Rep

rodu

ced

from

Vad

ose

Zon

e Jo

urna

l. P

ublis

hed

by S

oil S

cien

ce S

ocie

ty o

f Am

eric

a. A

ll co

pyrig

hts

rese

rved

.

Infrared Thermography and Fracture Analysis of Preferential Flow in Chalk

Annette E. Rosenbom* and Peter R. Jakobsen

ABSTRACT for the flow regime, and they may furthermore exhibita dynamic hydraulic activity (Commission on Geosci-Preferential flow in fractured Cretaceous chalk has been investi-ences, Environ. and Resour., 1996). Thus, identificationgated using infrared thermography (IRT) in a quarry in Denmark.of the most important hydraulically active fractures andFracture mapping along the vertical quarry walls shows numerous

horizontal fractures and four sets of vertical fractures respectively hereby the preferred fluid pathways are crucial (Bloom-striking 25�, 60�, 145�, and 175�. Water flows out of the exposed walls field, 1996).mainly through fractures and to a minor extent by seepage through the The standard technique for understanding and pre-matrix. To clarify which of the five fracture systems are hydraulically dicting the preferential flow patterns in chalk includesactive, the exposure has been investigated using IRT. By making use the following steps: (i) identification of the sources orof the contrast between the constant temperature of groundwater and mechanisms that are capable of producing the stressthe temperature of the exposed wall on a cold winter and a hot summer

field requisite for initiating and propagating the frac-day, zones of groundwater discharge have been delineated. The resultstures, (ii) formulation of a fracture network model basedof the IRT suggest that the vertical shear fractures are the mainon detailed quantitative descriptions of the geometryhydraulically active conduits. Furthermore, the horizontal fractures(orientation, trace length, spacing, and aperture) andare more hydraulically active in areas where they intersect a vertical

stained fracture or in association with some flint layers. Stains of the connectivity of different fracture sets in the fractureFe3� and Mn were observed primarily in fractures with a strike of network based on geological mapping of outcrops andapproximately 25� and 175�, suggesting that the water flows preferen- underground exposures and borehole data (Commis-tially in these fracture sets. The possible influence of the regional sion on Geosciences, Environ. and Resour., 1996), andsetting on the flow pattern is considered. The results suggest that the (iii) investigation of the flow patterns using indirectcombined use of detailed fracture analysis and IRT is a valuable methods such as pumping tests, well interference stud-method for providing the input data required for both modeling and

ies, and tracer tests (Downing et al., 1993; Price et al.,monitoring of fractured media.1993). The above indirect methods have the importantlimitation that they only provide aggregated informationon the effects of the fractures at scales significantlyThe depositional, diagenetic, and tectonic history oflarger than the fractures because of the data uncertaintychalk has created an unique dual-porosity reservoiron fracture locations, properties, connectivity, boundaryrock with a high-porosity matrix, which is relatively im-conditions, fracture–matrix interactions, and host matrixpermeable ≈8 � 10�15 m2 (Frykman, 2001) given poreproperties. To obtain data that can provide more de-throats typically in the range 0.05 to 1.0 �m (Price ettailed insight into the flow mechanisms in fractured me-al., 1976). The chalk stands apart from all other lime-dia, more refined techniques are needed.stones by virtue of its very fine grain size and relatively

In this respect, IRT is a promising technique thathigh friability (Selley, 1976) and can only function asprovides information on a refined spatial scale. Infraredan aquifer or a hydrocarbon reservoir because it is frac-thermography has been used in a variety of applicationstured. As opposed to the broader formation-type lime-(Miljure, 1992) because it is a nondestructive, noncon-stone, the hydraulic properties of chalk matrix vary littletact, and noninvasive technique that uses portable infra-between sites. The origin of the basic fracture systemred scanning equipment for detecting thermal changesis mainly tectonic, but a combination of factors such asin living organisms, objects, and domains. In relation todissolution, stress release as overburden is removed,hydrology, the method has been used in a variety ofand weathering can lead to significant enlargement ofways, e.g., in connection with mapping of submarinethese initial fractures.groundwater discharge (Bayari and Kurttas, 2002; Port-Fractures impose a dominating influence on the flownoy et al., 1998), mapping surface temperature and cir-mechanisms in many chalk systems by offering preferen-culation patterns in lakes (Anderson et al., 1995; Garretttial pathways for fluids. Chalk formations are of interestand Hayes, 1997; Lathrop and Lillesand, 1987; LeDrewfor engineering in the context of energy resources (petro-and Franklin, 1985), modeling heat dispersion in ther-leum, natural gas, geothermal water, and steam), watermal effluent plumes (Davies et al., 1997; Jensen et al.,supply, and environmental problems (migration of chemi-1988; Schott 1979), identifying subsurface springs (Rox-cal and radiation contamination, remediation, and protec-burgh, 1985), examining spatial patterns of stream tem-tion). However, all fractures are not of equal importanceperature (Torgersen et al., 2001), and studying moisture(water movement) in porous materials (Avdelidis et al.,

