college of engineering - high-resolution characterization of … · 2001. 11. 13. · spaced at...

High-resolution characterization of vadose zone dynamicsusing cross-borehole radar

Andrew Binley, Peter Winship, and Roy MiddletonDepartment of Environmental Science, Lancaster University, Lancaster, England, UK

Magdeline Pokar and Jared WestSchool of Earth Sciences, University of Leeds, Leeds, England, UK

Abstract. Characterization of the dynamics of moisture migration in the unsaturatedzone of aquifers is essential if reliable estimates of the transport of pollutants threateningsuch aquifers are to be made. Electrical geophysical investigation techniques, such asground-penetrating radar, offer suitable methods for monitoring moisture content changesin the vadose zone. Moreover, these tools permit relatively large measurement scales,appropriate for hydrological models of unsaturated processes, and thus they offer adistinct advantage over conventional measurement approaches. Ground-penetrating radar,when applied in transmission mode between boreholes, can provide high-resolutioninformation on lithological and hydrological features. The technique may be applied intomographic mode and in a much simpler vertical profile mode. Both modes ofmeasurement have been utilized using two boreholes 5 m apart located at a field site inthe UK Sherwood Sandstone aquifer. Radar transmission measurements have been usedto characterize the change in moisture content in unsaturated sandstone due to controlledwater tracer injection. Continual monitoring of cross-borehole radar measurements overan 18 month period has also permitted determination of travel times of natural loading tothe system and has revealed the impact of subtle contrasts in lithology on changes inmoisture content over time. The time series of inferred moisture contents show clearlywetting and drying fronts migrating at a rate of approximately 2 m month�1 throughoutthe sandstone.

1. Introduction

Geophysical methods have been widely employed for manyyears as an aid to hydrogeological characterization. Tradition-ally, such methods, such as surface-deployed DC resistivitysurveys, have permitted a relatively crude, albeit often useful,definition of hydrogeologically contrasting units. Recently, anumber of developments of several geophysical methods haveled to more detailed characterization capabilities and, in somecases, quantitative estimation of hydrogeological properties. Inparticular, tomographic surveys between available boreholeshave been shown to produce high-resolution information pre-viously unachievable through conventional surface methods[see, e.g., Daily et al., 1992; Slater et al., 2000]. One suchmethod is cross-borehole radar tomography, which offers notonly high spatial resolution but also estimation of hydrogeo-logical characteristics.

The velocity of high-frequency (10 MHz to 1 GHz) electro-magnetic waves is directly related to the bulk (composite)dielectric constant of the subsurface. In this frequency rangethe dielectric constant is primarily controlled by the polariza-tion of individual water molecules and is therefore stronglydependent on volumetric water content [Arulanandan, 1991].Thus the bulk dielectric constant changes with changes in thewater content (�) or, in the case of saturated media, changes inporosity (n). The chemical composition of the pore fluid (for

typical groundwaters) has relatively small influence on the di-electric constant, thus reducing ambiguity in the interpretationcompared with methods such as resistivity surveying. This re-lationship between dielectric constant and moisture contenthas been well established in the hydrological communitythrough widespread use of time domain reflectrometry (TDR).TDR is essentially a tool for measuring the velocity of electro-magnetic waves at scales up to tens of centimeters using high-frequency (gigahertz) signals. The empirical relationships be-tween dielectric constant and moisture content established byTopp et al. [1980] are widely used for interpretation of TDRdata as volumetric moisture content in near-surface soils.

Like TDR, ground-penetrating radar (GPR) works on theprinciple of transmitting a high-frequency electromagneticwave into the ground, often using frequencies in the range50–1000 MHz. The transmitted signal is measured by a re-ceiver antenna, either directly (transmission mode) or after areflection (reflection mode). Surface GPR surveys map con-trasts in geoelectrical properties of the subsurface from reflec-tion mode surveys. They are commonly carried out to maprelative changes in the subsurface stratigraphy, although anumber of attempts have been made [e.g., Greaves et al., 1996;Hubbard et al., 1997a; van Overmeeren et al., 1997] to determinemoisture content and even permeability by quantification ofthe velocity of the applied electromagnetic waves through thesubsurface.

A major limitation of GPR is the significant attenuation ofthe applied signal in electrically conductive media. In manysurface GPR applications, sensitivity is limited to the top few

Copyright 2001 by the American Geophysical Union.

Paper number 2000WR000089.0043-1397/01/2000WR000089$09.00

WATER RESOURCES RESEARCH, VOL. 37, NO. 11, PAGES 2639–2652, NOVEMBER 2001

2639

meters because of highly electrically conductive near-surfacesediments. A more reliable (and direct) means of estimatingthe velocity of radar waves is possible through transmissionmode surveys. In this case the measured signal is assumed to

travel directly from transmitter to receiver. Such a survey re-quires at least one but preferably two or more boreholes. Thesurveys can be carried out using open or nonmetallic casedboreholes. In the case of single borehole surveys the transmit-

Figure 1. Visual core log showing laminated and nonlaminated sections and median grain size. Dashed linesindicate erosive base for each sequence.

