geophysical and geological characterisation of karst hazards in urban environments, application to...

8/19/2019 Geophysical and Geological Characterisation of Karst Hazards in Urban Environments, Application to Orléans ,France

http://slidepdf.com/reader/full/geophysical-and-geological-characterisation-of-karst-hazards-in-urban-environments 1/12

Introduction

Among the drawbacks encountered in urban and sub-urban planning, the existence of subsurface karsts is

both common and particularly problematic. Difficultiestake the form of the risk of collapse or damage to con-struction through subsidence and increased developmentcosts due to geotechnical conditions that are not easily

Pierre ThierryNicole DebebliaArdnand Bitri

Geophysical and geological characterisationof karst hazards in urban environments:application to Orle ans (France)

Received: 12 February 2004Accepted: 6 April 2004Published online: 21 October 2004 Springer-Verlag 2004

Abstract Surface or shallow karstconstitutes a significant problem formany cities, including Orle ´ ans,France. However, the detection of cavities in an urban environmentpresents numerous difficulties (buriednetworks, reworked soils, geophysi-cal noise, etc.). A methodology hasbeen developed to respond to thisneed based on the integration of fourcomplementary methods: geologicaland geostatistical analysis of existingborehole descriptions, microgravi-metry, spectral analysis of surfacewaves (SASW) and ground penetrat-ing radar (GPR). This combinationof different methods, applied to a testsector in the city of Orle ´ ans and vali-dated by control borings, made itpossible to locate theprincipal karsticconduits beneath the study area andidentify a zone of mechanical weak-ness associated with one such feature.It also revealed that the presence of buried networks does not create sig-nificant gravity anomalies.

Keywords Urban environment Æ

Karst Æ Microgravimetry Æ SASW Æ

Orle ´ ans Æ Geostatistics

Re ´ sume ´ L’existence d’un karst su-perficiel ou profond pose un prob-

le ` me grave a ` beaucoup de villes, dontla ville d’Orle ´ ans, en France. En ef-fet, la de ´ tection de cavite ´ s dans unenvironnement urbain pre ´ sente denombreuses difficulte ´ s (re ´ seaux en-terre ´ s, sols remanie ´ s, bruit de fondge ´ ophysique, etc.). Pour re ´ pondre a `ce besoin, on a de ´ veloppe ´ une me ´ th-odologie basee sur l’integration dequatre methodes complementaires :l’analyse geologique et geostatistiquedes donnees des sondages existants,la microgravimetrie, l’analyse spect-rale des ondes de surface et lege ´ oradar. Cette combinaison dediffe ´ rentes me ´ thodes a e ´ te ´ applique ´ ea ` un secteur-test de la ville d’Orle ´ anset valide ´ e par des sondages de con-trole. Elle a permis de localiser lesprincipaux conduits karstiques de lazone e ´ tudie ´ e et d’identifier un secteurde faibles proprie ´ te ´ s me ´ caniques as-socie ´ a ` une telle particularite ´ . Cela aaussi montre ´ que la pre ´ sence de re ´ -seaux enterre ´ s ne cre ´ e pas d’anoma-lies gravimetriques significatives.

Mots-cle ´ fs Environnement urbain Æ

Karst Æ Microgravime ´ trie Æ

Ge ´ oradar Æ Orle ´ ans Æ Ge ´ ostatistique

Bull Eng Geol Environ (2005) 64: 139–150DOI 10.1007/s10064-004-0247-4 O R I G I N A L P A P E R

P. Thierry (&) Æ N. Debeblia Æ A. BitriBRGM, BP6009 45060,

Orle ´ ans Cedex 2, FranceE-mail: [email protected].: +332-386-43264Fax: +332-386-43399



anticipated such as variable bearing capacity. In somecases, cavities may constitute a vector facilitating themovement of pollutants into the groundwater, thusendangering drinking water resources. Karstic featurescan take the form of significant cavities several metreswide or networks of more limited open fractures (White1988). They can also include areas of surficial dissolution

on the uppermost zone of exposed karstified rocks(Klimchouck 1997). This epikarst is often filled byunconsolidated material.