The Geological Survey of Denmark and Greenland, Øster Voldgade 2003). Applications of IRT have also been used to detect10, DK 1350 Copenhagen K, Denmark. Received 3 May 2004. Original fractures/cracks and other deficiencies in materials (Moro-Research Paper. *Corresponding author (aer@geus.dk). poulou and Avedelidis, 2002; Shiratori et al., 1994). In

other words, IRT has been used for detecting fracturesPublished in Vadose Zone Journal 4:271–280 (2005).doi:10.2136/vzj2004.0074© Soil Science Society of America Abbreviations: BHTV, borehole televiewer; bsl, below sea level; IRT,

infrared thermography.677 S. Segoe Rd., Madison, WI 53711 USA

271

Published online April 26, 2005

Rep

rodu

ced

from

Vad

ose

Zon

e Jo

urna

l. P

ublis

hed

by S

oil S

cien

ce S

ocie

ty o

f Am

eric

a. A

ll co

pyrig

hts

rese

rved

.

272 VADOSE ZONE J., VOL. 4, MAY 2005

till. Below the glacial deposits, 45 m of Cretaceous chalk areand studying water flow and changes of water contenttemporarily exposed in the quarry. The chalk is a fairly uniformin different media but has so far not been used forcarbonaceous mud with layers of nodular flint. The nodularstudying flow in fractured porous media.flint outlines low mounds within the chalk (Surlyk, 1997; SurlykThe main objective of this study is to investigate howand Hakansson, 1999). Pronounced horizontal fracturing as wellthe combined use of detailed fracture analysis and IRTas numerous vertical fractures occur within the chalk.methods can be used for identifying hydraulically active The chalk of the study area and region is deposited in thefractures in chalk and hereby provide the input data Danish Basin (Fig. 2). The Danish Basin is developed adjacent

required for the construction of conceptual fracture net- to the Fennoscandian Border Zone. The zone is delineatedwork models. This is done by analyzing the fracture by large faults, which branch into the Danish Basin. During thenetwork of the study site based on previous regional Late Cretaceous and Early Cenozoic, dextral transpressioneltectonic and hydrologic knowledge (presented in the stresses induced partial inversion of the basin along the Fen-Site characterization section) and site-specific investiga- noscandian Border Zone as well as dextral (right lateral)

movement along the faults (Liboriussen et al., 1987). With ations using scanline and borehole mapping techniquesdextral strike-slip movement along the north–south faults, the(presented in the Methods section). The results of themaximum stress (�1) would be in northeast–southwest direc-detailed fracture analysis are combined and comparedtion (Fig. 2). In this stress regime, a number of regional fracturewith the results of IRT investigations. Through thissystems might be expected: a system of extensional fracturesstudy, it will be assessed whether and under which condi-parallel to the �1 direction, two systems of shear fractures withtions the combined use of detailed fracture analysis andan approximate angle of 30� from �1, and a system of stressIRT field techniques are useful as a practical field tool release fractures perpendicular to �1 (Hobbs et al., 1976) (Fig.in connection with studies of preferential flow in frac- 2b). During the Neogene, a regional large-scale uplift occurred

tured chalk. and 500 to 750 m of basinal deposits were eroded (Japsen,1993; Japsen and Bidstrup, 1999), resulting in horizontal exten-sional fractures.MATERIALS AND METHODS

The water table elevation of the study area is dominatedSite Characterization by the production of water (for the chalk treatment) from thetwo lakes in the quarry. The water table of the lakes was normallyThe study was performed at the Sigerslev chalk quarry, Den-

mark (Fig. 1). At the quarry, the uppermost 6 m are glacial artificially kept at approximately 5 m below sea level (bsl),

Fig. 1. Location map of the Sigerslev site with location of walls for infrared thermography (IRT) and wells for detailed IRT investigation.Additionally, a water table elevation map of the study area is presented.

Rep

rodu

ced

from

Vad

ose

Zon

e Jo

urna

l. P

ublis

hed

by S

oil S

cien

ce S

ocie

ty o

f Am

eric

a. A

ll co

pyrig

hts

rese

rved

.

www.vadosezonejournal.org 273

Fig. 1 (Aamand et al., 2001). During the study, the groundwater and an inclination of 20� (from vertical) and without casinglevel was lowered to 20 m bsl, resembling the bottom of the except for the upper 2 m. The well inclinations and the azimuthquarry. However, this artificial lowering of the groundwater of well ends were chosen for the following reasons: (i) to allowtable only 400 m from the coastline did not result in any for immersion of the televiewer image logging equipment; (ii)recordings of saltwater intrusion in a monitoring well situated to penetrate the 20-m-thick bed sequence visualized on thebetween the coast and the active part of the quarry. wall; (iii) to measure variations in the water table of the wells,

even at 20-m drawdown in the quarry; and (iv) to intersect amaximum amount of vertical fractures. To avoid direct crossScanline Mapping of Site-Specific Fracture Systemsflow between the wells, they were drilled with a separation