Figure 2. Variation in relative dielectric constant with moisture content for a medium-grained sandstonesample taken from the core. Shaded symbols indicate measurements made during wetting phase; opensymbols indicate measurements made during drying phase. Curves indicate the complex refractive indexmethod (CRIM) model for �s � 5, �s � 6.5, and �s � 9 and Topp et al. [1980] curve.

BINLEY ET AL.: HIGH-RESOLUTION CHARACTERIZATION OF VADOSE ZONE DYNAMICS2640

ter or receiver can be placed on the surface, although attenu-ation due to highly electrically conductive near-surface soilscan restrict such applications [see, e.g., van Overmeeren et al.,1997]. Cross-borehole (borehole to borehole) surveys will alsosuffer from signal attenuation, and so borehole separation isconstrained by the maximum electrical conductivity of the me-dia.

Early applications of cross-borehole GPR surveys (then la-beled geotomography) arose from pioneering work at Law-rence Livermore National Laboratory, Livermore, California,United States, by Lager and Lytle [1977]. Daily and Lytle [1983]and Daily and Ramirez [1984] document some initial field stud-ies applying the technique to geological site characterization.The increasing availability of commercial borehole radar sys-tems and growing acceptance of radar in the hydrological com-munity has led to a number of recent hydrogeological applica-tions of the technique in saturated systems [see, e.g., Lane etal., 1998; Peterson et al., 1999] and unsaturated media [e.g.,Hubbard et al., 1997b; Alumbaugh et al., 2000].

A significant hydrological advantage of such an approach isthe scale at which measurements are made. Conventional hy-drological measurement techniques, such as neutron probesand tensiometry, have measurement scales on the order of

centimeters. With appropriate probes, TDR may be used toobtain average water contents over tens of centimeters. How-ever, heterogeneity of the subsurface occurs at much largerscales, and thus more appropriate measurement techniquesare required. Furthermore, the measurement scales of mostconventional soil moisture sensors are several orders smallerthan the model grid scale for any practical hydrological pre-dictive tool. The need for improved measurement proceduresto address this issue of inconsistency of measurement andmodeling scale has been argued by Beven [1993] and others.Geophysical techniques offer such an appropriate measure-ment scale; furthermore, through appropriate petrophysicalrelationships they may be utilized to characterize hydrogeo-logical parameters, which may dictate the dynamics of thehydrological system. Unlike coring, geophysical methods areminimally invasive and provide in situ measurements on un-disturbed sediments. At appropriate resolution, geophysicaltechniques may also permit investigation of geostatistical prop-erties (as studied by Hubbard et al. [1999]), which is of obviousvalue to stochastic hydrological modeling.

The aims here are to examine the usefulness of cross-borehole radar in assessing the temporal variability of moisturecontent in the 12 m thick vadose zone at a specific field site inthe Permo-Triassic Sherwood Sandstone aquifer in the UnitedKingdom. This was achieved through monitoring of the changein response due to an artificially applied water tracer andcontinued monitoring over a period of 18 months in order tomap the dynamics due to natural loading. The investigationwas driven by the need to understand the unsaturated zonerecharge processes in this aquifer. The Sherwood Sandstone isa major groundwater resource in the United Kingdom and issubject to increasing demand for public and private watersupply. After the Chalk the Permo-Triassic Sandstone aquifergroup is the United Kingdom’s second most important aquiferfor groundwater resources. Licensed pumping from the Sher-wood Sandstone accounts for approximately 25% of all UKgroundwater abstraction [Allen et al., 1997]. Heavy chemicalloading from agricultural and industrial practices threaten thisresource. Despite this, studies of unsaturated processes in theSherwood Sandstone are rare.

2. Site DescriptionThe field site selected for detailed study of the vadose zone

dynamics in the sandstone is located 10 km NE of Doncaster,South Yorkshire, at the Lings Farm smallholding near thesmall town of Hatfield (National Grid Reference SE 653 078).The site was chosen because of the following criteria: (1) closeproximity to a sand/gravel quarry permitting detailed larger-scale hydrogeological surveys, (2) minimal drift cover, and (3)flat topography and reasonably undisturbed grass cover.

