Efficiency of Joint Use of MRS and VES to Characterize

ics xx (2006) xxx–xxx

+ MODEL

www.elsevier.com/locate/jappgeo

ARTICLE IN PRESS

Journal of Applied Geophys

Efficiency of joint use of MRS and VES to characterizecoastal aquifer in Myanmar

J.M. Vouillamoz a,⁎, B. Chatenoux b, F. Mathieu c, J.M. Baltassat c, A. Legchenko a

a Institut de Recherche pour le Développement, Indo-French Cell for Water Science, Indian Institute of Science, 560012 Bangalore, Indiab Action contre la Faim, 4 rue Niepce, 75014 Paris, France

c Bureau de Recherche Géologiques et Minières, BP 6009, 45060 Orléans Cedex, France

Received 30 August 2005; accepted 30 June 2006

Abstract

The productivity and the water quality of coastal aquifers can be highly heterogeneous in a complex environment. Thecharacterization of these aquifers can be improved by hydrogeological and complementary geophysical surveys. Such an integratedapproach is developed in a non-consolidated coastal aquifer in Myanmar (previously named Burma).

A preliminary hydrogeological survey is conducted to know better the targeted aquifers. Then, 25 sites are selected tocharacterize aquifers through borehole drillings and pumping tests implementation. In the same sites, magnetic resonancesoundings (MRS) and vertical electrical soundings (VES) are carried out. Geophysical results are compared to hydrogeologicaldata, and geophysical parameters are used to characterize aquifers using conversion equations. Finally, combining the analysis oftechnical and economical impacts of geophysics, a methodology is proposed to characterize non-consolidated coastal aquifers.

Depth and thickness of saturated zone is determined by means of MRS in 68% of the sites (evaluated with 34 soundings). Theaverage accuracy of confined storativity estimated with MRS is ±6% (evaluated over 7 pumping tests) whereas the averageaccuracy of transmissivity estimation with MRS is ±45% (evaluated using 15 pumping tests). To reduce uncertainty in VESinterpretation, the aquifer geometry estimated with MRS is used as a fixed parameter in VES inversion. The accuracy ofgroundwater electrical conductivity evaluation from 15 VES is enough to estimate the risk of water salinity. In addition, themaximum depth of penetration of the MRS depends on the rocks' electrical resistivity and is between 20 and 80 m at the study area.© 2006 Elsevier B.V. All rights reserved.

Keywords: Aquifer; Magnetic resonance soundings; Vertical electrical soundings; Conversion equation; Hydrogeophysics; Myanmar

1. Introduction

Coastal aquifers are of great relevance to humanneeds because coastal areas are often densely populated.However, a wide range of reservoir productivity can

⁎ Corresponding author. Tel.: +91 99 80 47 41 03.E-mail address: [email protected]

(J.M. Vouillamoz).

0926-9851/$ - see front matter © 2006 Elsevier B.V. All rights reserved.doi:10.1016/j.jappgeo.2006.06.003

exist within the same system that can be highlyheterogeneous according to its sedimentation history.Moreover, coastal aquifers can deliver salty water andsea water intrusion is always a risk. Finally, the mainquestions encountered by hydrogeologists working inthis field concern both the aquifer hydrogeologicalproperties and the water quality.

The main aquifer characteristics are usually knownby hydrogeologists from in situ surveys. One of the most

APPGEO-01605; No of Pages 13

2 J.M. Vouillamoz et al. / Journal of Applied Geophysics xx (2006) xxx–xxx

ARTICLE IN PRESS

reliable ways to gather information is the directobservation of the medium (e.g., geomorphology, fieldgeology, etc.) and drilling exploration boreholes that areused to check hydrogeological assumptions. Aquiferhydrogeological properties such as storativity andtransmissivity are usually quantified thanks to pumpingtests interpretation. A borehole and at least oneobservation well needs to be drilled to set up hydraulictests, and pumping operations can last several days.Hydraulic tests are time and money consumingcompared to non-invasive surface geophysics that canprovide rapid and low cost dense data coverage, but stillat a less reliable hydrogeological output.