Fracture data were collected along scanlines on the quarry distance at the surface of 10 m. When lowering the waterwalls and to some extent on the floor of the quarry (represent- table in the quarry to 20 m bsl, the level in the wells wereing levels from �30 m to –20 m). The orientation, distance approximately 10 m bsl.along the scanline, and length of the fracture traces were Cores from the wells and televiewer image logs [boreholemeasured. The fracture surfaces were also described with spe- televiewer (BHTV)] (Zemanek et al., 1970) were used to mapcial emphasis on Fe and Mn staining, which is an indirect the flint layers and fractures in the two wells. The BHTV usesindicator of current or past groundwater flow in the fracture. an ultrasonic beam with centerband frequency of 500 kHz toHydraulic active fractures on the floor of the quarry were

1.5 MHz and a beam diameter of 0.5 to 1.0 cm. The BHTValso registered.operates by scanning the borehole wall at a rate of about threerevolutions per second while the probe is steadily lowered in

Borehole Investigation for Fracture Properties the borehole. The transducer is fired at a rapid rate and servesas both the source of the acoustic pulse and the receiver forTwo wells (Well I and II) with a diameter of 6 cm and athe reflected signal. The BHTV image is oriented with respectlength of approximately 28 m (volume ≈0.08 m3) were drilledto the local magnetic field by a down hole magnetometer. Theinto the south-facing wall by the lake 2 to 3 m from the edgeimage is split along the apparent north azimuth and “un-of the wall (Fig. 1). The wells were drilled with an azimuth

of the well end on approximately N280 (parallel with the wall) wrapped” for display (Commission on Geosciences, Environ.

Fig. 2. (a) The tectonic setting in the Danish Basin. The dextral (right lateral) displacement along the faults delineating the tectonic block whereSigerslev is situated is indicated with red arrows. (b) Tectonic model showing the expected fractures formed within the given stress regimeprevailing in the Stevns block (Hobbs et al., 1976): two sets of conjugated shear fractures, one set of extensional fractures, and one set ofextensional fractures related to stress release.

Rep

rodu

ced

from

Vad

ose

Zon

e Jo

urna

l. P

ublis

hed

by S

oil S

cien

ce S

ocie

ty o

f Am

eric

a. A

ll co

pyrig

hts

rese

rved

.

274 VADOSE ZONE J., VOL. 4, MAY 2005

and Resour., 1996). With the image, it is possible to find thestrike and dip of the fractures identified in the well. For WellII, an additional BHTV log was conducted approximately 1.5yr after the detailed IRT investigation revealing accumulationof iron oxide along certain fractures.

Infrared Thermographic Investigations

Infrared thermography was used to map the groundwateroutflow pattern in the quarry. By making use of the contrastbetween the constant temperature of the groundwater (≈8�C)and the temperature of the surface of walls, the groundwaterdischarge zones were delineated.

Infrared thermography investigations were performed withan AGEMA 550 camera. With the high-performance, hand-held (horizontally), digital infrared camera, it is possible withthe given calibration to measure temperature within the rangeof –20�C to �65�C with an accuracy of �2�C and an automatictransmission correction based on distance, atmospheric tem-perature, and humidity. The camera was set up with an emis-sivity factor of 0.96, which represents the natural chalk surface.The spatial image resolution is 320 by 240 pixels, and theproduced thermo grams (heat images) have up to 256 colors,which clearly show thermal profiles and range of temperatures.The wavelengths that IRT primarily image are in the short-wave region from 3 to 5 �m and the long-wave region from7 to 14 �m. The camera used in this study operates in thespectral range 3.6 to 5.0 �m (AGEMA, 1997).

Fig. 3. Relative water table fluctuations and water temperature inAn IRT survey on the four walls in the quarry was per-Well I and Well II while injection tests are performed in Well Iformed during a cold winter morning (air temperature �14�C)and Well II, respectively. The reference level for the water columnto study the natural seepage from the walls with the aim of is approximately 10 m below sea level.

mapping large-scale preferential flow paths. The survey wasconducted under calm weather conditions and before sunrise

RESULTSto avoid uneven heating of the walls due to the wind and thesun. Stored heat from the sun the previous day was neglectable Results of the Fracture Analysissince it was cloudy and cold weather.