Following preliminary surface geophysical surveys and trialauguring, eight boreholes were drilled at the Hatfield site dur-ing June/July 1998. The boreholes were drilled using a 127 mmdiameter tip rotary air flush to a depth below the water table(approximately 12 m). Two of the boreholes (labeled R1 andR2) were drilled 5 m apart for the deployment of boreholeradar antennae. The other boreholes were installed for othercomplimentary geophysical investigations, the results of whichare to be presented elsewhere. Because of the weakness of thesandstone a 76 mm diameter PVC casing was installed in R1and R2, surrounded by a sand/cement backfill. Following dril-ling and prior to backfill and completion, each borehole was

Plate 1. Background radar velocity tomogram based on twomultiple-offset gather (MOG) surveys carried out on February1, 1999.

2641BINLEY ET AL.: HIGH-RESOLUTION CHARACTERIZATION OF VADOSE ZONE DYNAMICS

logged using Mount Sopris natural gamma and 39 KHz elec-tromagnetic induction conductivity tools.

3. Core CharacterizationIn October 1998 a 102 mm diameter core was extracted from

the site using polymer flush rotary wire line drilling, approxi-mately 20 m from R2 and 25 m from R1. The core was ex-tracted for detailed laboratory characterization of hydraulicand electrical properties. Figure 1 shows the lithological logtogether with a profile of mean particle size. Below the glacialdrift (sand, gravel, and cobbles) the core is essentially entirelysandstone. The two main distinguishable sublithologies aremedium-grained sandstone, which comprises the bulk of thecore, and fine and medium sandstone subhorizontally lami-nated on a millimeter scale (occurring in 0.2–0.5 m thick units,spaced at 1–3 m intervals). In addition, a series of thin (1–2 cm)discontinuous siltstone layers occurs within the sandstone be-tween 2.9 and 3.4 m. The grain size log (see Figure 1) showsthat the core consists of about five fining upward sequences,with each sequence being between 1 and 3 m thick. The finer,laminated units occur at the top of each fining upward se-

quence. The dashed lines in Figure 1 indicate the erosive basefor each sequence. As shown later, geophysical logs taken atother boreholes also show evidence of subtle contrast in lithol-ogy of the sandstone.

Four specimens from the core were extracted for measure-ment of dielectric response under varying moisture contents.Specimens were between 80 and 120 mm long, with 100 mmdiameter with a 45 mm coaxial hole to accommodate a centralelectrode, and were analyzed using a radio frequency vectornetwork analyzer (model HP 8712ET) in an arrangement de-scribed in detail by Huang et al. [1999]. Figure 2 shows therelative dielectric constant (�r) measured at 100 MHz over arange of moisture contents (0–28%) for a medium-grainedsandstone sample taken from the core, which was consideredrepresentative of the main lithology. The samples were satu-rated with increments of deionized water and allowed to equil-ibrate for 24 hours prior to taking the dielectric measurements.Measurements under desaturation were achieved with similarequilibration under stepped evaporative drying. Further detailsof the procedure are reported by West et al. [2001].

Measurements on methanol samples suggest accuracy of the

Figure 3. Schematic of field radar measurement procedure and location of tracer injection borehole usedfor February 1999 tracer experiment.


measurements to within 10% [see West et al., 2001]. Despitethis, significant scatter in the partially saturated sandstonemeasurements, particularly in the drying phase, is evident inFigure 2. This is likely to result from relatively inhomogeneousmoisture distribution in the samples during wetting and drying.This inhomogeneity is considered here to be macroscopic (i.e.,at the core scale), although it has been noted by Endres andKnight [1992] that microscopic fluid distribution may also affectobserved dielectric behavior.

A number of relationships between saturation and dielectricconstant have been established [see, e.g., Schon, 1976]. Topp’s[1980] equations are widely used in soil hydrology for moisturecontent estimation using TDR. These regression equationswere determined from measurements on several soils withmoisture contents ranging from 0 to 55% but not for consoli-dated sediments. In such media the Topp equations tend to

overestimate the volumetric moisture content, as observed byHokett et al. [1992] and Sakaki et al. [1998] and as shown for thedata in Figure 2. Another widely used relationship betweensaturation and dielectric constant is the complex refractiveindex method (CRIM) [see, e.g., Roth et al., 1990; Chan andKnight, 1999]. The CRIM has been found applicable to rocksincluding sandstones and tuff [Sakaki et al., 1998]. The CRIMequation gives the bulk dielectric constant (��) for an homoge-neous mixture as a weighted sum of the individual phases [Rothet al., 1990]:

�� 1 � n� �� s � nS ��w � n�1 � S� ��a , (1)

where �s is the dielectric constant of the sediment grains, �w isthe dielectric constant of water (assumed to be 81), �a is thedielectric constant of air (assumed to be 1), n is porosity, andS is saturation.

Figure 4. Inferred volumetric moisture contents for the zero-offset profile (ZOP) R1-R2 using meanvelocity from eight background ZOP surveys. The computed moisture contents are shown for various CRIMmodels. Natural gamma and induction conductivity logs in borehole R1 are included for comparison.


Pla

te2.

