identifying causes of ground-penetrating radar reﬂections ...rvd/pub/sed_2000.pdf · time-domain...

Identifying causes of ground-penetrating radar re¯ections usingtime-domain re¯ectometry and sedimentological analyses

REMKE L. VAN DAM and WOLFGANG SCHLAGERDepartment of Sedimentology, Faculty of Earth Sciences, Vrije Universiteit, De Boelelaan 1085,1081 HV Amsterdam, The Netherlands. (E-mail: [email protected])

ABSTRACT

Ground-penetrating radar (GPR) is a geophysical technique widely used to

study the shallow subsurface and identify various sediment features that re¯ect

electromagnetic waves. However, little is known about the exact cause of GPR

re¯ections because few studies have coupled wave theory to petrophysical

data. In this study, a 100- and 200-MHz GPR survey was conducted on aeolian

deposits in a quarry. Time-domain re¯ectometry (TDR) was used to obtain

detailed information on the product of relative permittivity (er) and relative

magnetic permeability (lr), which mainly controls the GPR contrast parameter

in the subsurface. Combining TDR data and lacquer peels from the quarry wall

allowed the identi®cation of various relationships between sediment

characteristics and erlr. Synthetic radar traces, constructed using the TDR

logs and sedimentological data from the lacquer peels, were compared with the

actual GPR sections. Numerous peaks in erlr, which are superimposed on a

baseline value of 4 for dry sand, are caused by potential GPR re¯ectors. These

increases in erlr coincide with the presence of either organic material, having a

higher water content and relative permittivity than the surrounding sediment,

or iron oxide bands, enhancing relative magnetic permeability and causing

water to stagnate on top of them. Sedimentary structures, as re¯ected in

textural change, only result in possible GPR re¯ections when the volumetric

water content exceeds 0á055. The synthetic radar traces provide an improved

insight into the behaviour of radar waves and show that GPR results may be

ambiguous because of multiples and interference.

Keywords Dielectric impedance, ground-penetrating radar, magnetic

permeability, relative permittivity, time-domain re¯ectometry.

INTRODUCTION

The seismic re¯ection method and ground-pen-etrating radar (GPR) are important geophysicalreconnaissance tools for the shallow subsurface.Both techniques are based on wave propagationand re¯ection, but wavelength ranges, and con-sequently resolution and penetration depths,differ signi®cantly. The seismic re¯ection methodhas a better penetration but lower resolutionthan GPR. Also, the contrast parameter for theseismic re¯ection method (acoustic impedance)differs from that for GPR (dielectric impedance).Yet, the two techniques are suf®ciently analog-ous for experience gained with one to be

applicable to the other, increasing knowledgeof the origin of re¯ections in both the seismicre¯ection method and GPR (Cardimona et al.,1998).

In the seismic re¯ection method, most re¯ec-tions are parallel to depositional bedding. Twodifferent types of re¯ection exist (Mayer, 1980):(1) re¯ections that occur at the location of majorchanges in acoustic impedance; and (2) re¯ec-tions that represent patterns resulting from con-structive and destructive interference betweenthe acoustic wave train and small impedancevariations in the sediments or rocks. Only the ®rsttype of re¯ection serves as a reliable guideto depositional history and geometry, and its

Sedimentology (2000) 47, 435±449

Ó 2000 International Association of Sedimentologists 435

amplitude is proportional to the magnitude ofchange. In contrast, the second type of re¯ectiondoes not represent a major impedance change atthe measured travel time (Mayer, 1979). Animportant method for distinguishing these twore¯ection types is detailed comparison of seismictraces with sediment and rock properties, asdetermined from cores and outcrop studies.These outcrop studies are conducted in depositsthat are considered to be analogous (Kenter et al.,1999) or on exposures in the vicinity of theseismic line (Sta¯eu & Sonnenfeld, 1994). Thesediment and rock properties from the outcropstudies can be used to generate lithologic andacoustic impedance models and are widely usedto construct synthetic seismic traces and sections(Fagin, 1991; Sta¯eu & Schlager, 1995; BraccoGartner & Schlager, 1999). These synthetic imagesimprove the understanding of seismic sectionsand re¯ections. It cannot be assumed, however,that the rock and sediment properties measuredin outcrops are completely similar to thosere¯ected in the seismic trace.

GPR measures changes in the dielectric pro-perties of sediments that cause re¯ection ofelectromagnetic energy. These changes in dielec-tric properties result primarily from changes inwater content, governed in turn by grain sizeand porosity (Topp et al., 1980; Roth et al., 1990;Sutinen, 1992; Huggenberger, 1993). As grain sizeand porosity changes are related to depositionalhistory, a clear relationship can be expectedbetween sedimentary structures and dielectricproperties, allowing accurate identi®cation ofradar facies and sequence boundaries (Gawthorpeet al., 1993).

Most work with GPR has focused either ongeological and sedimentological reconnaissance(Jol & Smith, 1991; Huggenberger, 1993; Bereset al., 1995; Bristow et al., 1996; Asprion &Aigner, 1997; Bridge et al., 1998; Rea & Knight,1998; Van Heteren et al., 1998; Van Overmeeren,1998) or on synthetic modelling of wave propa-gation (Carcione, 1996; Casper & Kung, 1996;Hollender & Tillard, 1998). Studies in whichwave theory is integrated with geological orpetrophysical data, common in re¯ection seis-mics, are sparse. As a result, little qualitative andquantitative knowledge exists on the cause ofGPR re¯ections. GPR shows sedimentary struc-tures and other features in the subsurface, but noinformation is available on the origin of re¯ec-tions or on the behaviour of the contrast param-eter, other than from the re¯ection image (and itsparameters, such as re¯ection strength). However,

many dielectric impedance changes are beyondthe resolution and sensitivity range of GPR. For afull understanding of the origin of GPR re¯ec-tions, dielectric impedance variations smallerthan those observed on radar images must beanalysed and explained.