The detection of karstic cavities in urban environ-ments and the evaluation of the degree of danger theyrepresent are thus a major issue for numerous cities inFrance and elsewhere in the world. This is particularlytrue for the city of Orle ´ ans in France.

In addition to the use of archives, geophysicalmethods can play an effective role in the detection andgeometric characterisation of karstic features. Densityand resistivity parameters are known to be very goodmarkers of the alteration and fissuring characteristic of

drainage zones associated with epikarst while electricbehaviour may be amplified by the presence of shallowgroundwater and strong water circulation. Microgravi-metry, electrical methods and electromagnetic imageryare proven methods for detecting karstic circulation,providing anthropogenic interference permits their use(Guerin and Benderitter 1995; Kaufmann 2000; Ioanniset al. 2002; Van Schoor 2002). Despite significant noiseand weak penetration due to intensive surficial fractur-ing, karst reservoirs have been successfully investigatedby seismic tomography (Rossi et al. 2002). This methodhas led to a description of the principal geologicalinterfaces present, namely soils and a poorly consoli-

dated cover, fractured limestone and unfractured lime-stone. Ground penetrating radar (GPR) is particularlysuitable where limestone formations crop out directlyand where soils or an argillaceous cover that absorbsradar waves are absent or highly discontinuous (Al-Fares et al. 2002). In this case, the structures charac-teristic of the karst environment, epikarst, fracturedzones, compact limestone and deep conduits can bedifferentiated on the radar section to a depth of about20 m. Unfortunately, such conditions are not common.The soils of the Orle ´ ans region are far too argillaceousfor this method to be effective in revealing deep con-duits. In addition, the urban setting is unsuitable forelectric imagery methods. The direct detection of deepconduits thus becomes problematic. The gravity methodis one of the few methods that can be considered inurban environments, providing that the size and depthof the voids are favourable (Crawford 2000; Beres et al.2001; Rybakov et al. 2001).

Nevertheless, no single method can guarantee theunambiguous detection of karst structures and thecombination of several geophysical methods is alwaysindispensable in obtaining a correctly constrained

model. For example, the exploration of a karst in theSwiss Jura was conducted by combining GPR, whichprovides a precise image of the most surficial voids andstructures, with gravimetry, which is sensitive to thepresence of deeper and larger heterogeneities (Beres et al.2001). Due to the weak penetration anticipated for GPRin a context with surficial argillaceous formations, it

seemed worthwhile to examine the possibility of com-bining gravimetry with the spectral analysis of surfacewaves (SASW). In addition, the characteristics of urbanisation (artificial nature of soils, developed sites,buried networks, geophysical noise) greatly increase thedifficulties encountered during the investigation andinterpretation of geophysical methods. It was thus nec-essary to find a way of overcoming these difficultiesthrough the integration of four complementary methods:geological and geostatistical analysis of existing boreholedescriptions, microgravimetry, SASW and GPR.

The study area

The study was conducted in Orle ´ ans, a city of about270,000 inhabitants in a 330-km2 area on the banks of the Loire River some 100 km to the south of Paris. Themarly-limestone substratum underlying the entire con-urbation is highly karstified. As proof, the LoiretSprings, amongst the most important karst exsurgencesin France, have an average discharge of 0.9 m3/s (Al-be ´ ric 2001, 2004). The significant discharge circulatingthrough this karst network constitutes the principaldrinking water resource of the city. Certain cavities arevery near the surface and a sinkhole recently appearedwithin the campus of the University of Orle ´ ans (Fig. 1).

The cavities also generated major unforeseen expensesduring the construction of a tramway line and a bridgeover the Loire.