Sequ

ence

ofch

ange

sin

moi

stur

eco

nten

tdu

eto

trac

erin

ject

ion

onF

ebru

ary

3,19

99,

dete

rmin

edfr

omM

OG

surv

eys.

Cha

nges

inm

oist

ure

cont

ent

are

base

don

the

CR

IMm

odel

usin

gth

ere

fere

nce

data

colle

cted

onF

ebru

ary

1,19

99.


Assuming an average porosity of 0.32 (derived from coremeasurements), Figure 2 shows the CRIM model applied tothe measured dielectric response for �s � 5, �s � 6.5, and�s � 9. This range is consistent with the findings of Hokett etal. [1992], Reynolds [1997], and Sakaki et al. [1998].

4. Field MeasurementsIn order to determine dielectric properties at the field scale,

borehole-to-borehole surveys may be conducted in two trans-mission modes. In both cases a radar signal is generated froma transmitter placed in one borehole with a receiver deployedin the other. Measurement of the travel time of the receivedwave permits determination of the first arrival and hence theapparent velocity of the electromagnetic wave (v). In onemode, using a multiple-offset gather (MOG), the receiver ismoved to different locations in one borehole while the trans-mitter remains fixed. The transmitter is then moved, and theprocess is repeated (see Figure 3). Following collection of alldata in this mode and determination of the travel time for eachwave path line, it is possible to derive a tomogram of velocitywithin the plane of the borehole pair. In contrast, a zero-offset

profile (ZOP) may be determined by keeping transmitter andreceiver at equal depths. By systematically lowering or raisingthe pair of antennae in the two boreholes, it is possible toobtain an average one-dimensional profile of travel time be-tween the two boreholes and over the length of the boreholes.In both cases in low loss materials and at high frequency thebulk dielectric constant is derived from

�� c/v�2, (2)

where c is the radar wave velocity in air (�0.3 m ns�1).The aim of the surveys reported here was to assess changes

in saturation of the sandstone due to both forced loading(following tracer input) and natural recharge. Borehole-to-borehole radar measurements in both MOG and ZOP modewere conducted at the Hatfield site between February 1, 1999,and July 11, 2000. The surveys were conducted in three stages:(1) preliminary survey for comparison with observed lithologyand determination of baseline measurements, (2) daily surveysfollowing injection of a water tracer on February 3, 2000, and(3) repeat surveys at approximately monthly intervals to assesschanges due to natural inputs. For all surveys the PulseEKKOborehole radar system was used with 100 MHz borehole an-

Figure 5. Changes in vertical moisture profile due to tracer injection on February 3, 1999, determined fromZOP surveys. Changes in moisture content based on the CRIM model and using reference data collected onFebruary 1 and 2, 1999, are shown.


tennae. Surveys were conducted using 0.25 m intervals betweenantennae positions over the range 1–12 m below ground level.Typical survey times were 2 hours for MOG mode and 5 minfor ZOP mode. Instrument calibration was carried out prior toand following each mode of survey and regularly during eachMOG survey in order to remove instrument drift. For bothsurvey modes, first time arrivals were picked manually. Tomo-graphic analysis of the MOG data was carried out using thestraight ray inversion procedure of Jackson and Tweeton [1994]using a discretization of 0.5 m in the horizontal and 0.25 m inthe vertical. Details of radar tomographic processing proce-dures are given by Peterson [2001].

4.1. Background Survey

On February 1, two MOG surveys were conducted: one withtransmitter in R1 and the other with transmitter in R2 5 maway from R1. The mean of the two resultant tomograms ofvelocity is shown in Plate 1. The difference between the tomo-grams from the two surveys is typically less than 2%. Note thatalthough data were collected with antennae located at depths

up to 11.75 m, the section of the tomograms covering depthsover 11 m are not shown because of likely artifacts created byhigh contrasts in dielectric constant at the water table at ap-proximately 12 m depth.

The velocity tomogram shows two distinct layers of relativelyhigh velocity, which indicate low moisture content. Compari-son with the lithological and particle size logs in Figure 1derived for the nearby core shows significant correlation be-tween observed sequences in the core and the velocity con-trasts. It appears that the observed fining-up sequences resultin subtle contrasts in moisture content perhaps because ofcapillary effects leading to greater water retention in the finer-grained units or, alternatively, because of ponding of water onfiner-grained units. The upper zone of the sandstone at 2 mdepth is also seen to be differentiated by a contrast in velocity,suggesting higher saturation in the sandstone immediately be-low the drift.