Electrical resistivity methods are widely used foraquifers characterization because of the link that existsbetween the electrical resistivity of the subsurface, therock water content and the water salinity. Recentlydeveloped magnetic resonance sounding method (MRS)provides additional useful hydrogeological informationas any MRS signal means groundwater presence(Legchenko et al., 2002). However, both MRS andelectrical resistivity methods are not self-sufficientbecause geophysical parameters need to be comparedwith borehole and pumping tests data before quantifyinghydrogeological characteristics like water electricalconductivity (EC) and aquifer properties (Vouillamozet al., 2002). Moreover, the uncertainty of geophysicalinterpretation can be notably reduced when severalmethods are jointly used (Albouy et al., 2001; Goldmanet al., 1994).

Fig. 1. Location of th

As a result, an integrated field survey was carried outin Myanmar (previously named Burma). This field wasselected because the NGO “Action contre la Faim”(AcF) drilled more than 1000 boreholes in the coastalarea, but 59% of them turned out to be unsuccessful. Theimplementation of new boreholes is the main problemhydrogeologists are confronted with, and the hydro-geophysical integrated approach is very attractive tocharacterize aquifers before drilling boreholes. Thispaper presents the main results of such an integratedsurvey carried out from April to June 2004.

2. Study area

Maungdaw and Buthidaung townships are located inthe North Rakhine State (NRS) of Myanmar (Fig. 1).This state is bordered by the Gulf of Bengal to the West,and it is partly a delta with low lands that can be floodedboth by river/runoff water during heavy rain, and by seawater at high tide. The climate is wet tropical with amonsoon. The average annual rainfall is 4900 mm andthe average air temperature is 26 °C.

NRS rocks are recent heterogeneous sedimentscomposed of clay, silt and sand that lie on Mioceneshales, claystones and siltstones (Chatenoux et al.,2004). The deposits sequence consists of both con-tinental and marine sediments that are strongly hetero-geneous at a large scale: the cuttings of drilled boreholesare often different from those of observation wells 20 maway.

e survey area.

3J.M. Vouillamoz et al. / Journal of Applied Geophysics xx (2006) xxx–xxx

ARTICLE IN PRESS

The population of the townships is about 720,000,mainly living in rural areas where agriculture is the firsteconomical sector with rice farming. Water is tradition-ally supplied by ponds dug by villagers to catch andstore rainwater. These ponds do not always supply safewater as it is very difficult to protect them from faecal–oral contamination. Shallow hand dug wells are alsoused in some areas, but their poor lining and sanitaryseals as well as the unsafe water lifting devices cause thewater to be also unsafe. Finally, safe drinking water isonly provided by AcF-constructed boreholes andprotected wells that are accessible to only 32% of thetownships inhabitants (2004).

3. Hydrogeological analysis

According to the analysis of data obtained from aselection of 551 boreholes, two main sets of aquiferswere identified. A first set of shallow aquifers extendsfrom a few meters below ground level to a depth ofabout 70 m. Their geometry and properties areheterogeneous at the scale of a few tens of meters.Both the specific capacity calculated from air liftdevelopment (that is well correlated to the transmissiv-ity) and the groundwater salinity estimated with its EChave various values (Fig. 2A and B). The second set ofaquifers is located below a depth of 70 m. It is morehomogeneous with almost always higher specificcapacity and less risk of high groundwater EC.

The EC of groundwater ranges from low mineralizedwater close to the local rainwater (60 μS/cm) to high

Fig. 2. Aquifer characteristics. A: depth against sp

mineralised water close to the gulf of Bengal water(28,000 μS/cm). This mineralization can be explainednot only by the rocks' sedimentation process withmarine sequences, but also by the actual infiltration ofocean water at high tide. The tide influence can bemeasured several tens of kilometers inside the low landsthrough the rising of salty water in rivers that infiltratethe aquifers.