In this study, time-domain re¯ectometry (TDR),a technique well known in soil sciences (Toppet al., 1982; Heimovaara et al., 1995), has beenused to correlate GPR data with subsurfaceinformation by measuring dielectric propertiesof sediment at small intervals along verticalsections in a quarry. TDR measures the propaga-tion velocity of an electromagnetic wave along asteel rod probe that is pushed into the sediment,and is best applied in unconsolidated sand-sizedand ®ner sediments. In coarse-grained sedimentthe TDR method functions less well due todisturbance of the sediment fabric during thepenetration of the TDR rods. The TDR measure-ments were interpreted using detailed sedimen-tological information from lacquer peels andgrain-size analyses, providing insight into thepossible causes of GPR re¯ections. Based on theTDR logs and the sedimentological informationfrom the lacquer peels, a dielectric impedancemodel of the subsurface was constructed. Thismodel, in which the sediment column is subdi-vided into distinct layers with characteristicdielectric properties, was used to construct one-dimensional synthetic radar traces. By analogywith the seismic re¯ection method, such syntheticimages can be compared with real GPR sectionsfor a better understanding and quanti®cation ofparameters controlling GPR re¯ections. Ultimate-ly, such comparisons will improve the knowledgeof waves and their re¯ections in general, whichwill also bene®t the seismic re¯ection method.

METHODS

Ground-penetrating radar

In this study, a Sensors & Software pulse-EKKO100 GPR system (Table 1), consisting of atransmitting and receiving antenna connected to aconsole and laptop computer (Fig. 1A), was used.The transmitted electromagnetic pulse is ideallymeant to penetrate the subsurface in a beam asnarrow as possible. Some of the energy, however,travels directly to the receiving antenna as air-wave and groundwave, giving the signal at the topof the resulting radar section (Fig. 1B). Part of theremaining energy, which enters the subsurface,

436 R. L. Van Dam and W. Schlager

Ó 2000 International Association of Sedimentologists, Sedimentology, 47, 435±449

re¯ects at layers of changing dielectric impedanceand travels back to the receiver. The quantity ofenergy received and the associated arrival timeare stored in the computer. The lateral extent andmorphology of re¯ectors can be delineated bymoving the portable equipment across the sur-face. The resulting radar section, on which eachmeasurement point is represented by a trace,shows time along its vertical axis and positionalong its horizontal axis. The velocity of radarwaves in different layers can be calculatedthrough common mid-point (CMP) measurements(Arcone, 1984), allowing conversion of traveltime to actual depth.

The transfer and re¯ection of electromagneticenergy in the subsurface depends on the dielec-tric impedance (Z) of the medium and can becalculated by (Brewster & Annan, 1994):

Z ��

jxlr� �jxe�� s

�1�

where j � ()1)1/2, x � angular frequency (Hz) �2pf, l � magnetic permeability (Henry m)1) � l0lr,r � electrical conductivity (S m)1), e � dielectricpermittivity (Farad m)1) � e0er.

The relative permittivity (er), which is mainlycontrolled by water content, is the most importantparameter governing the re¯ection process andwave velocity. When a signi®cant change inrelative permittivity is encountered, part of the

electromagnetic energy is re¯ected, the re¯ectionbeing proportional to the magnitude of change.For most materials, the relative magnetic per-meability (lr) is near unity. Consequently, themagnetic permeability in the subsurface is nearthe free-space value (lo) of 4p * 10)7 Henry m)1

(Von Hippel, 1954; Powers, 1997) and plays norole in the electromagnetic energy behaviour.However, under certain conditions, such as thepresence of iron and iron oxides, relative mag-netic permeability can be enhanced signi®cantly(Von Hippel, 1954; Olhoeft & Capron, 1994). Theelectrical conductivity of a material in¯uencespenetration depth as well as resolution. Low-conductivity materials, such as unsaturated andcoarse-grained sediments, cause little attenuationand, under ideal circumstances, penetration is ofthe order of tens of metres (Davis & Annan, 1989).However, wave velocity and length are highest inlow-conductivity materials, leading to a decreasein resolution (Table 1). Penetration depth andresolution are also in¯uenced by the GPR fre-quency used for measurement. Lower antennafrequencies are favourable for greater penetration,but result in a decrease in resolution. Resolutionis approximately a quarter of the GPR wavelength,and ranges from 0á08 m for saturated sands and200-MHz antennas to 0á4 m for dry sands and100-MHz antennas (Table 1).

Time-domain re¯ectometry

The TDR method is based on the propagationvelocity of an electromagnetic signal along asediment probe and was developed to character-ize the water content of soils using the relativepermittivity (Topp et al., 1980). The propagationvelocity can be calculated by:

m � c0��erlrp �2�

where c0 is the electromagnetic wave velocity in avacuum (3 * 108 m s)1). The product of the rela-tive permittivity and the relative magnetic per-meability can thus be calculated from the traveltime (Dts) of the TDR signal and the length (L) ofthe probe (Roth et al., 1990):

Table 1. GPR set up and wave parameters for the pulseEkko100 system.

Frequencies(MHz)

Antennaseparation (m)

Pulservoltage (V)

Wavelength indry sand (m)

Resolution(m)

Wavelength insaturated sand (m)

Resolution(m)

100 1á0 400 1á5 0á4 0á6 0á15200 0á5 400 0á75 0á2 0á3 0á08

Fig. 1. (A) Ground-penetrating radar set up and methodand (B) resulting radar section.