In the quarry, one horizontal and four vertical frac-Another IRT investigation, combined with a flow experi-ment (induced seepage), was conducted on a hot summer day ture systems are recognized. The upper 9 m below thewhen the weather conditions were calm, but the sun energy glacial till are crushed and heavily fractured by the Qua-heated up the surface of the wall to approximately 25�C. The ternary glaciers, which have overridden the region. Thesouth-facing wall by the lake was chosen for this investigation, number of horizontal fractures increases rapidly upwardand Well I and II were drilled. During this investigation, ap- in the upper 9 m of chalk. Below this upper glaciallyproximately 2.25 m3 of 13�C water was injected into Well I fractured zone, the spacing of the horizontal fractureswith rates of 20 l/min for 60 min and 32 l/min for 30 min over a is more uniform although it increases slightly with depthtotal time period of approximately 1.5 h (referred to as Test 1).

from approximately 20 cm at 20 m above sea level toDuring the highest injection rate, Well I was filled with water.approximately 40 cm at 20 m bsl. The horizontal frac-Simultaneously, thermal videotape and digital images weretures are usually more than 20 m long and pass throughtaken documenting the temperature changes on the wall. Waterflint layers, which outline low mounds in the chalk.for the injection was supplied by a garden hose connected toAlong subhorizontal flint layers, horizontal fracturesthe nearest water supply line. During the transport in the

garden hose, the injection water was heated from approxi- occur at the interface between flint and chalk. At themately 8�C to 15�C because of the sun. The flow rate, which wall, water is primarily seeping out from the horizontalwas measured by a flow meter installed at the end of the fractures at the intersection points with a vertical frac-garden hose, was increased by 12 l/min after 1-h injection ture or in association with some of the subhorizontalperiod because hardly any outflow on the wall was observed flint layers. At the intersections, channels are oftenvia the thermal videotape. Approximately 1.5 h after the injec- eroded into the horizontal fracture plane, indicating ation into Well I, water was injected in Well II using an injection transport of sediment and hereby flow.rate of 32 l/min over 70 min for a total amount of 2.25 m3

The orientations of the four vertical fracture systems(referred to as Test 2). The starting time for Test 2 was deter-are 25�, 60�, 145�, and 175� (Fig. 4a), which closely resem-mined on the basis of the thermal video showing re-establish-ble the suggested fracture systems in the regional frac-ment of the initial temperature pattern of the wall (�1�C)ture model (Fig. 2b). The trace length of the verticalbefore Test 1 and the re-establishment of the water tablefractures varies from 20 cm to more than 20 m with anin the Well I and II. Combined pressure and temperatureaverage of about 1.5 m.transducers were installed in each well for monitoring changes

in water table and temperature every 2 s (Fig. 3). The horizontal fractures divide the chalk into me-

Rep

rodu

ced

from

Vad

ose

Zon

e Jo

urna

l. P

ublis

hed

by S

oil S

cien

ce S

ocie

ty o

f Am

eric

a. A

ll co

pyrig

hts

rese

rved

.

www.vadosezonejournal.org 275

Fig. 4. Fracture analysis of vertical fractures. (a) Rose diagram (showing orientation and frequency) of all vertical fractures measured in theSigerslev Chalk Quarry. (b) Rose diagram of vertical fractures with estimated trace length, grouped into multilayer fractures and single-layerfractures. (c) Rose diagram of vertical fractures registered for staining, with staining, and without staining. (d) Diagram showing orientationof hydraulic active fractures in the bottom of the quarry.

chanical layers. The vertical fractures are categorized dized and precipitates as it reaches the borehole. Thelargest accumulation of the iron oxide is, however, ob-as multilayer, if they cross several mechanical layers, andserved in the fractures with a north–south orientationsingle-layer fractures when confined to one mechanicalcorresponding to the 175� fracture system. Some of thelayer. The orientations of the multilayer fractures mainlyhorizontal fractures also have this iron oxide coating,correspond to the 175� and the 25� fracture systemsespecially the three fractures that are closely connectedwhile the single-layer fractures mainly correspond to theto the flint layer at level approximately �6.75, �16.5,60� and the 145� fracture systems (Fig. 4b). Multilayerand �23.75 m (Fig. 5).vertical fractures have a higher connectivity than single-

The orientations of the four vertical fracture systemslayer fractures, and they are thus expected to have aresemble the expected fracture systems suggested by thelarger effect on the groundwater flow.tectonic model with the maximum stress (�1) pointingFerric iron and Mn staining on the fracture surfacetoward south–southwest. The 25� fracture system corre-suggests that the fracture has been hydraulically active.sponds to the extensional fractures parallel to �1, theThe orientations of fractures with staining primarily cor-175� and 60� fracture systems correspond to the shearrespond to the 25� and 175� fracture systems, and frac-fractures, and the 145� fracture system corresponds totures with no staining correspond to the 60� and 145�the stress release fractures. This implies that the exten-fracture systems (Fig. 4c).sional fractures (25�) and one of the shear fracture systemsAt the bottom of the quarry, hydraulically active frac-(175�) in combination are the main groundwater-con-tures are recognized in shallow ponds and lakes becausetrolling fracture systems. The 60� shear fracture systemthe water seeping out of the vertical fractures creates is less well developed than the 25� extensional fracturealignments of craters in the mud, which has settled in system, and the stress release fractures are more scat-the ponds. The orientations of these hydraulically active tered and seem to play a minor role.