Looking at the data of 1044 boreholes, severalreasons explain unsuccessful boreholes: (1) Holes aredry in 26% of the drilled sites, (2) Borehole productivityis very low in 3.4% of the holes (yield less than 0.5 m3/h),(3) Water EC is too high for human consumption in 22%of the sites (EC>3000μS/cm) and (4) Technical problemsdo not allow the completion of drilling. Finally, theaverage drilling success rate for the 1995–2003 period is41%, which corresponds to the completion of an averageof 54 successful boreholes per year.

The main questions encountered by hydrogeologiststo implement new boreholes are identified from thehydrogeological and drilling analyses: (1) Is there a deepaquifer available in the selected area? If yes, at whatdepth and of what thickness? (2) What is the EC ofgroundwater? (3) What is the productivity of theaquifer?

4. Use of geophysics for hydrogeology

To answer the hydrogeological questions, geophy-sical measurements are implemented at 25 sites thatwere selected to represent as well as possible the NRS

ecific capacity. B: depth against water EC.

4 J.M. Vouillamoz et al. / Journal of Applied Geophysics xx (2006) xxx–xxx

ARTICLE IN PRESS

context (Chatenoux et al., 2004). Boreholes andobservation wells are drilled and pumping tests areconducted at these sites to characterize the saturatedaquifer depth, thickness, storativity, transmissivity, andthe water EC. Then, conversion equations constructedwith geophysical parameters are calibrated withhydrogeological properties.

Electrical resistivity measurements are essential toassess water salinity in coastal environment. Directcurrent (DC) method has major disadvantages incoastal heterogeneous areas since it is sensitive tolateral and near-surface heterogeneities, and is alsostrongly affected by the equivalence, i.e., a measure ofthe non-uniqueness of the inverse problem solution(Goldman et al., 1989). Time domain electromagneticsoundings (TDEM) method is more efficient todetermine the depth of fresh/saline water interface ina coastal environment because it is very sensitive toconductive targets. TDEM data interpretation is alsosignificantly less ambiguous as compared with DCmethod, and it is less sensitive to lateral heterogeneitiesthan vertical electrical sounding (VES; Goldman et al.,1989; Goldman and Neubauer, 1994). However, VESequivalence can be notably reduced if additionalinformation is used to fix parameters, namely resistivityor thickness of layered model. In this study, thicknessand depth of saturated aquifers derived from MRSinversion were introduced as fixed parameters toinverse VES data. Methods and tools used for thesurvey are presented in Table 1.

4.1. Use of MRS parameters to estimate properties ofaquifers

Magnetic resonance sounding (MRS) aims toenergize the nucleus of the hydrogen of groundwatermolecules and to measure the magnetic resonance signalthat is sent out by protons after the stimulation signal is

Table 1Methods and tools used in the survey

Method Material Array Interpretationsoftware

VES Syscal R1+ Schlumberger IPI2Win(Iris Instruments) ABmax=600 m (Moscow State

University)MRS Numisplus Square loop Samovar

(Iris Instruments) 75 m side (Iris Instruments)Drilling PAT 201 and 301

(PAT company)4″ and 6″1/2drilling diameter

–

Pumpingtest

2″ submersiblepump

Average pumpingduration of 4.6 h

Aquifer test(WHI software)

cut off (Legchenko et al., 2002). The geophysicalparameters obtained from MRS are the MRS watercontent (wMRS) and the signal decay constants (T2⁎ andT1) that are linked to the mean size of pores that containgroundwater (Schirov et al., 1991). These outputparameters provide two types of hydrogeologicalestimators: storage related parameters and flow relatedparameters.

4.1.1. Storativity calculationMost rocks contain a percentage of empty spaces that

is filled with water if it is saturated. This is the totalporosity n of rocks. In comparison with the totalporosity, the MRS water content obtained after inversionof the MRS signal is defined as (Legchenko et al.,2002):

wMRS ¼ Vlong

Vtotald100 ð1Þ

where Vlong is the volume of water with sufficient longdecay constant T2⁎ that is measured with the actualinstrumentation (>30 ms), and Vtotal is the total volumesampled by the sounding that is determined by the localand environmental conditions (geomagnetic field,electrical resistivity of the sub-surface, dimension ofMRS loop, energizing pulse). Eq. (1) means that MRSwater content is less than the total porosity of saturatedaquifer because the decay of the MRS signal makespart of the signal shorter than the actual instrumentationis able to detect.