Identifying causes of GPR re¯ections 437


erlr �c0Dts

2L

� �2

�3�

For most sediments, the relative magnetic per-meability (lr) is near unity (Roth et al., 1990).Factors other than water content, such as soildensity, texture and temperature, have a negligi-ble in¯uence on the relative permittivity (Toppet al., 1980). Using these assumptions, the volu-metric water content (h) is found by substitutionof er in the empirical relationship (Topp et al.,1980):

h � �ÿ5�3 � 10ÿ2� � ��2�92 � 10ÿ2� � �er��ÿ ��5�5 � 10ÿ4� � �er�2� � ��4�3 � 10ÿ6� � �er�3�

�4�

Early laboratory studies (e.g. Fellner-Feldegg,1969) used soil columns in coaxial transmissionlines, while ®eld probes were later developedand improved (Dalton et al., 1984; Zegelin et al.,1989; Brisco et al., 1992; Heimovaara, 1993). Inthis study, the TDR equipment developed byHeimovaara & Bouten (1990) was used (Fig. 2A).A 0á05-m-long, three-rod probe, connected to aTektronix Cabletester and laptop computer forsystem control and data storage, was pushedinto the soil. The cable tester transmits a fast-rise voltage pulse through the transmission lineand probe. The frequency band ranges from300 kHz to 3 GHz (Heimovaara et al., 1996),which encompasses all GPR frequencies. Atchanges in electrical properties, part of thepulse is re¯ected to the cable tester. After

calibration for the epoxy casing, the two re¯ec-tion points of the resulting wave form (Fig. 2B)give the travel time along the rods (Dts), neces-sary to obtain erlr (Eq. 3).

Sedimentological and textural characteristics

Sedimentary structures were studied both directlyin the ®eld and using lacquer peels and thinsections. The lacquer peels were made to studysedimentary structures and parameters macro-scopically. On these lacquer peels, sedimentarystructures or bedding planes appear as narrowridges or lows, owing to different grain and poresizes. Other elements, such as organic materialand iron oxide, appear as (root) relicts or as colourchanges on the lacquer peel. Thin sections, madefrom undisturbed sediment samples, were used tostudy sediment properties microscopically. Twosamples were taken for grain-size analyses.

Synthetic radar traces

Using TDR and sedimentological data, dielectricimpedance models of the subsurface can beconstructed. The models, in which values fordepth and erlr are given, form the input forpulseEKKO software (Sensors & Software 1996)that construct synthetic one-dimensional GPRtraces. The program assumes vertically incidentelectromagnetic waves and calculates all generatedre¯ections, including multiples and interlayerre¯ections. First, the program transforms theimpedance model from a depth scale into a timescale. Then, the impulse response for the layeredmodel is computed, and a correction for

Fig. 2. (A) Time-domain re¯ecto-metry equipment and (B) resultingwave form.



spherical-waveform spreading losses is applied.The ground response is obtained by convolutionof a standard pulseEKKO wavelet (both 100 and200 MHz) with the impulse response. Attenua-tion differences were not incorporated in themodel. To ensure that small re¯ections were alsovisualized, the one-dimensional synthetic traceswere plotted with an automatic gain control(AGC), applying a gain inversely proportional tothe signal strength. A gain limit was applied toprevent very small signals from producing verylarge gains.

SEDIMENTOLOGY OF THE STUDY SITE

Fieldwork was conducted in the Boudewijnquarry in Ossendrecht (Fig. 3). This 15-m-deepquarry can be subdivided into two units: a lowerunit of tidal deposits and an upper unit of aeoliandeposits.

The lower unit has a thickness of »7á5 m andconsists of tidal sediments (Tegelen Formation)deposited during several Tiglian interglacials ofEarly Pleistocene age (Kasse, 1988; Fig. 4). Anunconformity, possibly formed during the regres-sion associated with the Beerse Glacial, separatesthe two subunits. The lower subunit (Hooger-heide Member) consists of large-scale cross-bedded

very ®ne to ®ne sand. Locally, numerous smallripple marks with clay drapes, resulting fromfrequent ¯ow reversals, are present (Kasse, 1988).The upper subunit (Woensdrecht Member) con-sists of large-scale cross-bedded very ®ne tomedium sand. The top of this subunit is alaterally continuous clay layer with a maximumthickness of 3 m (Kasse, 1988). At the end of theTiglian interglacial, the climate became cooler,and the sea level fell. As a result, a substantialhiatus, locally marked by a coarse-grained resid-ual deposit, is present on top of the Woensdrechtmember at the base of the upper aeolian unit. Thishiatus represents an extended time of erosion andnon-deposition (Kasse, 1988).

The upper quarry unit has a maximum thick-ness of 7á5 m and consists of aeolian sands of thePleistocene Twente and Holocene Kootwijk For-mations (Kasse, 1988). These formations areseparated by a peat layer of the GriendtsveenFormation in blowout hollows or by a palaeosolof podsol type in adjacent higher areas (Fig. 4).The sand from the Twente Formation dates fromthe Late Dryas Stadial to the Early Holocene(Schwan, 1991). Locally, a thin soil or peat layerfrom the Allerùd Interstadial is present near thebottom of this subunit. The sand from the TwenteFormation originates from the banks of the riverScheldt, which ¯owed just to the west of thestudy area (Fig. 3). This sand was deposited assand sheets and dunes and has a large- tomedium-scale cross-bedded character. The prom-inent peat bed and palaeosol separating theTwente and Kootwijk Formations represents mostof the Early and Middle Holocene, a timespan of»6000 years (Schwan, 1991). Deposition of thesand from the Late Holocene Kootwijk Formationstarted as early as 3000 years ago through thebuild-up of small dunes (Schwan, 1991). Inplaces, lamination has been obliterated by abun-dant roots that can be assigned to several discon-tinuous soil levels.

The steplike excavation of the quarry wouldhave allowed investigation of all units, from boththe top and the side wall. In this study, a sectionin the aeolian unit was chosen for detailedanalysis (Fig. 5) because: (1) it is made up ofhighly resistive sand; (2) the palaeosol betweenthe two aeolian subunits and the transition to theTegelen clay at the base of the aeolian sand wereexpected to give clear GPR re¯ections and pro-vide good reference points in the radar sections;(3) the variety of sedimentary structures and theirpossible re¯ections were considered to be importantin establishing which features cause re¯ection of

Fig. 3. Map of The Netherlands showing location ofstudy area.



GPR waves; and (4) the terrain at this site waseasily accessible and even.