Fig. 1 Sinkhole within the university campus, Orle ´ ans la Source, inMarch 1986 (photo: P. Albe ´ ric, ISTO—CNRS-University of Orle ´ ans)

140



The study was conducted along a principal profile1 km in length near the Parc Floral of Orle ´ ans LaSource where the Loiret Springs are located (see Fig. 2).This sector was chosen for several reasons:

– Urbanization of the ground (significant anthropo-genic artificial reworking, road surfacing, presence of numerous buried networks)

– Presence of significant karstic conduits partiallymapped by cave divers

– Sufficiently open space to facilitate measurement – A lack of buildings—the only difference from a classic

urban environment—makes it possible to better definethe effects of human activities on the geophysicalsignals.

Geological context

The study area is located on either side of a slope that isan ancient terrace on the margin of the alluvial plain of the Loire. Relief is very low, 95 m NGF on the alluvialplain, rising to about 100 m for the ancient terrace.

Five different geological formations are recognized(Gigout 1970). From youngest to oldest, they are:

– Present-day formations: principally loam or topsoil,as well as man-made fill and reworked colluvium.

Various types exist, but sandy or sandy-argillaceousare the most common. They are often thin, generallyless than a metre.

– Recent Loire alluvium: grain size ranges from fine

sand (overflow loam) to pebbles. The alluvium issiliceous. The pebbles are flint, quartz, sandstone andother siliceous rocks. The modern alluvium generallybecomes finer with distance from the present-dayriverbed.

– Ancient terrace alluvium dating from the Riss (likelyage). This is composed of siliceous and argillaceousmaterial with grain sizes from clay to pebbles and evenboulders; no limestone is present. The bedding islenticular with horizons of clay-deficient gravelly sand(‘‘red sand’’) and clay or pebble lenses. In the studyarea, this formation is a polygenetic alluvial depositwith the alluvium commonly composed of a fine sandyfacies.

– Sands and marls of the Orle ´ ans region (Burdigalian):these are described from various boreholes andunderlie the ancient alluvium in the southern half of the study area. This formation has a lenticular bed-ding containing lenses of sand and green or white marlwith limestone concretions.

– The Beauce Formation, of Aquitainian age, consti-tutes the substratum of the conurbation. It is alacustrine limestone with marly lenses and bands of

Fig. 2 Location of the studyarea; extract of the 1/50,000scale geological map ‘‘La Ferte ´ -Saint-Aubin’’ (black lineshows location of the princi-pal profile of the study and thetwo secondary profiles; starsshow known boreholes)

141



millstone grit. The Beauce Formation can reach athickness of more than 50 m in the study area. In itsuppermost zone, the Formation is essentially repre-sented by marl several metres thick, before gradinginto compact limestone. Although marl can be a rec-ognized facies of the Aquitainian Formation, it isoften a product of weathering and reworking: an

agglomerate of limestone and marly-limestone nod-ules, with elements of millstone grit within marly clay.This frost-fractured colluvial deposit developed at thesurface at the expense of the Beauce Formation. Theselimestone and marly-limestone formations have pro-duced significant karst dissolution features. Anabundance of karst depressions has resulted fromthem including: avens, chasms and sinkholes, swallowholes and dolines. Karst dissolution also occurs at thesurface that separates the Beauce Formation and thealluvium (or the base of the Burdigalian Marl). Thissurface is very irregular and pockmarked.

Geostatistical modelling

The geological understanding of a site is based on thematerial identified in a limited number of boreholes. Tointerpolate or extrapolate this information to cover anentire study area, geological modelling can be used: thesurfaces (base and top) of the different geological for-mations are calculated as regular grids. These grids givethe elevation Z of the surface at each node of the grid.These nodes are spaced at constant intervals based onthe X and Y axes.

The geological modelling developed for this study

uses geostatistical tools (Chiles and Delfiner 1999). Itshould be noted that in any problem requiring automaticinterpolation techniques, the technique itself has a majorinfluence on the quality of the results; hence, the choiceof interpolation parameters cannot be neglected if thegoal is to make the best use of the available information.

Two major reasons justify the use of a geostatisticalapproach:

– It is possible to characterise the spatial variability of the phenomenon under study, in this case topography,using the variable ‘‘elevation’’. The tool used is thevariogram, which synthesises the structure of thevariable. In an established direction, it indicates thestandard deviation of the values at two points X andX +h as a function of the distance h; it providesinformation on possible anisotropies and on the var-iable’s degree of regularity.