Effective porosity nc quantifies the portion of waterthat contributes to flow in saturated aquifer. Effectiveporosity is less than the total porosity because part ofthe total porosity is bound water attached to the surfaceof rocks by molecular attraction forces, and becauseunconnected and dead-end pores do not contribute tothe flow. Because the MRS signal is longer for mobilewater (several tens to several thousands of millise-conds) than for bound water (several units to severaltens of milliseconds), the MRS water content that isderived from the long signal (>30 ms) is a roughestimation of the effective porosity nc (in non-consolidated sediments the role of unconnected anddead-end porosity is negligible):

wMRSfnc ð2Þ

The release from storage in unconfined aquifers isdue to the de-watering of pores, whereas the releasefrom storage in confined aquifers is because of the effectof water expansion and aquifer compaction caused bythe change in fluid pressure. In confined aquifers,

5J.M. Vouillamoz et al. / Journal of Applied Geophysics xx (2006) xxx–xxx

ARTICLE IN PRESS

storativity is measured by the so-called storagecoefficient that depends on the elastic properties ofaquifers and water. It is quantified by (Freeze andCherry, 1979):

S ¼ qdgdbdðaþ nbÞ ð3Þ

where ρ is the mass per unit volume of water, g is theacceleration of gravity, b is the saturated thickness, α isthe aquifer compressibility, β is the compressibility ofwater and n is the total porosity. In unconfined aquifers,the amount of water released from an aquifer storage bywell abstraction is mainly due to gravitational forces(the elastic component is neglected). The storativity isquantified by the specific yield Sy. If a confined aquiferbecomes locally unconfined because of the pumpingeffect, the storability is the sum of the elastic storageand specific drainage. The latter is defined as thespecific yield of confined aquifers (Lubczynski andRoy, 2004).

To calculate storativity, two MRS estimators havebeen proposed but they have not yet been validatedwith sufficient experiments (Vouillamoz, 2003). Inconfined aquifers, storage coefficient SMRS could bederived from its hydrogeological formulation (Eq. (3)),replacing the total porosity n with the MRS watercontent wMRS, and the thickness b with the thicknessof saturated layer Δz obtained from MRS:

S ¼ qdgdbdðaþ nbÞYSMRS ¼ C1dwMRSdDz ð4Þwhere C1 is a parametric factor that needs to becalculated comparing SMRS with storativities derivedfrom pumping test. In unconfined coarse grainsaquifers, specific yield is comparable in value toeffective porosity but it can differ substantially forclayey material where it is only part of the effectiveporosity (Lubczynski and Roy, 2005). Thus, an MRSestimator of specific yield for unconfined sandy aquiferSy_MRS can be proposed:

wMRSfnczSyYSy MRS ¼ wMRSdC2 ð5Þwhere C2 is a parametric factor.

4.1.2. Hydraulic conductivity and transmissivitycalculation

Kenyon (1997) proposed an empirical formula of theintrinsic permeability of rocks:

kMRS ¼ CdwadTb ð6Þwhere C, a and b are parametric factors that are sitespecific. The intrinsic permeability k depends only on

rock properties while the hydraulic conductivitydepends also on water properties:

K ¼ kdgv

ð7Þ

where K is the hydraulic conductivity, g is theacceleration of gravity and v is the cinematic viscosityof the fluid. Since the intrinsic permeability is linked tothe hydraulic conductivity, Legchenko et al. (2002)proposed to use Eq. (6) to estimate K. Variousexperiments were conducted to check the appropriatevalue of the factors a and b, and the formula obtainedfor sandy aquifers is (Legchenko et al., 2004; Vouilla-moz, 2003):

KMRS ¼ CpdwdT21

TMRS ¼ KMRSdDz ð8Þwith a=1, b=2, Cp is a parametric factor that needs tobe calibrated from pumping test data, and KMRS andTMRS are respectively the MRS hydraulic conductivityand transmissivity.