The schematic representation of sedimentaryand pedogenic structures in the studied section(Fig. 5B) re¯ects the general strati®cation for theupper quarry unit given above. In the sectionstudied, the Allerùd soil is absent. Three inor-ganic facies, deposited on a gradually dryingdepositional surface (Schwan, 1991), comprisethe Twente Formation: (1) a basal subhorizontallybedded sand-sheet facies; (2) a gently dippingsand-sheet facies; and (3) a cross-strati®ed dune

facies. Sand sheet facies B, at the base of theformation, is characterized by alternation ofmedium sand and ®ner grained beds. This facies

Fig. 4. Schematic stratigraphiccolumn for quarry Boudewijn,Ossendrecht. Data from Kasse(1988), Schwan (1991) and thepresent study. GF, GriendtsveenFormation; EMH, Early and MiddleHolocene; Ad, Allerùd.

Fig. 5. (A) Picture of section wall showing the loca-tions of lacquer peels and TDR sections. (B) Schematicrepresentation of sedimentary and pedogenic struc-tures. In the Twente Formation, left of lacquer peelOSL02, several diagenetic iron oxide bands that cross-cut depositional bedding are problematic for the inter-pretation of sedimentary structures.



re¯ects a varying wind regime, a damp deposi-tional surface (caused by the underlying semi-pervious Tegelen Formation) and the availabilityof both sand and silt in the source area. Sandsheet facies A, located above the basal facies, ischaracterized by metre-scale gently dipping sigm-oidal strata, with a maximum dip angle of 140°/13°. The dune facies at the top of the formation isseparated from sand sheet facies A by a coarse-grained bounding surface. The dune facies stratahave a maximum dip angle of 350°/37°. However,the strike of the strata show considerablevariability, in line with the inferred parabolicshape of the dunes. The top of this cross-strati®edsand is marked by a 0á3-m-thick palaeopodsol(Palaeosol B1, Fig. 5B; hidden by a small ledgein Fig. 5A). Except for some faint horizontallamination, sedimentary structures in the overly-ing sand from the Kootwijk Formation have beenobliterated by several soil levels (e.g. PalaeosolB2, Fig. 5B).

RESULTS

Ground-penetrating radar

GPR data were sampled along a line that waslocated 5 m from the section wall (Fig. 6) and hada length of 35 m. The line was sampled with twofrequencies (Table 2). On the 100-MHz radarsection (Fig. 7A), some of the largest features of

the unit are visible. At a depth of approximately6 m, two gently left-dipping re¯ections representthe boundary between the aeolian sands and thetidal clays. These re¯ections become discontinu-ous and less pronounced towards the left, whichmay be a result of increasing re¯ector depth or adecreasing change in grain size or water contentat this re¯ector. Below the boundary, the electro-magnetic signal is attenuated quickly and, hence,no deeper re¯ectors can be seen (1 in Fig. 7A).The subhorizontal re¯ections above the sand±clay transition (2 in Fig. 7A) originate in sandsheet facies B (Fig. 5B). Above this GPR facies,numerous discontinuous and mostly right-dip-ping re¯ections (3 in Fig. 7A) mark sand sheetfacies A (Fig. 5B). The overlying left-dippingcross-strati®ed dune facies (Fig. 5B) is presentonly in the left part of the interpreted section (4 inFig. 7A). This might be caused by the overlyingpalaeosol re¯ection, which may keep the dunecross-strati®cation re¯ections in its shadow. Thispalaeosol between the Twente and Kootwijkformations (Fig. 5B) is delineated by the contin-uous, dipping re¯ection at an approximate depthof 2 m (5 in Fig. 7A). Towards the right, thisre¯ection merges with the two horizontal re¯ec-tions at the top of the radar section (6 in Fig. 7A),which represent the direct airwaves and ground-waves.

On the 200-MHz radar section (Fig. 7B), moredetail is visible, especially in the 2- to 4-m depthrange. However, depth of penetration hasdecreased, and the sand±clay transition is onlybarely visible in the lower right of the pro®le (1and 2 in Fig. 7B). The re¯ections representingsand sheet facies A (3 in Fig. 7B) generally dip tothe right as expected (Fig. 5B). Left-dippingre¯ections (4 in Fig. 7B) mark the overlyingcross-bedded dune facies (Fig. 5B), but arepresent only in the left part of the radar section.As the radar section is not an exact representationof the section wall, because it is located 5 m away(Fig. 6), this under-representation of left-dippingre¯ections is attributed to the lateral variability ofthe facies, in both thickness and strike of thestrata. Even across this small horizontal distance,the thickness of the dune facies may have

Fig. 6. Sketch to illustrate location of measurements inthe Boudewijn quarry.

Table 2. Characteristics and locations of GPR measurements.

LineDistanceto wall (m)

Frequency(MHz) Mode X0 (m) Xmax (m) Xtot (m) DX (m)

OSR01 5 100, 200 Re¯ection 0 35 35 0á25100, 200 CMP 11 ± 10, 7á8 0á2



decreased to below the resolution of the GPRwaves. The palaeosol at the base of the KootwijkFormation (Fig. 5B) is represented by the re¯ec-tion at a depth of almost 2 m (5 in Fig. 7B). There¯ections between the palaeosol and the twodirect waves (6 in Fig. 7B) may represent either(horizontal) bedding or one of the discontinuoussoil horizons B2 (Fig. 5B).

Textural characteristics

Laser grain-size analyses (Konert & Vandenber-ghe, 1997) and thin sections (from lacquer peelOSL01) provide information on sediment charac-teristics. The sandy intervals consist of ®ne to

medium sand (229á1±287á5 lm) that is subroundedand moderately to well sorted. Pore space occa-sionally exceeds maximum grain size, indicatingrooting or burrowing. In the palaeosol, a largeamount of organic material (»5% of total sedi-ment volume) occupies part of the pore space.