vertical fractures correspond to the 25� fracture system The finding that the dominant flow direction in the(Fig. 4d). network of fractures is parallel with the coast is consis-

By looking at the strike/dip of fractures observed in tent with the observation that no saltwater intrusionthe image logs and cores from the two wells, both vertical was recorded in the monitoring well situated betweenand horizontal fractures were registered (see the Wulff the coast and the active part of the quarry where thenets, Fig. 5). Only one vertical fracture was observed groundwater table was lowered down to 20 m bsl.in the lower part of Well I while Well II containednumerous vertical fractures. The information concern- Results of the Infrared Thermographic Analysising the accumulation of iron oxide in fractures from the

Natural Seepagesaturated part of Well II indicates that most of thevertical fractures are hydraulically active since dissolved By performing an IRT survey on four walls of the

quarry representing two different levels, ALevel 5 to 20 m andFe is transported with reduced groundwater and is oxi-

Rep

rodu

ced

from

Vad

ose

Zon

e Jo

urna

l. P

ublis

hed

by S

oil S

cien

ce S

ocie

ty o

f Am

eric

a. A

ll co

pyrig

hts

rese

rved

.

276 VADOSE ZONE J., VOL. 4, MAY 2005

Fig. 5. Image logs from Well I and Well II, indicating fractures, flint layer, and the water table in the two wells. The strike/dip of the fracturesin the two wells are represented in Wulff nets.

Rep

rodu

ced

from

Vad

ose

Zon

e Jo

urna

l. P

ublis

hed

by S

oil S

cien

ce S

ocie

ty o

f Am

eric

a. A

ll co

pyrig

hts

rese

rved

.

www.vadosezonejournal.org 277

Fig. 6. Thermographic images of two corners in the chalk quarry seen from southeast. The upper image shows Level A (5 to 20 m), and lowerimage shows Level B (�20 to 0 m). The images are taken on a cold winter morning with the relatively warm groundwater seeping out onthe cold wall surface.

BLevel–20 to 0 m (Fig. 1), on a cold winter morning (air temper- water table observed in the two wells established 2 toature �12�C), it was possible to construct a thermal 3 m from the wall edge. Above 10 m bsl, the temperaturemap of the natural seepage of ≈8�C groundwater from of the wall adjacent to the lake is approximately thewalls (Fig. 6). The thermal map shows that the south- same at both faces and warmer than the air temperaturefacing walls are generally warmer than the east-facing of �14�C. This is the domain of the capillary fringe,walls. Additionally, the temperature generally increases which can extend more than 30 m beyond the waterwith depth, and the hot spots are aligned horizontally table due to the pore-throat diameters in chalk (Priceon the south-facing wall. et al., 1993) and is affected by the 8�C warm ground-

Since the chalks matrix permeability is negligible, the water. The same temperature pattern is seen at Level Amain flow and associated heat conduction toward the (Fig. 5), but the level of the water table and the zonewalls must be transported along the active hydraulic of the capillary fringe in the wall is unknown. The east-fractures/pathways in the chalk. The fact that the south- facing wall in Level A is also a bit colder in the upperfacing walls are warmer than the east-facing walls is in part than on the south-facing wall. This seems to be aagreement with the fracture analysis that suggested that result of some protruding chalk parts on the east-facingthe preferential flow pathways in the chalk trend north- wall, which have a larger surface area.east–southwesterly. In the unsaturated zone, these frac-ture systems will act as barriers for the capillary flow Flow-Induced Seepagebetween the matrix blocks and hereby prevent ground-

To further document the preferential flow pathwayswater flow and associated heat conduction toward thein relation to the mapped fracture systems, a detailedeast-facing walls to the same degree as for the south-investigation was performed that included borehole in-facing walls. The horizontal alignment of the hot spotsformation and IRT image series of the lower south-is only seen on the south-facing wall and is more pro-facing wall by the lake.nounced with depth. The increasing temperature with