– It facilitates the resolution of the optimal interpola-tion problem, i.e. of the optimal estimate of an un-known value Z (x0), on the basis of the availableexperimental data.

Kriging was used for the interpolation, as it is theonly technique that simultaneously provides the optimalestimate and the precision of this estimate, characterisedas the kriging standard deviation (which represents a

measurement of the error of the estimation).It is necessary to emphasise that surfaces calculated inthis way represent a statistically plausible ‘‘image’’ of reality. The representation makes it possible to betterapproximate the situation, in particular by showing thegeometric tendencies and detecting possible discrepan-cies in the data.

For the study area, descriptions of 50 borings werefound in the different archives. The major source was theBSS (subsurface databank), a national database of borings managed by the BRGM. Of these, two wereunacceptable due to inaccurate information hence thevariogram analysis was done without integrating these

two borings (Fig. 3). It should be noted that this is avery reduced statistical population for this type of analysis.

Under these conditions, the variogram can be ad- justed on the basis of a spherical model and the resultobtained is satisfactory. Note the nugget effect of

Fig. 3 Variogram of the top of the Tertiary

142



±0.5 m. This confidence interval, expressed as the valueof the nugget effect, indicates elevation variations of thetop of the limestone caused by phenomena such aserosion or weathering.

In addition, it is possible to calculate a predictivegeological cross-section of the principal study profile

(Fig. 4), along which a relatively constant thickness of about 5 m is observed for the surficial deposits. Krigingstandard deviation calculated along the same profileranged between 1.5 and 1 m.

Choice of methods, array and exploration strategy

Predictive gravity modelling of an explored conduitshowed that its effect, although weak, is detectable. Asingle feature can cause an anomaly of about 15 lGal.The modelling also shows that a spacing of 5 m is suf-ficient for this type of exploration. This was verified inthe Parc Floral by conducting two gravity calibration

profiles to intersect known conduits (Fig. 2). Subse-quently a profile was explored outside the Parc Floralperpendicular to the assumed extension of the exploredkarstic conduits. There was a risk that numerous buriednetworks known to be present at the site, but whoselocations are imprecise, would distort the anomalies.Radar profiles were used to locate these networksaccurately. Along these same profiles, gravity measure-ments on a smaller, 2 m grid were undertaken to helpverify the effect of the different structures detected byradar and to estimate the potential interference they maycause. The interpretation of gravity anomalies is notunique. In order to reduce the ambiguity, gravimetry

coupled with seismic SASW was used to detect possiblezones of weakness caused by the presence of karsticcavities.

Results of the geophysics

The uncertainty of the gravity measurements, inferredfrom a statistical analysis of deviations observed at re-peated stations, was on average 7 lGal, lower than the

anomaly expected from an isolated karstic conduit. Onthe other hand, ‘‘geological noise’’ cannot be estimatedby gravity measurement alone.

In fact, the two profiles studied in the La Source ParcFloral showed that negative gravity anomalies arecommonly associated with known karstic conduits

(Fig. 5). These anomalies have amplitudes and dimen-sions that are greater than anomalies modelled for asingle feature. Other anomalies with similar character-istics were also shown where no feature had been ex-plored. These anomalies may be explained by thepresence of systems that are more complex than antici-pated, characterised by the juxtaposition of severalconduits and the association of decompressed zones.

The gravity-radar profiles showed that the influenceof buried networks on measured gravity anomalies isnegligible as the few gravity effects observed in correla-tion with radar anomalies only slightly exceeded thestatistical uncertainty of the measurements (Fig. 6).

Fig. 4 Geological cross-sectionbased on geostatistical model-ling and calculations

Fig. 5 Location map and results of the two gravity profiles runover the explored Loiret karst (Black line represents karst system;white stars represents negative residual anomaly; grey stars indicatepositive residual gravity anomaly)

143



Except for the settling tank—the source of a 10 lGalanomaly—the conduits identified by the radar methoddid not create significant gravity anomalies.