Time-domain re¯ectometry and lacquer peels

Lacquer peels were taken from representativeparts of the wall covering the full height of thesection (Table 3), and then TDR data (Table 4)and sediment samples (Table 5) were collectedfrom these transects (Fig. 5). Comparison of TDRdiagrams with associated lacquer peels (Fig. 8)

Fig. 7. Radar sections for line OSR01 (left) and interpretation (right). (A) 100 MHz. (B) 200 MHz. The outlined partcorresponds with the interpreted section shown in Fig. 5. CMPs indicate that the electromagnetic wave velocity was0á11 m ns)1 in the upper highly resistive subunit, and averaged 0á09 m ns)1 for the total unit. Radar sections for lineOSR01 are printed with a velocity of 0á09 m ns)1. As this velocity is an average for the entire unit, the plot isstretched and compressed in its higher and lower parts respectively. Average vertical exaggeration is 1á4. Thenumbered labels in the interpreted radar sections refer to the text and represent: 1, tidal clays; 2, sand sheet facies B;3, sand sheet facies A; 4, dune facies; 5, palaeosol; 6, overlapping airwave and groundwave; 7, re¯ections dipping inthe opposite direction to the sedimentary structures; and 8, partly overlapping wave forms.



illustrates several relationships between sedi-ment characteristics and dielectric properties.The product erlr (Eq. 3) increases from a baselinevalue of about 4 in the upper part of the section toalmost 25 in its lower part. The baseline value of4 represents the relative permittivity of dry sand,whereas the value of 25 is characteristic of therelative permittivity of water-saturated sand(Davis & Annan, 1989; Wensink, 1993), bothhaving a relative magnetic permeability of unity.According to Eq. 4, these values for er correspondto volumetric water contents of 0á055 and 0á4respectively. Numerous excursions in erlr aresuperimposed on the general trend, and most ofthese can be linked to features recorded in thelacquer peels.

The palaeosol at a depth of 1á8 m gives anincrease in erlr to about 8 (Fig. 8A). This is causedby the fact that humic material, present in thispalaeosol, holds water, which increases relativepermittivity. Figure 8B shows a similar responseto humic material when two individual cross-sets, with windblown humic material, cause therelative permittivity to increase. In the lower partof Fig. 8B and the upper part of Fig. 8C, severaliron oxide bands, identi®ed in the lacquer peels,result in sharp increases in erlr. Most of these ironbands follow depositional bedding, but some,such as the band at a depth of 3á4 m (Fig. 8B),cross-cut the bedding, indicating a diageneticorigin of the iron bands. As the erlr peakstypically occur directly on top of these cementedand less permeable layers, they may be caused bystagnating water occupying pore space, resultingin an increase in er. Alternatively, an increase in

the relative magnetic permeability, caused by thepresence of iron oxide minerals, may bring aboutthe observed peaks in erlr. Both causes seem toplay a role in this case. The iron oxide-relatedexcursions in the TDR log associated with lacquerpeel OSL02 (Fig. 8B) are superimposed on abaseline value for erlr of about 4. This suggeststhat the absence of any moisture and changes inrelative magnetic permeability are the main caus-es of these excursions. In the TDR log associatedwith lacquer peel OSL03 (Fig. 8C), the amplitudeof the iron oxide-related excursions is higher.This suggests that, as water saturation increasesdownwards in lacquer peel OSL03, stagnatingwater and thus relative permittivity contributes tothe increase in erlr, in addition to enhancedrelative magnetic permeability.

In the zone in which average water contentremains low and changes in erlr are caused byeither organic material or iron oxide (Fig. 8A andB), it is noteworthy that none of these changes canbe related directly to grain size. In lacquer peelOSL02, large-scale cross-bedding is obvious, butnone of the bedding planes cause erlr to changesigni®cantly, except where humic material or ironis present. For example, the gravel layer at 2á85 mdepth does not result in any change in erlr in theTDR log (Fig. 8B). In contrast, erlr ¯uctuatessigni®cantly where water saturation increases(Fig. 8C). In this case, ¯uctuations that coincidewith the presence of iron oxide bands as well assedimentary structures are superimposed on ageneral increase in erlr with depth. Here, ®ne-grained layers from sand sheet facies B (Fig. 5)enhance capillary forces and therefore support

Table 4. Characteristics and locations of TDR measurements.

No.TDR section(Fig. 5) Y0 (m) Ymax (m) Ytot (m) DY (m)

Probelength (m) X (m) Location in grid

z970605.001 1 0á8 1á95 1á15 0á05 0á1 12á5 Lacquer peel OSL01z970605.002 1 0á8 2á0 1á2 0á05 0á05 12á5 Lacquer peel OSL01z970606.003 1 1á6 2á3 0á7 0á025 0á05 12á5 Lacquer peel OSL01z970606.004 2 2á1 4á7 2á6 0á05 0á05 11 Lacquer peel OSL02z970606.005 3 3á7 6á0 2á3 0á05 0á05 4á5 Lacquer peel OSL03

Table 5. Locations of thin section samples.

Name Y (m) X (m) Location in grid

OSS1 0á6 12á5 Lacquer peel OSL01OSS2 1á6 12á5 Lacquer peel OSL01OSS3 1á6 12á5 Lacquer peel OSL01

Table 3. Locations and size of lacquer peels.

Name Ytop (m) Ybot (m) Ytot (m) X (m)

OSL01 0 2á3 2á3 12á5OSL02 2á1 4á7 2á6 11OSL03 3á7 5á6 1á9 4á5



changes in relative permittivity. Thus, GPRre¯ections from sedimentary structures, not relat-ed to mineralogical change, can be expected onlywhen moisture is present in the sediment andvalues for relative permittivity and volumetricwater content exceed the baseline values of 4 and0á055 respectively.

Synthetic radar traces

Using the three TDR logs (Fig. 8), one log cover-ing the full section was compiled. From thiscomposite TDR log and the lacquer peels, givinginformation on erlr and sediment characteristics,respectively, four dielectric impedance modelswere constructed, each focusing on a particularcharacteristic of the section (Fig. 9). Using thefour impedance models, one-dimensional syn-thetic GPR traces were constructed. The layeredimpedance models have a depth scale, whereasthe synthetic traces have a time scale. As theproduct of relative permittivity and relative mag-netic permeability, and thus the wave velocity,

varies over depth, the relative thickness of thelayers also varies.