Relative changes of the water level in the wells weredepth is a consequence of the gradient of the waterobserved while water was injected with flow rates oftable toward the wall, implying that the vertical fractures20 and/or 32 l/min into each well and observed in thewill drain as an effect of the capillary forces to theiropposite well (Fig. 3). Since the well setting, the frac-surroundings before reaching the wall surface. The out-tures, and the lowering of the water table in the quarryflow on the wall is therefore primarily observed in thedo not warrant applications of analytical solutions forhorizontal fractures, which are intersected by a verticalestimating hydraulic parameters from water level obser-fracture (seen as the hot spots). Some outflow is alsovations in wells (Duffield, 1996), we have only per-seen in association with subhorizontal flint layer and atformed a qualitative assessment. By looking directly atlocations where the low permeable chalk base impedesthe recovery data for the wells, both seem to recoververtical drainage. These observations are in agreementrelative quickly. For Well I, Fig. 3, the water seems towith the location of the hot spots at approximately 10 mdrain below the initial water table during recovery. Thisbsl and below on the wall adjacent to the lake (Fig. 5,

Level B), which also corresponds to the level of the kind of rebound effect is normally only seen in high-

Rep

rodu

ced

from

Vad

ose

Zon

e Jo

urna

l. P

ublis

hed

by S

oil S

cien

ce S

ocie

ty o

f Am

eric

a. A

ll co

pyrig

hts

rese

rved

.

278 VADOSE ZONE J., VOL. 4, MAY 2005

conductivity formations (Butler, 1998), which indicate ing and propagating the fractures in the study area,that Well I is highly hydraulic active in the lower part. regional fracture systems were suggested, and their ori-

The temperature measurements in the wells showed entations were consistent with the obtained site-specificthat injection of 13�C water in Well I did not generate fracture information. Five regional fracture systems werean increase in the water temperature in Well II. When identified. The fracture analysis based on fracture orien-starting the injection in Well II, the water temperature tation, fracture trace length, staining on fracture sur-had decreased from 13 to 10�C in Well I, and the temper- faces, and hydraulic active fractures suggests preferen-ature continued decreasing to 9�C by the end of the tial pathways in the chalk parallel to the coastline. Theinvestigation. Figure 3 also shows that 15�C water enters dominant pathway orientation was corroborated by theWell II in the beginning of Test 2. This is water that observation that no saltwater intrusion was recorded inhas stayed in the garden hose 1.5 h between the injection the active part of the quarry. Since the identification ofof water in Well I and Well II and thus has been subject the most important hydraulic fractures and thereby theto more heating by the sun. preferred fluid pathways is crucial for the development

The thermal energy from the injection well was not of a conceptual model (Bloomfield, 1996), IRT investi-noticed in the observation well, indicating that the two gation was introduced to test the suggested preferentialwells are not connected by hydraulically active fractures.groundwater flow direction and the regional fractureThese observations are corroborated by the hydraulicmodel. The suggested preferential groundwater flow di-head observations and a tracer test (using Na2SO4) (Aa-rection was confirmed by IRT where the actual outflowmand et al., 2001), also indicating that the two wells arewas mapped on the walls.not in direct hydraulic contact.

With the nearly constant temperature of the ground-The thermal images from the wall illustrating areaswater and a temperature contrast between the ambientwhere cold water (less than 16�C) is seeping onto thegroundwater and the rock of approximately 8�C, it iswarm exposure at given times show a different outflowpossible to use IRT investigations to identify fluid path-pattern from each of the wells to the surface of the wallways. This temperature contrast minimized the temper-(Fig. 7). The image series during injection in Well Iature errors introduced by factors such as sun energy,show a very late and minimal breakthrough of the coldwind cooling, and a rough wall surface, which may gen-water at spots along some of the horizontal fractures.erate an uneven heating of the wall surface. With theThese cold spots seem to be located in the lower part

of the wall and in the area where some of the vertical natural seepage wall survey, it was possible to observefractures were mapped on the wall surface. The cold and verify the suggested large-scale preferential flowspots seen at higher levels at a later time indicate a paths. At a smaller scale, the induced seepage experi-gradual saturation of the chalk. The weak temperature ment combined with borehole information showed tworesponse at the wall surface is probably caused by drain- different outflow patterns on the wall (Test 1 and 2),ing of cold water from the bottom of the Well I as which again suggested that the vertical fractures domi-suggested by the recovery data of the water table for nate the flow regime. The investigation also underlinedthis well. The image series collected during injection in the importance of having the right position and orienta-Well II shows an early and pronounced outflow on the tion of the wells to get the maximum information con-wall surface at nearly all levels. The pronounced outflow cerning the dominating fractures of the flow regime andfrom blocks of horizontal fractures in the upper part of that there are certain limitations connected with interpre-the wall is due to the presence of fractures in the upper tation of borehole data alone in fractured chalk. Individ-part near Well II that provide preferential conduits to