Along the principal profile outside the park, twosignificant zones of anomalies were detected in the carpark: one at the slope along the extension of the

Fig. 7 Principal gravity profile

in the car park of Parc Floral.The control borings are repre-sented by vertical segments indashed line (borings that did notencounter voids) and solid line(borings that encounter voids)

Fig. 6 Example of a gravityand radar profile run over someburied networks (EDF electric-ity, GDF gas, CGE water, vatsettling tank)

144



anomalies detected inside the park and the other at thebase of the car park, along the assumed axis of one of the explored features (Fig. 7).

Here also, the anomalies were wider and more intensethan those expected for a single karst conduit having thecharacteristics of the tunnels explored by divers. Theyare probably the sum of the effects of juxtaposed karst

conduits, to which are probably added the effects of zones of weakness and uncompacted ground producedby karstic circulation. In fact, the two cored boreholes(SC1 and SC2) located near the two principal negativegravity anomalies in this profile encountered severalsuccessive levels of water-saturated open voids. Thesevoids were found at depths of 5–20 m. The modelling atborehole SC2 showed that these voids do not explain thetotal anomaly observed. The effects of the deep exploredconduit not encountered in the borehole must be added,in addition to that of other unrecognised voids oruncompacted zones (Fig. 8). According to the model-ling, the width of the low-density area disturbed by

karstic drainage could be in the order of 60 m.On the same profile outside the park, the SASWmethod was used to measure a vertical profile of shearwave velocity and to visualise zones of ‘‘weak’’mechanical properties such as uncompacted areas or

possible sinkholes in the process of formation. TheSASW method is a rapid means of determining rigidity-depth profiles of surficial formations, which can com-plement or partially replace pressure tests in boreholes(Matthews et al. 1996). Indeed, in the case of horizontalplanar beds, SASW can be used to evaluate the verticalprofile Vs of shear waves. These Vs velocities are directly

related to the maximum shear modulus (Gmax=q Vs2

,q being the volumetric mass of soil), which after cali-bration can be correlated to the pressure modulus as it ismeasured during geotechnical drilling tests.

Initial reconnaissance (Fig. 9) was conducted on theprincipal profile using the following acquisition param-eters:

– 24 trace acquisition array – 5 m distance between geophones – 50–200 m of distance between shots.

This reconnaissance profile identified an anomalouszone that corresponds to a negative anomaly located to

the north of the profile. The spacing was then reduced inthis sector to obtain a very high-resolution image.During this second phase, the array was made up of 24traces, the spacing of geophones was 2 m and the dis-tance between shots was 4 m.

These high-resolution data show a disturbed zonesuperimposed on the gravity anomaly and characterised

Fig. 8 Gravity modelling at borehole SC2

145



by the appearance of several dispersion modes(Fig. 10b). This last feature probably represents suddenvariations in rigidity.

The profile also shows a 70 m wide zone character-

ised by attenuation of the continuity of seismic markers(Fig. 11), probably due to the fracturing of epikarst.The karstification also seems to distort the Vs velocityfield by the introduction of velocity inversions. Theanomalous gravity and seismic zones have the samewidth, the seismic anomaly being slightly offset to thesouth with respect to the gravity anomaly. This offsetcan perhaps be explained by a local obliqueness of thekarst structure, which was demonstrated by a parallelprofile about 15 m to the west of the principal profile;this obliqueness would affect the two methodsdifferently.