Figure 9A shows the graph of erlr vs. depth thatwas used as input for the different impedancemodels. Figure 9B represents the simplest situa-tion with a basic value of 4 and two excursions:the palaeosol and the increase in erlr above thesand±clay transition. In the synthetic radar traces,all layer boundaries appear as radar re¯ections,and there are also multiples of earlier re¯ections(indicated by arrows). The situation in Fig. 9Crepresents the slightly more complex palaeosol(as found in the TDR logs) and a stepped increasein erlr towards the sand±clay transition. The thinlayer, with an erlr value of 4, within the palaeosolleads to two overlapping waveforms (especiallyin the 100-MHz synthetic trace) and to somemultiples in the lower part of the synthetic radartraces (arrowed). Figure 9D focuses on the smallexcursions in the TDR logs caused by the ironoxide bands. In the 200-MHz synthetic trace, allevents can be distinguished but, in the 100-MHzsynthetic trace, overlapping waveforms produce a

Fig. 8. Lacquer peels and associated TDR measurements. (A) TDR section 1 and lacquer peel OSL01. (B) TDR section2 and lacquer peel OSL02. (C) TDR section 3 and lacquer peel OSL03. Lacquer peels are printed in mirror image, suchthat the structures are shown in their true orientation.



complicated pattern. The sediment model com-bining Fig. 9C and D leads to highly complicatedsynthetic radar traces with overlapping wave-forms and multiples (Fig. 9E).

DISCUSSION AND CONCLUSION

Various reasons exist for changes in erlr and, thus,for radar re¯ections, but changes in relativepermittivity are the most important factor. Therelative permittivity is predominantly controlledby water content, which is a result of the large er

contrast between air and water. Relative permit-tivity is 1 for air and 80 for water, whereas it is 4±6 for most sedimentary minerals (Davis & Annan,1989). The volume fractions of the differentconstituents that form the sediment determineer. Consequently, the sediment porosity as well asthe ability of the sediment to hold water areimportant factors controlling the relative permit-tivity (Knoll & Knight, 1994). When water contentis high, such as in the lower part of the section(Fig. 8C), higher porosities cause both watercontent and relative permittivity to increase.When water content is low, such as in the higherpart of the section (Fig. 8A and B), the relativepermittivity is controlled by the ability of thesediment to retain water. Both organic material,which holds water, and ®ne-grained sediment,which enhances capillary forces, cause watercontent as well as relative permittivity toincrease. The role of relative magnetic permeabil-ity in the GPR re¯ection process is less clear. VonHippel (1954) reported a relative magnetic per-meability of 1á09 for a magnetite-bearing sedi-ment, which would result in an increase in erlr

from 4 to 4á4 for a dry sandy soil. This result is notsupported by data from the present study, inwhich iron oxide-related excursions in erlr from 4to about 6 were found (Fig. 8B). The requiredrelative magnetic permeability values of 1á5 havenot been reported in the literature. This suggeststhat a combination of iron oxides, enhancing lr,and water, enhancing er, cause the peaks in erlr.

In the zone with a baseline value of 4 for erlr,none of the changes can be related directly tograin size (Fig. 8A and B). Thus, when volumetricwater content does not exceed 0á055, grain size isnot important as a cause of GPR re¯ections.However, numerous re¯ections in the zone justbelow the palaeosol re¯ect sedimentary beddingof sand sheet facies A (3 in Fig. 7) and the dunefacies (4 in Fig. 7). Both windblown humicmaterial and iron oxide bands increase erlr andcould cause these re¯ections. Some re¯ections,however, dip in the opposite direction to thesedimentary structures. For example, this isshown by the re¯ection at position 9 m at 4 mdepth in the 100-MHz section (7 in Fig. 7A) andthe re¯ection at position 15 m at 3 m depth in the200-MHz section (7 in Fig. 7B). These re¯ectionsare most probably caused by iron oxide bands thatcross-cut sedimentary bedding.

The TDR logs could be used to calculate detailedsynthetic radar traces. Owing to the non-verticalquarry wall, the depth in the dielectric impedancemodel, and consequently the time in the syntheticGPR traces, is overestimated by about 20%.Nevertheless, the synthetic traces (Fig. 9) can becompared with the actual GPR sections (Fig. 7).Waveforms that overlap in part in the synthetictraces (Fig. 9B), because of the complex palaeosol,are also present in the actual GPR sections. This is

Fig. 9. (A) Composite TDR log for the complete quarry section. Layered dielectric impedance models, constructedfrom this log, and resulting one-dimensional synthetic radar traces; (B) simple, (C) soil, (D) iron and (E) combination.Arrows indicate multiples. Note that vertical scales are in depth for the TDR log and layered sediment models and intime for the synthetic radar traces.



well exempli®ed in the 200 MHz section betweenposition 10 and 20 m at a depth of 1±1á5 m (8 inFig. 7B). In Fig. 9D, the small excursions in theTDR logs caused by the iron oxide bands weremodelled. This zone is dif®cult to compare withthe actual radar sections, as the variability inthickness and intensity of the iron oxide bands ishigh, and the changes in erlr are small. Neverthe-less, the re¯ection strengths in the synthetic tracesand actual GPR sections (especially for sand sheetfacies A) match quite well and are smaller than thedune facies above and sand sheet facies B below.The increase in water content towards the sand±clay transition was modelled in the synthetictraces using a three-step increase in erlr (Fig. 9C).This resulted in three re¯ection events in thesynthetic traces, such as that observed in theactual 100-MHz radar section (Fig. 7A). However,in the 200-MHz section (Fig. 7B), no clear re¯ec-tions are visible from the sand±clay transition.This discrepancy might be caused by the fact thatno attenuation was incorporated in the calculationof the synthetic traces. Nevertheless, syntheticGPR traces can help in understanding the variousre¯ections in actual GPR sections.

Not all GPR re¯ections can be matched withevents on TDR logs, lacquer peels and one-dimen-sional synthetic images. The internal lateralvariability of the facies is the main problem incorrelating these data sets. Within the 5 m distancefrom the GPR sections to the quarry face, both thegeometry and the depth of the facies, as well as thewater content, could have changed laterally. Inspite of the above, this study shows that thecombination of TDR and sedimentological analy-sis is a very useful technique for qualifying andquantifying erlr in the subsurface and achieving abetter insight into the origin of GPR re¯ections.

