in the future.

ACKNOWLEDGEMENTS

Gordon Clarke acknowledges a DEL Ph.D. scholar-ship, and Lee Barbour an EPSRC Visiting Fellow-ship. The authors would particularly like to thankEamon O’Loughlin and Michael Sexton who spentmany hours in the field in summer 2003 collectingthe geophysical data. The Geophysical InversionFacility of the University of British Columbia pro-vided a free academic license for the code DCIP2D,which is gratefully acknowledged. The Constructionand Road Services of Northern Ireland providedsupport in many different ways, which made thisstudy possible. We thank Neil Glasser and JohnWoodward for their detailed comments on theoriginal version of this manuscript.

REFERENCES

Birch, F.S. (1989) A geophysical study of Quaternary sediments near the Late Pleistocene marine limit

in Epping, New Hampshire. Northeast Geol., 11,124–132.

Eyles, N. and McCabe, A.M. (1989) The Late Devensian(<22,000 BP) Irish Sea Basin: The sedimentary recordof a collapsed ice sheet margin. Quaternary Sci. Rev.,8, 307–351.

Knight, J. (2002) Bedform patterns, subglacial meltwaterevents, and Late Devensian ice sheet dynamics innorth-central Ireland. Global Planet. Change, 35, 237–253.

Li, Y. and Oldenburg, D.W. (2000) 3-D inversion ofinduced polarisation data. Geophysics, 65, 1931–1945.

McCabe, A.M. (1987) Quaternary deposits and glacialstratigraphy in Ireland. Quaternary Sci. Rev., 6, 259–299.

McCabe, A.M. and Dardis, G.F. (1989) A geological viewof drumlins in Ireland. Quaternary Sci. Rev., 8, 169–177.

Menzies, J. (1987) Towards a general hypothesis on the formation of drumlins. In: Drumlin Symposium(Eds J. Menzies and J. Rose), pp. 9–24. Balkema,Rotterdam.

Mulholland, P. (2004) Cross-borehole seismic transmissionsurveys at the EERC test site, Queen’s University Belfast.Unpublished B.S. thesis, Queen’s University Belfast,Belfast, Ireland.

O’Loughlin, E. (2003) The use of seismic refraction forbedrock detection at an engineering site, Loughbrickland,County Down. Unpublished M.S. thesis, Queen’sUniversity Belfast, Belfast, Ireland.

Reynolds, J.M. (1997) An introduction to applied and environmental geophysics. John Wiley & Sons.

Sexton, M. (2003) The use of electrical resistivity surveyingas a complementary technique for traditional geotechnicalinvestigations. Unpublished M.S. thesis, Queen’sUniversity Belfast, Belfast, Ireland.

Sharma, P.V. (1997) Environmental and Engineering Geo-physics. Cambridge University Press.

Sharpe, D., Pugin, A., Pullan, S. and Shaw, J. (2004)Regional unconformities and the sedimentary archi-tecture of the Oak Ridges Moraine area, southernOntario. Can. J. Earth Sci., 41, 183–198.

Sheehan, J.R., Doll, W.E. and Mandell, W.A. (2005) Anevaluation of methods and available software forseismic refraction tomography analysis. J. Environ. Eng.Geoph., 10, 21–34.

Sutinen, J. (1985) Application of radar, electrical resistiv-ity, and seismic soundings in the study of morainiclandforms in northern Finland. In: INQUA TillSymposium, Geological Survey of Finland, SpecialPaper, 3, 65–75.

Van der Veen, M., Bueker, F., Green, A.G. and Buness, H.A.(2000) Field comparison of high-frequency seismicsources for imaging shallow (10–250 m) structures. J. Environ. Eng. Geoph., 5, 39–56.

GSP_4_C11.qxd 11/5/07 13:33 Page 176

glacial sedimentary processes and products (hambrey/glacial sedimentary processes and products) ||...

Documents