Comparison between geophysical resultsand geological modelling

Figure 12 shows a comparison of the results obtained bythe different methods along the principal profile:

– Results of gravity measurements (Bouger and residualanomalies)

– Velocity section of surface waves directly belowSASW reconnaissance arrays

– Simplified geological cross-sections of destructive andcored boreholes

– Pressure measurements recorded in the destructiveboreholes

As previously noted, the boreholes confirmed thelow-density anomalies with cavities encountered in coredboreholes SC1 and SC2. A very good overall agreementwas found between the geological model and the SASWprofile. This allowed a preliminary characterisation of the different formations in terms of Vs. Table 1 shows

Fig. 10 Dispersion diagrams associated with shots recorded at theParc Floral site along the high-resolution profile. a southern shots,b central part near the gravity anomaly, c northern shots)

Fig. 9 Vs section resulting fromthe interpolation of the resultsof the SASW inversion on theprincipal reconnaissance pro-file. Location of the profile atvery high resolution

146



the Vs ranges as a function of the different facies. Itshould also be noted that for some facies in the massivelimestone, Vs values can reach or exceed 1,000 m/s.

There appears to be a clear difference in the SASWresponse along the two negative gravity anomalies. For

the southern anomaly, the elevated values of Vs (up to700 m/s) extend to a depth of about 10 m, correspond-ing to a limestone of relatively good mechanical prop-erties. The northern anomaly, on the contrary, ischaracterised by a significant interval, about 15 m, of soils of low velocity (approximately 400 m/s). Pressuretests conducted on the top of the Beauce marl and marlylimestone yielded values for the E pressure modulus of 1.5–7 MPa. In comparison, the values of this modulusmeasured nearby, away from the gravity anomalies, atthe same depths and in the same formations, range from7–35 MPa. Similar correlations between the presence of incompetent weathered material characterised by anelevated depth of refusal and low-density/conductivezones have been detected in the Tournaisis directlyabove particular palaeo-karst features in which karstcollapses are concentrated (Kaufmann 2000).

As a consequence, a high-resolution SASW profilewas undertaken on this northern anomaly that con-firmed the range of Vs values observed for the alteredupper part of the Beauce limestone, of the order of 300 m/s. It also showed notable degradation of thesurficial formations at the northern gravity anomaly.

Discussion of results

Microgravity data

In general, gravity anomalies are considered significantif they have an amplitude greater than the measure-

ment uncertainty and are observed at two or threestations, or if they have an amplitude two to threetimes the measurement uncertainty observed at a sin-gle station. The use of an expanded uncertainty ob-tained by multiplying the standard uncertainty by acoverage factor of two or three is recommended byinternational measurement authorities such as theBIPM (International Bureau of Weights and Mea-sures) and the CIPM (International Committee onWeights and Measures). For gaussian error distribu-tion, it is assumed that a coverage factor of twoprovides a 95% confidence level, whereas a factor of three provides a 99% confidence level (guidance on

the expression of uncertainty in measurement, Inter-national Organisation of Normalisation, 1995).

In the Parc Floral, with a standard uncertainty of 7 lGal resulting from repeated gravity measurements,anomalies having an amplitude greater than 15– 20 lGal can be considered and significant features havebeen detected by the three profiles. Anomalies with lowto medium amplitude (15–30 lGal), sometimes at thedetection limit, are located directly above karst con-

Fig. 11 Reconnaissance of theLoiret karst by very high-reso-lution SASW and gravimetry. agravity anomaly, b SASW im-age, c Vs velocity section

147



duits recognised by SSL1 divers and very probablyindicate the presence of these conduits. In the car park,the two most intense anomalies (50 and 30 lGal) cor-respond well to voids empirically confirmed by twocontrol boreholes, while the three other boreholes, lo-cated on positive anomalies, have encountered unfrac-tured rock with no cavities. Other anomalies have beendetected where it has been impossible to verify theexistence of voids because they have not been drilled.Numerical simulation of gravity anomalies shows thatan isolated karst conduit with the dimensions and

depth of conduits explored by cave divers cannot itself explain the entire 30-lGal-anomaly measured north of the car park. It is necessary to add other cavities ordecompressed ground to this conduit to account for theanomaly. It seems that the gravity signatures observedrepresent epikarst rather than isolated deep karstic

conduits whose effect here is at the limit of the meth-od’s detection threshold.