Figure 4 shows the moisture content profile derived from theaverage velocity (at a given depth) from eight surveys con-

Figure 6. Moisture breakthrough profiles integrated over selected depths using ZOP profiles shown inFigure 5.


ducted on February 1 and 2, 1999, using (1) and (2). Theprofile is shown as a range by using �s � 5, �s � 6.5, and �s �9 in the CRIM model, assuming a constant porosity of 32%.Also shown in Figure 4 is the natural gamma and inductionconductivity log for borehole R1. A strong positive correlationbetween the moisture profile and bulk electrical conductivity isclearly evident. A marked peak is seen in both profiles at 4 mdepth possibly indicating the location of a lower permeabilitylayer. Increase in apparent moisture content and conductivitycan be seen over the interval 6 to 6.5 m which may be a resultof the contrast in sedimentary sequence reported earlier. Slightincreases in natural gamma, which can be associated with in-creased clay content, in the same interval also support thesuggestion of possible reduction in permeability at approxi-mately 6.5 m depth. Farther down the profile (8.5–11 m) in-creased moisture content and conductivity can be seen, per-haps a result of close proximity to the capillary fringe but alsoa possible result of further layering.

4.2. Response due to Tracer Input

A tracer experiment was set up during February 1999 toassess the response of the sandstone to forced loading. Usinga short borehole drilled to 3.5 m, positioned in the plane of R1and R2 (see Plate 1), 1000 L of water were injected duringFebruary 3, 1999. The electrical conductivity of the water wasraised to 600 �S cm�1 by addition of a weak NaCl solution soas to achieve similar conductivity of the native pore water. A1 m screened section and gravel pack around the injectionborehole permitted tracer migration directly into the sand-stone. Between February 4 and February 10, MOG surveyswere carried out each day together with duplicate ZOP sur-veys. Surveys were repeated on February 17 and February 27,1999. Using the established baseline surveys carried out onFebruary 1 and 2, 1999, it is possible to determine changes inmoisture content due to tracer loading.

The CRIM model of the dielectric-saturation relationshipshow significant sensitivity of predicted moisture content todielectric parameters, as illustrated earlier in Figure 4 for thebackground ZOP survey results. The adopted CRIM modelparameter (�s) was derived from samples of the main lithol-ogy; however, West et al. [2001] show that finer-grain size sam-ples taken from the core have larger values and thus limit theuse of such an approach for determination of absolute mois-ture contents. Changes in saturation, however, are insensitive

to the dielectric constant of the sediment grains since, for theCRIM model given in (1), the change in volumetric moisturecontent is given by

n�St � S0� �� t � �� 0

��w � ��a, (3)

where St is saturation at time t , S0 is background saturation, �� t

is the observed bulk dielectric constant at time t , and ��0 is theobserved background bulk dielectric constant.

4.2.1. MOG survey results. Plate 2 shows the change inmoisture content determined using (3) applied to the tomo-grams from the MOG surveys. All the tomograms shown inPlate 2 are based on the same background image determinedfrom the pretracer surveys. From the difference images thewetted zone can be seen to be separated into two distinctplumes: a dominant zone of increased saturation steadily mi-grates vertically downward from the source, while a stationaryregion of increased moisture content is apparent over thedepth interval 2.5–3 m. This separation indicates the imped-ance to flow or increased soil moisture retention in the uppersandstone caused by fine-grained sediments such as the finesandstones and siltstones observed in the core between 2.9 and3.4 m. In contrast, it would appear that tracer directly injectedbeneath this layer moves rapidly downward to 6 m after a fewdays. Note that from the background tomogram in Plate 1 thisdepth coincides with the base of a layer of reduced saturationwhich is perhaps due to relatively high permeability. Duringthe later stages of the experiment the wetting front continuesto move vertically at a slower rate with significant changes at8 m depth observable 14 days after the start of the experiment.Decay of the tracer signal with time is a result of migration offplane, particularly as further impeding layers are encounteredwith depth. Nevertheless, the set of tomograms shows an ex-tremely consistent sequence of moisture migration throughoutthe sandstone.

4.2.2. ZOP survey results. The movement of the wettingfront is perhaps more clearly seen from the sequence of mois-ture difference profiles derived from the ZOP surveys. Figure5 shows the sequence from which it can be seen that a frontinitially travels at approximately 0.15–0.2 m d�1 during theperiod February 4–9. On February 10 the sharpness of thewetting front signal is reduced, although increases in saturationto 8 m depth can be seen on February 17. It is recognized that

Figure 7. Net rainfall for Hatfield site over survey period. Data are supplied courtesy of EnvironmentAgency, Leeds, England, United Kingdom.


the changes in saturation due to the tracer are three-dimensional, whereas the ZOP sequence is effectively an inte-grated one-dimensional profile. The impact of subtle contrastsin grain size (and inferred contrasts in permeability) can beclearly seen, however. This is more striking when presented inthe form of breakthrough curves, as shown in Figure 6. Herechanges in moisture content from the ZOP surveys are aver-aged over 1 m intervals (four antennae positions at 0.25 mintervals). The rapid response to depths of 5 m contrasts thediffusive signal at greater depths. In particular, a markedchange in arrival time distribution and mass recovery can beseen in the 6.25–7 m interval, again a result of impedance tovertical migration. An increase in saturation in the near-surface layers after tracer infiltration had ceased is also appar-ent from the plots in Figure 6. This is due to natural inputs,which became the focus of continued surveys at the site.