ACKNOWLEDGEMENTS

The Netherlands Institute of Applied Geoscience(NITG-TNO) and the Physical Geography Depart-ment of the University of Amsterdam are thankedfor kind permission to use their equipment. Thelate Thom Roep guided us to the outcrop and wasinstrumental in conducting the ®eldwork. Wealso gratefully acknowledge the help of Fabricio`scalebar' De Jonge and Kim Cohen (UtrechtUniversity), Evert Slob (Delft University of Tech-nology) and Albrecht Weerts and Stefan Dekker(University of Amsterdam) during the ®eldwork.The work was supported ®nancially by NITG-TNO and the Vrije Universiteit Industrial Asso-

ciates in Sedimentology. Useful comments weremade by Mark Dekkers (Utrecht University),Adrian Immenhauser and Kees Kasse (Vrije Uni-versiteit) and Jan Van Der Kruk (Delft Universityof Technology). The manuscript bene®ted fromcritical reviews by Joep Storms (Delft Universityof Technology) and Sytze Van Heteren (VrijeUniversiteit) on an earlier version and by SteveCardimona (University of Missouri-Rolla) andPeter Huggenberger (University of Basel).

LIST OF SYMBOLS

c0 Electromagnetic wave velocity in vacuum(ms)1)

f Frequency (Hz)j ()1)1/2

L Length of the TDR probe (m)t Time (s)m Propagation velocity of electromagnetic

signal (ms)1)X Horizontal position (m)X0 Horizontal position of ®rst measurement

(m)Xmax Horizontal position of ®nal measurement

(m)Xtot Total length of line (m)Y Vertical position on quarry face (m)Y0 Vertical position of ®rst measurement (m)Ymax Vertical position of ®nal measurement (m)Ybot Vertical position of bottom of lacquer peel

(m)Ytop Vertical position of top of lacquer peel (m)Ytot Total height (m)Z Dielectric impedanceDX Step size (m)DY Sampling interval (m)e Dielectric permittivity (Farad m)1)e0 Dielectric permittivity of free space

(Farad m)1)er Relative permittivityh Volumetric water contentl Magnetic permeability (Henry m)1)l0 Magnetic permeability of free space

(Henry m)1)lr Relative magnetic permeabilityr Electrical conductivity (Sm)1)x Angular frequency (Hz)

REFERENCES

Arcone, S.A. (1984) Field observations of electromagneticpulse propagation in dielectric slabs. Geophysics, 49,1763±1773.



Asprion, U. and Aigner, T. (1997) Aquifer architecture

analysis using ground-penetrating radar: Triassic and

Quaternary examples (S. Germany). Environ. Geol., 31,66±75.

Beres, M., Green, A., Huggenberger, P. and Horstmeyer, H.(1995) Mapping the architecture of glacio¯uvial sediments

with three-dimensional georadar. Geology, 23, 1087±1090.

Bracco Gartner, G.L. and Schlager, W. (1999) Discriminatingbetween onlap and lithologic inter®ngering in seismic

models of outcrops. AAPG Bull., 83, 952±971.

Brewster, M.L. and Annan, A.P. (1994) Ground-penetratingradar monitoring of a controlled DNAPL release: 200 MHz

radar. Geophysics, 59, 1211±1221.

Bridge, J.S., Collier, R.E.L. and Alexander, J. (1998) Large-

scale structure of Calamus River deposits (Nebraska, USA)revealed using ground-penetrating radar. Sedimentology,

45, 977±986.

Brisco, B., Pultz, T.J., Brown, R.J., Topp, G.C., Hares, M.A.and Zebchuck, W.D. (1992) Soil moisture measurementusing portable dielectric probes and time domain re¯ec-

tometry. Water Resources Res., 28, 1339±1346.

Bristow, C., Pugh, J. and Goodall, T. (1996) Internal structureof aeolian dunes in Abu Dhabi determined using ground-

penetrating radar. Sedimentology, 43, 995±1003.

Carcione, J.M. (1996) Ground-penetrating radar: wave theory

and numerical simulation in lossy anisotropic media.Geophysics, 61, 1664±1677.

Cardimona, S.J., Clement, W.P. and Kadinsky-Cade, K. (1998)

Seismic re¯ection and ground-penetrating radar imaging of

a shallow aquifer. Geophysics, 63, 1310±1317.Casper, D.A. and Kung, K.-J.S. (1996) Simulation of ground-

penetrating radar waves in a 2-D soil model. Geophysics, 61,1034±1049.

Dalton, F.N., Herkelrath, W.N., Rawlins, D.S. and Rhoades,J.D. (1984) Time-domain re¯ectometry: Simultaneous meas-

urement of soil water content and electrical conductivity

with a single probe. Science, 224, 989±990.Davis, J.L. and Annan, A.P. (1989) Ground-penetrating radar

for high resolution mapping of soil and rock stratigraphy.

Geophys. Prospecting, 37, 531±551.

Fagin, S.W. (1991) Seismic Modeling of Geologic Structures ±Applications to Exploration Problems. Society Exploration

Geophys., Tulsa, 267 pp.

Fellner-Feldegg, H. (1969) The measurement of dielectrics in

the time domain. J. Phys. Chem., 73, 616±623.Gawthorpe, R.L., Collier, R.E.L., Alexander, J., Leeder, M.R.

and Bridge, J.S. (1993) Ground penetrating radar: applica-

tion to sandbody geometry and heterogeneity studies. In:Characterisation of Fluvial and Aeolian Reservoirs (Eds C.P.

North and D.J. Prosser), Spec. Publ. Geol. Soc. London, 73,421±432.

Heimovaara, T.J. (1993) Design of triple-wire time domainre¯ectometry probes in practice and theory. Soil Sci. Soc.Am. J., 57, 1410±1417.

Heimovaara, T.J. and Bouten, W. (1990) A computer-con-

trolled 36-channel time domain re¯ectometry system formonitoring soil water contents. Water Resources Res., 26,2311±2316.

Heimovaara, T.J., Focke, A.G., Bouten, W. and Verstraten,J.M. (1995) Assessing temporal variations in soil water

composition with time domain re¯ectometry. Soil Sci. Soc.Am. J., 59, 689±698.