In addition to the effects of voids, the amplitude,shape and position of gravity anomalies are also influ-enced by topography (if the topographic corrections areinsufficient) and by geological heterogeneity. In thiscase, precise topographic corrections have eliminated thefirst source of uncertainty. In contrast, for some geo-metric configurations of surficial heterogeneity, theminimum gravity anomaly may not be located exactlyabove the cavity, or may even be masked. In addition,ambiguities regarding the interpretation of a gravityanomaly are always possible. For example, an identical80-lGal-gravity anomaly can be created either by acavity 20 m in diameter and 3 m high, at a depth of

1 Subaquatic Speleology of the Loiret.

Table 1 VS ranges for the principal lithological facies encountered

Facies VS

Anthropogenic fill 200–350 m/sSandy-argillaceous alluvium 250–350 m/sWeathered marly limestone 300–500 m/sMarly limestone 500–700 m/sLimestone >600 m/s

Fig. 12 Juxtaposition of geophysical and geological results alongthe principal profile

148



10 m, for a density contrast of )2; or by a sandy zonethat extends from the surface to a depth of 20 mencompassing a density contrast of )0.2. This is why anysignificant gravity anomalies must always be checked byborings.

Other parameters also influence the capacity of amethod to detect a void, in particular the measurement

interval. For example, a well 20 m deep and 1.5 m indiameter produces a gravity anomaly of about 50 lGal,but over a very short distance (about 2 m). This is whypreliminary modelling is indispensable in the selection of a suitable measurement array. In this case, predictivemodelling showed that an interval of 5 m was appro-priate for the detection of deep tunnels with broad lat-eral extent and the choice was made to measure gravityanomalies along three profiles perpendicular to knownkarst features. This allowed comparison of measure-ments of the first two profiles with speleological surveysin order to estimate the gravity signature of the karstfeatures and to propose an extension of known conduits

along the third profile.

Radar data

Detailed radar and gravity profiles to estimate interfer-ence caused by urban networks and infrastructure,showed that the conduits present are too small to have adetectable gravity influence. Thus, they will not interferewith the effects being studied. Only the oil-settling tank,whose dimensions are larger, caused an anomaly of theorder of 10-lGal, close to the threshold of the precisionof the method and thus capable of slightly distorting themeasurements. Conversely, if a developer wanted to

determine the location of certain shallow networks, theradar method is an advantageous option that has beenused for a long time.

SASW imagery

The results obtained by SASW have been fully con-firmed by control borings installed along the principalprofile (cored boreholes and pressure measurementborings); this efficiency was partly due to the tabularnature of the geological formations.

In addition, this method offers two advantages foruse in an urban environment:

– The use of gimbals (sensors placed on the ground)means that it is totally non-intrusive. The road sur-facing thus remains intact and the implementationtime is likewise decreased.

– It is not necessary to disrupt traffic. The measure-ments can be done between the passage of two vehi-cles (in the case of average traffic density).

Conclusions

The results obtained from this study show the advan-tages of an approach integrating geological and geo-statistical analysis with different methods of geophysicalmeasurement in an urban environment in order todetermine the geometric and mechanical characteristics

of a subsurface composed of a substratum of tabularand karstified marly-limestone covered by alluvium andman-made formations.

– Geological and geostatistical analysis of existing ar-chive data enables a deductive approach to the localgeological context (thickness and position of differentformations). Such analysis constitutes a frameworkthat is extremely useful for the interpretation of geo-physical measurements.

– Microgravimetry makes it possible to locate theprincipal karst conduits. Quantitative interpretationof the results can be made on the basis of knowledgeof the geometry of the geological formations and

information provided by SASW (identification of fractured zones forming a link between the karstfeatures and the surface).

– SASW can be used to represent the vertical profile of shear wave velocity through a cross-section of thesubsurface. Here again the results of the geologicaland geophysical analysis provide valuable support forthe interpretation of the results and the identificationof sensitive zones. Very high-resolution profiles onsensitive zones lead to very precise localization of thedifferent disturbed areas.

– Radar allows the very precise location of buried sys-tems.

– The comparison of the SASW profiles with the resultsof the geostatistical analysis aids in the validation of the model and may allow its use in zones where thereis very limited information or too much variability.