ual borehole records cannot provide any informationthe wall surface. The flint layer F3 seems to deliver aabout the connectivity and trace length of the fractures.pronounced amount of cold water to the wall surface asHowever, the accumulation of iron oxide in some frac-suggested by the highly conductive horizontal fracturetures registered via BHTV in Well II seems to make itfound in connection with F3 in Well II (Fig. 7). Sincepossible to outline the hydraulically active fractures inthe time period between Test 1 and 2 is 1.5 h, injectedthe well. The induced seepage experiment also suggestswater from Test 1 may contribute to the seepage inthat indirect methods, such as the well interference test,Test 2. Since the recovered water table in the wells arecannot provide detailed information of the complex flowlocated approximately 10 m bsl, the majority of theregime prevailing in fractured chalk. The weakness offractures over 10 m bsl will be unsaturated before start-the IRT method lies in the necessity of an exposure toing Test 2, but dead-end pockets in the chalk could be

filled and thereby contribute in the seepage of Test 2. map the outflow and in the lack of information concern-Given the information from the wells and the wall ing the groundwater-flow distribution in the vertical

surface, the vertical shear fractures striking 175� seem fractures. Furthermore, the method requires that thereto dominate the flow to the wall surface by acting as is a pronounced temperature contrast between the flow-preferential pathways for water flow. Only a few of the ing fluid and the rock media to delineate the flow paths.horizontal fractures associated with flint layers seem However, when these circumstances are present, an IRThydraulically active. investigation in combination with extensive fracture

analysis seems to be a valid and useful tool for distin-DISCUSSION AND CONCLUSIONS guishing between hydraulically active/inactive fractures

and for outlining preferential flow paths in the fracturedBy identifying the sources or mechanisms that arecapable of producing the stress field requisite for initiat- media.

Rep

rodu

ced

from

Vad

ose

Zon

e Jo

urna

l. P

ublis

hed

by S

oil S

cien

ce S

ocie

ty o

f Am

eric

a. A

ll co

pyrig

hts

rese

rved

.

www.vadosezonejournal.org 279

Fig. 7. Thermographic image series illustrating the temperature changes with time of the south-facing wall by the lake in the quarry (Level B,Fig. 6) under injection of cold water in Well I (Test 1, upper image series) and Well II (Test 2, lower image series).

Rep

rodu

ced

from

Vad

ose

Zon

e Jo

urna

l. P

ublis

hed

by S

oil S

cien

ce S

ocie

ty o

f Am

eric

a. A

ll co

pyrig

hts

rese

rved

.

280 VADOSE ZONE J., VOL. 4, MAY 2005

Hobbs, B.E., W.D. Means, and P.F. Williams. 1976. An outline ofACKNOWLEDGMENTSstructural geology. John Wiley & Sons, Hoboken, NJ.

This research was in part supported by the EC Environment Japsen, P. 1993. Influence of lithology and Neogene uplift on seismicvelocities in Denmark. Implications for depth. Am. Assoc. Pet.and Climate Research Programme (Contract ENV4-CT97-Geol. Bull. 77(2):194–211.0441) and in part by The Geological Survey of Denmark and

Japsen, P., and T. Bidstrup. 1999. Quantification of late CenozoicGreenland, GEUS. Thanks to Faxe Kalk A/S for access toerosion in Denmark based on sonic data and basin modelling. Bull.the Sigerslev quarry. We also thank J.C. Refsgaard of TheGeol. Soc. Den. 46:79–99.

Geological Survey of Denmark and Greenland and K.H. Jen- Jensen, J.R., E. Ramsey, H.E. Mackey, and M.E. Hodgson. 1988.sen of Geological Institute, University of Copenhagen, for Thermal modelling of heat dissipation in the Pen Branch Deltadiscussing and reviewing the manuscript. using thermal infrared imagery. Geocarto Int. 3:17–28.

Lathrop, R.G., and T.M. Lillesand. 1987. Calibration of ThematicMapper thermal data for water surface temperature mapping: CaseREFERENCES study on the Great Lakes. Remote Sens. Environ. 22:297–307.

LeDrew, E.F., and S.E. Franklin. 1985. The use of thermal infraredAamand, J., E. Adar, J. Bloomfield, A. Erichsen, I. Gale, A.S. Grader,imagery in surface current analysis of a small lake. Photogramm.K. Griffiths, H. Hotzl, P.R. Jakobsen, G.B. Kristensen, R. Nativ,Eng. Remote Sens. 51:565–573.E. Nygaard, A. Polak, B. Reichert, Z. Ronen, A.E. Rosenbom,

Liboriussen, J., P. Ashton, and T. Tygesen. 1987. The tectonic evolu-S.R. Sørensen, R. Wallach, S. Watson, A. Williams, and K. Witt-tion of the Fennoscandian Border Zone in Denmark. Tectonophy-hueser. 2001. Fracflow: Final report. Contaminant transport, moni-sics 137:21–29.toring techniques, and remediation strategies in cross European