4.2. Use of electrical resistivity to estimate water EC

4.2.1. Water electrical resistivity calculationMany attempts have been made to quantitatively

determine the porosity of the aquifer and the waterconductivity by electrical resistivity measurements. Thebasis of these calculations is the Archie equation(Archie, 1942):

qw ¼ qaqdnmaa

ð9Þ

where ρw is the resistivity of the water, ρaq is theresistivity of the saturated aquifer, a and m are materialempirical factors and na is the Archie porosity. TheArchie porosity is assumed to be the specific yield Sy ascited in Singha and Gorelick (2005).

For geophysicists, Eq. (9) remains difficult to resolveas it still contains two unknowns, the water resistivityρw and the specific yield Sy. Only the use of a prioriinformation such as groundwater EC obtained fromborehole, or specific yield known from pumping tests,can help to solve it. But experience shows that a prioriinformation is often insufficient and that the accuratedetermination of ρw and Sy is only possible in simpleconditions (Kafri and Goldman, 2005). Moreover, weknow from boreholes that the NRS aquifers are not freeof clay, which invalids the direct use of Eq. (9) asillustrated by Worthington (1993). As a result, we did

6 J.M. Vouillamoz et al. / Journal of Applied Geophysics xx (2006) xxx–xxx

ARTICLE IN PRESS

not use the Archie equation in this survey but we lookedfor a rough estimation of the water resistivity using theaquifer resistivity calculated from VES:

qw ¼ 1=EC ¼ C3dqaq ð10Þwhere C3 is a parametric factor.

5. Field results

5.1. Interpretation methodology

The survey conducted at site S01 is presented toillustrate the methodology used to characterize aquifers(Fig. 3).

5.1.1. Drilling boreholesAn observation well was drilled at a distance of

19.9 m from an existing borehole (Fig. 4A). The rockswere clay and sand, and the medium was hetero-geneous as the thickness of the different layersdiffered in the borehole and the well. The sandyaquifer, confined by a clayey layer, was found 10 to13 m below the ground level. The thickness of theaquifer was estimated to be between 30 and 40 m.

Fig. 3. Hydrogeophysical survey and interpretation methodology. Exam

5.1.2. Pumping testA hydraulic test was conducted, pumping for

4.5 h at 1.95 m3/h in the borehole, and monitoringthe water level in both borehole and well. Themaximum drawdown did not reach the bottom of theconfining layer at the observation well (Fig. 4A). Atthe borehole, the depletion of water below theconfining layer was negligible compared to theaquifer thickness as the head loss due to thepumping was considered (Fig. 4A). Consequently,both Theis and Jacob methods were used to interpretthe raw data (Fig. 4B and C; Kruseman and deRidder, 2000). The hydrogeological properties esti-mated from this interpretation were pumping testtransmissivity TQ=2.4 10−4 ±18% and pumping teststorativity SQ=2.4 10−4 ±5%. The water EC, thatwas monitored during the pumping period, wasstable ranging between 955 and 977 μS/cm, that isan average water resistivity of ρw=10.4 Ω · m.

5.1.3. MRSA square loop of side 75 m was laid around the

borehole to set up a MRS (Fig. 5). MRS inversionproduced a multilayered aquifer with a shallowsaturated level between 3 and 9 m deep and a main

ple of borehole data and geophysical inversion results at site S01.

Fig. 4. Borehole and pumping test data, site S01. A: raw data of borehole and water level during the pumping test. B: Jacob method applied to theobservation well (drawdown data). C: Theis method applied to the observation well data.