The results obtained in Orle ´ ans during this first phaseof the study thus made it possible to:

– Specify, at several points, the actual position of thetop of the Beauce limestone with respect to the cal-culated model

– Determine the position of two principal karst featuresthat supply the Loiret Spring on topographic surveysprepared by SSL divers

– Detect a zone of mechanical weakness, marked byboth a notable reduction of the mechanical quality of the limestone and a probable degradation of thematerial

Acknowledgements The spelaeological divers of the SSL (Sub-aquatic Spelaeology of the Loiret), affiliated with the French

149



Federation of Spelaeology, kindly made available the topographicsurveys of the principal karst conduits. In addition, BernardBourgine contributed his expertise in geostatistics and P. Albe ´ ric

and Mr. Lepiller contributed their thorough understanding of thehydrogeology of the Loire Valley. Our thanks go also to R. Steadfor translating the initial text.

References

Albe ´ ric P (2001) L’inversac (perte-e ´ mer-gence) de la re ´ surgence du Bouilon(source du Loiret, France). Sci TechEnviron 13:1–4

Albe ´ ric P (2004) River backflooding into akarst resurgence (Loiret, France). JHydrol 286:194–202

Al-fares W, Bakalowicz M, Guerin R,Duckan M (2002) Analysis of the karstaquifer by means of a ground pene-trating radar (GPR): example of theLamalou area (He ´ rault, France). J ApplGeophys 51(2–4):97–106

Beres M, Luetscher M, Olivier R (2001)Integration of ground-penetrating radarand microgravimetric methods to map

shallow caves. J Appl Geophys 46:249– 262

Chiles JP, Delfiner P (1999) Geostatistics,modelling spatial uncertainty. Wiley,New York, 693 pp

Crawford NC (2000) Microgravity investi-gations of sinkhole collapses underhighways. Proc of the 1st Int Conf onthe application of geophysical method-ologies to transportation facilities andinfrastructures, 11–15 December 2000,St. Louis, Missouri

Gigou M (1970) Notice explicative de lacarte ge ´ ologique N398 ‘‘La Ferte ´ SaintAubin’’ a ` l’e ´ chelle 1/50,000. BRGM,Orle ´ ans

Guerin R, Benderitter Y (1995) Shallowkarst exploration using MT-VLF andDC resistivity methods. Geophys Pros-pec 43:635–653

Ioannis FL, Filippos IL, Bastou M (2002)

Accurate subsurface characterizationfor highway applications using resistiv-ity inversion methods. J Electric Elec-tron Eng 1:43–55

Kaufmann O (2000) Les effondrementskarstiques du Tournaisis : gene ` se,e ´ volution, localisation, pre ´ vention.The ` se 3e ` me cycle, Faculte ´ Polytech-nique de Mons, Mons, Belgique, 349 pp

Klimchouck A (1997) The nature andprincipal characteristics of epikarst.Proc 12th Int Congr Speleology, August1997, La Chauds-de-Fonds, Switzer-land, pp 306

Matthews MC, Hope VS, Clayton RI(1996) The use of surface waves in thedetermination of ground stiffness pro-files. Geotech Eng 119:84–95

Rybakov M, Goldshmidt V, Fleischer L,Rotstein Y (2001) Cave detection and 4-D monitoring: a microgravity case his-tory near the Dead Sea. Leading Edge20(8):896–900

Rossi G, Picotti S, Baradello L, Bolcato S,Bratus A, Wardell N (2002) Geophysi-cal integrated approach to the study of an aquifer in a Karstic area. SEG ex-panded abstracts, SEG 72nd AnnMeeting, October 2002, Salt Lake City,pp 1480–1483

Van Schoor, M( 2002) Detection of sink-holes using 2D electrical resistivityimaging. J Appl Geophys 50:393–399

White WB (1988) Geomorphology andhydrology of karst terrains. OxfordUniversity Press, Oxford, 464 pp

150

geophysical and geological characterisation of karst hazards in urban environments, application to...

Documents