4.3. Response due to Natural Inputs

A controlled point source tracer injection permits charac-terization of hydraulic processes within the sandstone; it is,however, the response due to natural nonpoint source inputswhich is of primary importance. Environmental regulators re-quire accurate estimates of travel times of potential pollutants

in the unsaturated zone in order to predict impacts on ground-water resources. Following the tracer experiment described insection 4.2, repeat MOG and ZOP surveys were carried outwith an approximate sampling interval of 1 month up until July2000 to assess the change in moisture conditions throughoutthe entire thickness of the unsaturated zone.

Daily rainfall, measured locally, coupled with actual evapo-transpiration estimates (based on the Meteorological OfficeRainfall and Evapotranspiration Calculation System, Brack-nell, England, United Kingdom values) have been used to com-pute monthly net rainfall values prior to and over the monitoringperiod and are shown in Figure 7. The data shown in Figure 7reveal seasonal characteristics over the 2 year monitoring period:The main inputs occur during September to January with a num-ber of spring events during March or April producing a delayedrecharge source followed by relatively high recharge losses due toevapotranspiration during summer.

4.3.1. MOG survey results. Plate 3 shows the change inmoisture content determined using (3) applied to selectedMOG survey results. The data set determined prior to thetracer injection on February 1 and 2, 1999, has been used as thebackground image for all tomograms. Note that, unlike thetracer sequence, the images now show a response which is

Figure 8. Changes in vertical moisture profile due to natural inputs, determined from ZOP surveys. Changesin moisture content based on CRIM model and reference data collected on February 1 and 2, 1999, are shown.


Pla

te3.

Sele

cted

sequ

ence

ofch

ange

sin

moi

stur

eco

nten

tdu

eto

natu

rali

nput

s,de

term

ined

from

MO

Gsu

rvey

s.C

hang

esin

moi

stur

eco

nten

tba

sed

onth

eC

RIM

mod

elus

ing

refe

renc

eda

taco

llect

edon

Feb

ruar

y1

and

2,19

99,a

resh

own.


laterally continuous because of the nature of the loading. Theimpact of the high net rainfall during March 1999 can beclearly seen, although it is recognized that this is likely to bedisplacement of water originating from heavy winter rainfall.As the front moves down during the summer of 1999, emer-gence of a drying front can be seen which develops through towinter 1999.

4.3.2. ZOP survey results. As in the tracer experimentthe moisture dynamics are more clearly seen from the se-

quence of ZOP survey results. Figure 8 shows this sequenceusing data collected up until June 2000. The winter/springrecharge and summer drying fronts are clearly seen. Also no-ticeable is the apparent increase in saturation at depths of 10 mand greater. This is due to a rise in the water table (andassociated capillary fringe) not necessarily due to natural re-charge but is likely to be a response caused by decreasedabstraction of a public supply well located some 500 m awayfrom the site.

Figure 9. Moisture breakthrough profiles integrated over selected depths using ZOP profiles shown inFigure 8.


Figure 9 illustrates the temporal dynamics throughout theprofile over the complete monitoring period, including thetracer experiment. Ignoring the effect of the tracer, the shapeof the response in the upper sandstone (2.25–3 m depth) fol-lows the seasonal pattern of the net rainfall in Figure 8. Notethe reduction in moisture content changes for the peak in 2000in comparison to the peak in 1999: a result of significantlyreduced rainfall during winter 1999. The delays in wetting anddrying patterns are also clearly visible over this interval whencompared with the rainfall input series, as is the smooth cyclicpattern, both demonstrating the impact of drift cover. Moresignificant, however, is the transmission of this signal through-out the profile. A delay of 2–3 months in the initial peak isapparent when comparing series at 2.25–3 m and 7.25–8 m. Asimilar delay is also apparent in the propagation of the dryingfront. It is interesting to note that the transmission of thewetting front appears geologically controlled at about 6 msince a marked change is apparent when comparing time seriesfor the 5.25–6 m and 6.25–7 m intervals. This is consistent withthe observed response due to tracer input (Figure 6) and back-ground tomogram (Plate 1).

5. ConclusionsCross-borehole radar may provide supplementary informa-

tion about lithological features through appropriate petro-physical relationships between radar wave velocity and mois-ture content. These relationships, such as the CRIM modelused here, are sensitive to accurate determination of appro-priate parameters, suggesting that determination of absolutevalue of moisture content will suffer high uncertainty limits.Such sensitivity, however, does not exist when applied to thestudy of changes in saturation levels, as demonstrated here.Through repeated measurements of borehole-to-boreholetransmission GPR the moisture dynamics in the unsaturatedzone of a regionally important sandstone aquifer have beenreported.