7J.M. Vouillamoz et al. / Journal of Applied Geophysics xx (2006) xxx–xxx

ARTICLE IN PRESS

saturated level between a depth of 12 and 55 m. Thedeeper aquifer, that was obviously the target (Fig. 3),has two distinct levels: it has a short T1 decayconstant between the depth of 12 and 26 m(T1≈ 50 ms) and a longer T1 below 26 m(T1≈200 ms). The short and long T1 may beunderstood as a fine to clayey sand overlaying acoarser sand respectively.

Fig. 5. Raw geophysical data, site S01.

5.1.4. VESA Schlumberger VES was conducted nearby the

borehole (Fig. 5). The aquifer geometry (i.e., depthand thickness) obtained from MRS interpretation wasused as fixed parameter to inverse the VES data.According to the analysis of equivalence the resistivityof the targeted aquifer ranges between 23 and89 Ω · m.

A: MRS. B: Schlumberger VES

Table 2Geophysical calibration

Conversion equation Calibration factor Max.(%)

First quartile(%)

Median(%)

Mean(%)

Third quartile(%)

Min.(%)

MRS storage coefficient SMRS=(wMRS ·Δz) ·C1 C1=2.42 10−4 76 11 6 8 11 8SMRS=ρ · g · (C4+wMRS ·β) ·Δz C4=1.6 10−9 76 4 2 6 44 2

MRS transmissivity TMRS=Cp ·w(z) ·T12(z) ·Δz Cp=6.68 10−9 185 4 31 45 54 1

VES water EC ρw=1/EC=C3 ·ρaq C3=0.43 233 28 51 75 117 2

The percentages are the differences between the hydrogeological parameters derived from pumping tests and from geophysical estimators.

8 J.M. Vouillamoz et al. / Journal of Applied Geophysics xx (2006) xxx–xxx

ARTICLE IN PRESS

5.1.5. Hydrogeophysical characterization of the aquiferA comparison between geophysical data and

pumping test interpretations was done for all thesurveyed sites. The parametric factors in Eqs. (4) (8)and (10) were determined, which leads to theestimation of storativity, hydraulic conductivity andtransmissivity with MRS, and estimation of ground-water EC with VES (Fig. 3 and Table 2).

5.2. MRS estimators

5.2.1. MRS storativitySeven test sites were selected for storativity

calibration (MRS storativities were calculated fromEq. (4) because of confined aquifers). We found arelationship between storativity estimated with MRSand pumping tests as illustrated in Fig. 6A and Table2. To check the validity of Eq. (4), the hydrogeolo-gical formulation of storage coefficient is used

Fig. 6. Calibration of MRS estimators.

replacing b with Δz, n with wMRS and α with aparametric factor C4:

S ¼ qdgdDzdðC4 þ wMRSdbÞ ð11ÞUsing typical values of ρ=1000 kg·m−3, g=9.81 m·s−2

and β=5.10−10 Pa−1 (De Marsily, 1986), the fit isslightly improved compared to that obtained using Eq.(4) with C4 =1.6 10− 9 (Table 2). This value iscomparable to the compressibility factor of a sandyreservoir (α=1.6 10−9 Pa−1), that is in accordance withthe local geology. It indicates that Eq. (4) is acceptable toestimate storativity from MRS.

5.2.2. MRS tranmissivityThe best fit to estimate the transmissivity from MRS

using 15 soundings with Eq. (8) is obtained with theparametric factor Cp=6.68 10−9 (Fig. 6B and Table 2).To compare this Cp value with the value obtained in

A: storativity. B: transmissivity.

Fig. 7. Water EC against calculated aquifer resistivity.

Fig. 8. Maximum investigation depth of MRS.

9J.M. Vouillamoz et al. / Journal of Applied Geophysics xx (2006) xxx–xxx

ARTICLE IN PRESS

France in similar geological conditions, the hydraulicconductivity and the MRS signal that depend on watertemperature need to be corrected (the average watertemperature encountered in Myanmar was 30 °C against12 °C in France). After temperature correction, the bestparameterization of T is obtained with the new factorCp=5 10−9. This factor is very close to Cp=4.9 10−9

obtained for similar sandy formations in France byVouillamoz (2003). It indicates that the MRS transmis-sivity estimator is robust and depends mainly on thereservoir's nature and structure.