Heimovaara, T.J., De Winter, E.J.G., Van Loon, W.K.P. andEsveld, D.C. (1996) Frequency-dependent dielectric per-

mittivity from 0 to 1 GHz: time domain re¯ectometry meas-

urements compared with frequency domain network ana-

lyzer measurements. Water Resources Res., 32, 3603±3610.

Hollender, F. and Tillard, S. (1998) Modeling ground-pen-

etrating radar wave propagation and re¯ection with theJonscher parameterization. Geophysics, 63, 1933±1942.

Huggenberger, P. (1993) Radar facies: recognition of facies

patterns and heterogeneities within Pleistocene Rhine

gravels, NE Switzerland. In: Braided Rivers (Eds J.L. Bestand C.S. Bristow), Geol. Soc. Spec. Publ., 75, 163±176.

Jol, H.M. and Smith, D.G. (1991) Ground penetrating radar of

northern lacustrine deltas. Can. J. Earth Sci., 28, 1939±1947.Kasse, K. (1988) Early-Pleistocene Tidal and Fluviatile Envi-

ronments in the Southern Netherlands and Northern Bel-gium. PhD Thesis, Vrije Universiteit Amsterdam, The

Netherlands.Kenter, J.A.M., Van Hoe¯aken, F., Bahamonde, F., Bracco

Gartner, G.L. and Besems, R.E. (1999) Anatomy and

lithofacies of an intact and seismic-scale Carboniferous

carbonate platform (Asturias, NW Spain): analogues ofhydrocarbon reservoirs in the Pricaspian basin (Kazakh-

stan). SEPM Spec. Vol., (in press).

Knoll, M.D. and Knight, R. (1994) Relationships betweendielectric and hydrogeologic properties of sand-clay mix-

tures. In: GPR ¢94 ± Proceedings of the 5th InternationalConference on Ground Penetrating Radar (Eds J.D. Redman,

A.P. Annan, J.P. Greenhouse, T. Klym and J.R. Rossiter), 1,45±61. Waterloo Center for Groundwater Research, Kitch-

ener, Ontario.

Konert, M. and Vandenberghe, J. (1997) Comparison of laser

grain size analysis with pipette and sieve analysis: a solu-tion for the underestimation of the clay fraction. Sedimen-tology, 44, 523±535.

Mayer, L.A. (1979) Deep sea carbonates: acoustic, physicaland stratigraphic properties. J. Sedim. Petrol., 49, 819±836.

Mayer, L.A. (1980) Deep-sea carbonates: physical property

relationships and the origin of high-frequency acoustic

re¯ectors. Mar. Geol., 38, 165±183.Olhoeft, G.R. and Capron, D.E. (1994) Petrophysical causes of

electromagnetic dispersion. In: GPR ¢94 ± Proceedings of the5th International Conference on Ground Penetrating Radar(Eds J.D. Redman, A.P. Annan, J.P. Greenhouse, T. Klymand J.R. Rossiter), 1, 145±152. Waterloo Center for

Groundwater Research, Kitchener, Ontario.

Powers, M.H. (1997) Modeling frequency-dependent GPR. TheLeading Edge, 16, 1657±1662.

Rea, J. and Knight, R. (1998) Geostatistical analysis of ground-

penetrating radar data: a means of describing spatial varia-

tion in the subsurface. Water Resources Res., 34, 329±339.Roth, K., Schulin, R., FluÈhler, H. and Attinger, W. (1990)

Calibration of time domain re¯ectometry for water content

measurement using a composite dielectric approach. WaterResources Res., 26, 2267±2273.

Schwan, J. (1991) Palaeowetness indicators in a Weichselian

Late Glacial to Holocene aeolian succession in the south-

western Netherlands. Z. Geomorph. N.F., (Suppl.)-Bd. 90,155±169.

Sensors&Software. (1996) Pulse EKKO Synthetic RadargramUser's Guide. Sensors&Software, Mississauga, ON, Canada.

Sta¯eu, J. and Schlager, W. (1995) Pseudo-unconformities inseismic models of large outcrops. Geol. Rundschau, 84,761±769.

Sta¯eu, J. and Sonnenfeld, M.D. (1994) Seismic models of a

shelf-margin depositional sequence: upper San AndresFormation, Last Chance Canyon, New Mexico. J. Sedim.Res., 64, 481±499.



Sutinen, R. (1992) Glacial deposits, their electrical properties

and surveying by image interpretation and ground pen-

etrating radar. Geol. Surv. Finland Bull., 359, 1±123.

Topp, G.C., Davis, J.L. and Annan, A.P. (1980) Electromagneticdetermination of soil water content: measurements in coaxial

transmission lines. Water Resources Res., 16, 574±582.

Topp, G.C., Davis, J.L. and Annan, A.P. (1982) Electromag-

netic determination of soil water content using TDR: I.applications to wetting fronts and steep gradients. Soil Sci.Soc. Am. J., 46, 672±678.

Van Heteren, S., Fitzgerald, D.M., McKinlay, P.A. and Buy-nevich, I.V. (1998) Radar facies of paraglacial barrier sys-

tems: coastal New England, USA. Sedimentology, 45,181±200.

Van Overmeeren, R.A. (1998) Radar facies of unconsolidatedsediments in the Netherlands: a radar stratigraphy inter-

pretation method for hydrogeology. J. Appl. Geophys., 40,1±18.

Von Hippel, A.R. (1954) Dielectric Materials and Applications.

John Wiley and Sons, New York.Wensink, W.A. (1993) Dielectric properties of wet soils in the

frequency range 1±3000 MHz. Geophys. Prospecting, 41,671±696.

Zegelin, S.J., White, I. and Jenkins, D.R. (1989) Improved ®eldprobes for soil water content and electrical conductivity

measurement using time domain re¯ectometry. WaterResources Res., 25, 2367±2376.

Manuscript received 22 February 1999;

revision accepted 15 September 1999.



identifying causes of ground-penetrating radar reﬂections ...rvd/pub/sed_2000.pdf · time-domain...

Documents