Tomographic measurement and analysis permits detailedspatial structures to be characterized and may be extendedeasily to three-dimensional investigations [Jackson andTweeton, 1996; Eppstein and Dougherty, 1998]. In contrast, ra-dar velocity profiling through ZOP surveys allows rapid mea-surement of moisture content changes with little postprocess-ing required. The major advantage, however, is that ZOPsurveys may be viewed as large-scale TDR-like profiles whereaveraging is performed in the horizontal direction. For thefield site described here, estimates were made of moisturecontent and changes in moisture content with time over a 5 mhorizontal distance and a vertical discretization of 0.25 m.Provided the bulk electrical conductivity of the subsurface islow enough, increased borehole separation is possible withoutsignificant deterioration of the received signal through atten-uation (we have successfully applied the same technique atother similar sites using borehole separation up to 15 m).Increased transmitter power will help improve the scale ofproblems which may be investigated using this technique.

The ZOP survey may then offer a valuable tool for verticalprofiling of vadose zone hydrological dynamics. Cross-borehole GPR measurements at the Hatfield site and othersimilar UK sandstone sites are continuing with a view to im-proving regulatory models of this important element of thehydrological cycle which previously has received limited atten-tion. Computation of residence times should be achievable by

appropriate modeling of the dynamics throughout the profile,which would provide useful insight into the impact of nonpointsource pollutants on the local groundwater resource.

Acknowledgments. We are grateful to John Aldrick (EnvironmentAgency, Leeds, England, United Kingdom) for continued support forour work. Ted Mould and Albert Walmsley (Environment Agency,Leeds, England, United Kingdom) brought drilling expertise to theproject. The work would not have been possible without agreement ofsite access by John Cunliffe of Ling’s Farm, Hatfield. Many Lancasterstudents and research staff have contributed to many days of datacollection, in particular, Jamie Baron, Hamza Bouabdallah, and MarcoVaccari. We are grateful to the anonymous reviewers for their con-structive comments. This work was funded by the Natural Environ-ment Research Council, United Kingdom, under NERC grant GR3/11500.

ReferencesAllen, D. L., L. J. Brewerton, L. M. Coleby, B. R. Gibbes, M. A. Lewis,

A. M. MacDonald, S. J. Wagstaff, and A. T. Williams, The physicalproperties of major aquifers in England and Wales, Tech. Rep.WD/97/34, 312 pp., Br. Geol. Surv., Wallingford, Oxfordshire, En-gland, 1997.

Alumbaugh, D., L. Paprocki, J. Brainard, and C. Rautman, Monitoringinfiltration within the vadose zone using cross-borehole ground pen-etrating radar, in Proceedings of the Symposium on Applications ofGeophysics to Engineering and Environmental Problems (SAG-EEP2000), pp. 273–281, Environ. and Eng. Geophys. Soc., Denver,Colo., 2000.

Arulanandan, K., Dielectric method for prediction of porosity of sat-urated soil, J. Geotech. Eng., 117(2), 319–330, 1991.

Beven, K. J., Estimating parameters at the grid scale: On the value ofa single measurement, J. Hydrol., 143, 109–124, 1993.

Chan, C. Y., and R. J. Knight, Determining water content from di-electric measurements in layered materials, Water Resour. Res., 35,85–93, 1999.

Daily, W., and J. Lytle, Geophysical tomography, J. Geomagn. Geo-electr., 35, 423–442, 1983.

Daily, W. D., and A. L. Ramirez, In situ porosity distribution usinggeophysical tomography, Geophys. Res. Lett., 11(6), 614–616, 1984.

Daily, W. D., A. L. Ramirez, D. J. LaBrecque, and J. Nitao, Electricalresistivity tomography of vadose water movement, Water Resour.Res., 28, 1429–1442, 1992.

Endres, A., and R. J. Knight, A theoretical treatment of the effect ofmicroscopic fluid distribution on the dielectric properties of partiallysaturated rocks, Geophys. Propospect., 40, 307–324, 1992.

Eppstein, M. J., and D. E. Dougherty, Efficient three-dimensional datainversion: Soil characterization and moisture monitoring from cross-well ground-penetrating radar at the Vermont test site, Water Re-sour. Res., 34, 1889–1900, 1998.

Greaves, R. J., D. P. Lesmes, J. M. Lee, and M. N. Toksoz, Velocityvariations and water content estimated from multi-offset, groundpenetrating radar, Geophysics, 61(3), 683–695, 1996.

Hokett, S. L., J. B. Chapman, and C. E. Russell, Potential use of timedomain reflectrometry for measuring water content in rock, J. Hy-drol., 138, 89–96, 1992.