5.3. VES estimator

The water EC measured with a light conductivity-meter in water samples is compared with the waterresistivity calculated with Eq. (10) for 15 VES (Fig. 7).The best fit is obtained using the parametric factorC3=0.43 (Table 2).

6. Discussion

6.1. Geophysical results

6.1.1. MRS methodThe maximum penetration depth of a sounding can

be estimated as:

h ¼ 500dffiffiffiffiffiffiffiffiffiffiffiqaq=f

qð12Þ

with h as the maximum penetration depth that is thedepth where the signal is reduced by about 30% (skin

depth), ρaq is the electrical resistivity of the media and fis the signal frequency.

The maximum penetration depth was computed withEq. (12) for every site assuming the frequency off=1900 Hz, and limited to the side length of transmis-sion/reception (Tx/Rx) loop (Legchenko et al., 1997).We found that deep aquifers that are in the “blind zoneof MRS” (shadow zone of Fig. 8) can not be identifiedby this method. As an overall result, MRS was useful in68% of the sites (10 sites in the blind zone and 1 rejectedbecause of poor signal to noise ratio over a total of 34).

Previous works (Vouillamoz, 2003 and Table 3)assessed the equivalence on geometry derived fromMRS for a sample of 30 soundings: the averageuncertainty of the depth and thickness of a saturatedaquifer was ±79% and ±39% respectively. For thecurrent survey these uncertainties will mainly affect theuse of electrical resistivity parameters since MRSgeometry is used as a fixed parameter to inverse VES.But the uncertainties on hydrogeological propertiesestimated from MRS are acceptable because the mainMRS equivalence concerns the product (w ·Δz) that ispart of MRS storativity and transmissivity estimators(Legchenko and Shushakov, 1998). Moreover, theaverage MRS uncertainties on storage coefficient andtransmissivity are comparable to the uncertainties onhydrogeological properties calculated from pumpingtests (Table 3).

We found that the main limitation to the use of MRSin NRS context is an insufficient depth of investigationbecause of shallow electrically conductive layers thatmay screen a deeper aquifer (Roy and Lubczynski,2003). When not affected by the screening effect, the

Table 3Uncertainty of geophysical and hydrogeological characterization

Symbol Max.(%)

First quartile(%)

Median(%)

Mean(%)

Third quartile(%)

Min.(%)

Resistivity of aquifer derived from VES with fixed geometry ρaq 230 32 2 18 10 58Transmissivity estimated from pumping test TQ 284 32 48 57 66 11Transmissivity estimated from MRS TMRS 231 39 49 64 72 6Storativity estimated from pumping test SQ 43 8 17 18 25 2Storativity estimated from MRS SMRS 65 11 16 19 22 2

10 J.M. Vouillamoz et al. / Journal of Applied Geophysics xx (2006) xxx–xxx

ARTICLE IN PRESS

storativity and the transmissivity of aquifers wereestimated by MRS with acceptable average accuraciesof ±6% and ±45% respectively (Table 2). This is themain advantage of the method.

6.1.2. VES methodThe VES measurements provided sufficient signal to

noise ratio in 83% of the sites (6 VES were rejected overa total of 35). To minimize equivalence in VESinterpretation, we used the geometry derived fromMRS as a fixed parameter. As illustrated for site S01,using fixed parameters reduces drastically the uncertaintyof aquifer resistivity from 2.5 Ω ·m<ρaq<141 Ω ·m to23Ω ·m<ρaq<89Ω ·m (Fig. 9). The average uncertaintyon ρaq was ±18% for this survey (Table 3).

It is not possible to estimate a reliable water ECfrom aquifer resistivity because of the equivalence andlateral heterogeneities that add uncertainty to VESinterpretation. For example, the resistivity of thetargeted aquifer at site S01 was 23Ω ·m<ρaq<89Ω ·m,corresponding to an EC of groundwater calculated withEq. (10) between 261 μS/cm<EC<1011 μS/cm. Thislow accuracy will not always allow to distinguishbetween acceptable and too high EC for human use.