Miljure, R. 1992. Infrared thermography finds process hot spots.fractured chalk. EC Environment and Climate programme 1994– Chem. Eng. Prog. 88(3):81–85.1998. Geol. Survey of Denmark and Greenland, Copenhagen, Moropoulou, A., and N.P. Avedelidis. 2002. Active thermography inDenmark. the assessment of repaired aircraft panels. Insight 44(3):145–149.AGEMA. 1997. Thermovision� 550 Basic I. Publ. 557 252 Ed B. FLIR Portnoy, J.W., B.L. Nowicki, C.T. Roman, and D.W. Urish. 1998. The

Syst., Portland, OR. discharge of nitrate-contaminated groundwater from developedAnderson, J.M., R.W. Duck, and J. McManus. 1995. Thermal radiome- shoreline to marsh-fringed estuary. Water Resour. Res. 34(11):

try: A rapid means of determining surface water temperature varia- 3095–3104.tions in lakes and reservoirs. J. Hydrol. (Amsterdam) 173:131–144. Price, M., M.J. Bird, and S.S.D. Foster. 1976. Chalk pore-size measure-

Avdelidis, N.P., A. Moropoulou, and P. Theoulakis. 2003. Detection ments and their significance. Water Serv. (London) 80:596–600.of water deposits and movement in porous materials by infrared Price, M., R.A. Downing, and W.M. Edmunds. 1993. The chalk as animaging. Infrared Phys. Technol. 44:183–190. aquifer. p. 267–273. In R.A. Downing, M. Price, and G.P. Jones

Bayari, C.S., and T. Kurttas. 2002. Coastal and submarine karstic (ed.) The hydrogeology of the chalk of North-West Europe.discharges in the Gokova Bay, SW Turkey. Q. J. Eng. Geol. Hy- Clarendon Press, Oxford, UK.drogeol. 35(4):381–390. Roxburgh, I.E. 1985. Thermal infrared detection of submarine springs

Bloomfield, J. 1996. Characterisation of hydrogeologically significant associated with the Plymouth Limestone. Hydrol. Sci. J. 30:185–196.Schott, J.R. 1979. Temperature measurement of cooling water dis-fracture distributions in the Chalk: An example from the Upper

charged from power plants. Photogramm. Eng. Remote Sens. 45:Chalk of southern England. J. Hydrol. (Amsterdam) 184:355–379.753–761.Butler, J.J. 1998. The design, performance, and analysis of slug tests.

Selley, R.C. 1976. An introduction to sedimentology. AcademicCRC Press, Boca Raton, FL.Press, London.Commission on Geosciences, Environment and Resources. 1996.

Shiratori, M., Y. Takahashi, and Y. Qiang. 1994. Application of infra-Rock fractures and fluid flow: Contemporary understanding andred thermography to detection of cracks. JSME Int. J., Ser. Aapplications. Natl. Acad. of Sci., Washington, DC.37(3):296–302.Davies, P.A., L.A. Mofor, and M.J.V. Neves. 1997. Comparisons of

Surlyk, F. 1997. A cool-water carbonate ramp with bryozoan mounds:remotely sensed observations with modelling predictions for theLate Cretaceous–Danian of the Danish Basin. p. 293–307. In N.P.behaviour of wastewater plumes from coastal discharges. Int. J.James and J.A.D. Clarke (ed.) Cool-water carbonates. Special Publ.Remote Sens. 18:1987–2019.56. Soc. for Sedimentary Geol., Tulsa, OK.Downing, R.A., M. Price, and G.P. Jones. 1993. An aquifer at risk. Surlyk, F., and E. Hakansson. 1999. Maastrichtian and Danian strata

p. 267–273. In R.A. Downing, M. Price, and G.P. Jones (ed.) The in the southeastern part of the Danish Basin. p. 29–58. In G.K.hydrogeology of the chalk of North-West Europe. Clarendon Press, Pedersen and L.B. Clemmensen (ed.) Proc. Reg. Eur. Meet. ofOxford, UK. Sedimentology, 19th, Copenhagen, Denmark. Excursion A3. Univ.

Duffield, G.M. 1996. AQTESOLV for Windows. HydroSOLVE Inc., of Copenhagen, Copenhagen, Denmark.Reston, VA. Torgersen, C.E., R.N. Faux, B.A. McIntosh, N.J. Poage, and D.J.

Frykman, P. 2001. Spatial variability in petrophysical properties in Norton. 2001. Airborne thermal remote sensing for water tempera-Upper Maastrichtian chalk outcrops at Stevns Klint, Denmark. ture assessment in rivers and streams. Remote Sens. Environ.Mar. Pet. Geol. 18:1041–1062. 76:386–398.

Garrett, A.J., and D.W. Hayes. 1997. Cooling lake simulations com- Zemanek, J., E.E. Glenn, L.J. Norton, and R.L. Caldwell. 1970. For-pared to thermal imagery and dye tracers. J. Hydraul. Eng. mation evaluation by inspection with the borehole televiewer. Geo-

physics 35:254–269.123(10):885–894.