Huang, Y., Y. T. Zhang, and L. J. West, A large one-port co-axialdielectric line system for dielectric property measurement, in Pro-ceedings of the Ninth International Conference on ElectromagneticMeasurement, pp. 31/1–31/4, Natl. Phys. Lab., Brighton, England,UK, 1999.

Hubbard, S. S., Y. Rubin, and E. Majer, Ground-penetrating-radar-assisted saturation and permeability estimation in bimodal systems,Water Resour. Res., 33, 971–990, 1997a.

Hubbard, S. S., J. E. Peterson, E. L. Majer, P. T. Zawislanski, K. H.Williams, J. Roberts, and F. Wobber, Estimation of permeable path-ways and water content using tomographic radar data, Leading Edge,16(11), 1623–1628, 1997b.

Hubbard, S. S., Y. Rubin, and E. Mayer, Spatial correlation structureestimation using geophysical and hydrogeological data, Water Re-sour. Res., 35, 1809–1825, 1999.

Jackson, M. J., and D. R. Tweeton, MIGRATOM—Geophysical to-


mography using wavefront migration and fuzzy constraints, Rep.RI9497, 35 pp., Bur. of Mines, Minneapolis, Minn., 1994.

Jackson, M. J., and D. R. Tweeton, 3DTOM: Three-dimensional geo-physical tomography, Rep. RI9617, 84 pp., Bur. of Mines, Minneap-olis, Minn., 1996.

Lager, D. L., and R. J. Lytle, Determining a subsurface electromag-netic profile from high-frequency measurements by applying recon-struction-technique algorithms, Radio Sci., 12, 249–260, 1977.

Lane, J. W., P. K. Joesten, F. P. Haeni, M. Vendl, and D. Yeskis, Useof borehole-radar methods to monitor the movement of a salinetracer in a carbonate rock at Belvidere, Illinois, in Proceedings of theSymposium on Applications of Geophysics to Engineering and Envi-ronmental Problems (SAGEEP98), pp. 323–328, Environ. and Eng.Geophys. Soc., Denver, Colo., 1998.

Peterson, J., Pre-inversion corrections and analysis of radar tomo-graphic data, J. Environ. Eng. Geophys., 6(1), 1–18, 2001.

Peterson, J. E., E. L. Majer, and M. D. Knoll, Hydrogeological prop-erty estimation using tomographic data at the Boise hydrogeophysi-cal research site, in Proceedings of the Symposium on Applications ofGeophysics to Engineering and Environmental Problems (SAG-EEP99), pp. 629–638, Environ. and Eng. Geophys. Soc., Denver,Colo., 1999.

Reynolds, J. M., Introduction to Applied and Environmental Geophysics,796 pp., John Wiley, New York, 1997.

Roth, K. R., R. Schulin, H. Fluhler, and W. Attinger, Calibration oftime domain reflectometry for water content measurement using acomposite dielectric approach, Water Resour. Res., 26, 2267–2273,1990.

Sakaki, T., K. Sugihara, T. Adachi, K. Nishida, and W. Lin, Applica-

tion of time domain reflectometry to determination of volumetricwater content in rock, Water Resour. Res., 34, 2623–2631, 1998.

Schon, J. H., The Physical Properties of Rocks: Fundamentals and Prin-ciples of Petrophysics, Pergamon, New York, 1976.

Slater, L., A. Binley, W. Daily, and R. Johnson, Cross-hole electricalimaging of a controlled saline tracer injection, J. Appl. Geophys., 44,85–102, 2000.

Topp, G. C., J. L. Davis, and A. P. Annan, Electromagnetic determi-nation of soil water content: Measurement in coaxial transmissionlines, Water Resour. Res., 16, 574–582, 1980.

van Overmeeren, R. A., J. C. Gehrels, and S. V. Sariowan, Groundpenetrating radar for determining volumetric soil water content:Results of comparative measurements at two test sites, J. Hydrol.,197, 316–338, 1997.

West, L. J., Y. Huang, and K. Handley, Dependence of sandstonedielectric behavior on moisture content and lithology, in Proceedingsof the Symposium on Applications of Geophysics to Engineering andEnvironmental Problems (SAGEEP2001), [CD-ROM], Environ. andEng. Geophys. Soc., Denver, Colo., 2001.

A. Binley, R. Middleton, and P. Winship, Department of Environ-mental Science, Lancaster University, Lancaster LA1 4YQ, England,UK ([email protected]; [email protected];[email protected])

M. Pokar and J. West, School of Earth Sciences, University ofLeeds, Leeds L52 9JT, England, UK. ([email protected];[email protected])

(Received November 10, 2000; revised March 20, 2001;accepted March 26, 2001.)


college of engineering - high-resolution characterization of … · 2001. 11. 13. · spaced at...

Documents