Fig. 9. VES equivalence at site S01. A: filed data and synthetic curves for mmodels with aquifer geometry fixed from MRS inversion.

However, a guideline can be proposed as illustrated inFig. 10: (1) if ρaq<5 Ω ·m the water will be probablysalty, (2) if 5 Ω ·m<ρaq<12 Ω ·m the risk of salinity ishigh, and (3) if ρaq>12 Ω ·m the water will beprobably used by the people.

6.1.3. Joint use of MRS and VES: a geophysicalmethodology

MRS and VES can complement each other effi-ciently. MRS is used to locate productive aquifers and toquantify their storativity and transmissivity. VES is usedto estimate the salinity risk of groundwater identifiedwith MRS. Finally, the maximum penetration depth ofMRS is calculated with VES resistivity and compared tothe depth that was targeted.

The exploration site S31 illustrates the proposedmethodology (Fig. 11). Two aquifers are identified byMRS. The deeper reservoir is located from 25 to 65 mdeep and contains water that is less mineralized than theshallower one according to the VES interpretation(EC=110 μS/cm against EC=560 μS/cm on average).This reservoir will be targeted by the drilling, and itsstorativity is estimated with MRS to be SMRS=6·10

−4

and its transmissivity to reach TMRS=1·10−3 m2/s.

odels without fixed parameter. B: models without fixed geometry. C:

Fig. 10. Evaluation of salinity risk from VES measurement. Theshadow area represents the unusable water for human purpose. Thesolid line is the link between ρaq and EC calculated with Eq. (10). Thebroken lines represent the uncertainty on the ρaq calculation due toequivalence in VES inversion.

11J.M. Vouillamoz et al. / Journal of Applied Geophysics xx (2006) xxx–xxx

ARTICLE IN PRESS

6.2. Economical considerations

The presented economical analysis aims to estimatethe financial impact of geophysics on the drilling

Fig. 11. Example of aquifer identificati

programme. The calculations are carried out consideringthe real AcF costs (year 2004) that include the staff, thelogistics, the geophysical equipment (buying, mainte-nance and depreciation) and the administrative costs.The life period is assumed to be 2 years for the drillingrig, 5 years for the VES apparatus and 8 years for theMRS instrumentation. Eight working months and 54successful borehole implementations per year areassumed. The local average cost of a successfulborehole is 664€ and 502€ for an unsuccessful one.

Calculations show that geophysics saves moneywhen (Vouillamoz et al., 2002):

Gzbh�dr2r1

� 1

� �ð13Þ

with r2 the actual borehole success rate, r1 the successrate with the use of new geophysics, bh− the averagecost of an unsuccessful borehole and G the average costof geophysical surveys per borehole.

Starting from the known or expected drilling successrate, one can calculate the minimum improvement ofsuccess rate that is needed to save money usinggeophysics. For example, the number of successfulboreholes drilled by AcF in Myanmar between 1995 and2003 ranges between 30% and 55% on average.Calculation shows that the joint use of VES and MRSis economically acceptable in complex areas where thesuccess rate of drilling is not more than 30%.

on and characterization, site S31.

12 J.M. Vouillamoz et al. / Journal of Applied Geophysics xx (2006) xxx–xxx

ARTICLE IN PRESS

7. Conclusions

In North Rakhine State of Myanmar, the combineduse of MRS and VES allows reliable detection of coastalaquifers, and acceptable estimation of their storativity,transmissivity and water salinity risk. Average accuracyof the estimation was found to be ±6% for the storativityand ±45% for the transmissivity. The MRS transmis-sivity estimator seems to be robust, but the MRSstorativity estimator still need be validated with a largernumber of data. The estimation of water salinity is notaccurate enough (±75% on average) but a salinity riskcan be estimated.

Existence of shallow electrically conductive layersmay screen MRS signal and reduce the maximum depthof investigation. During the survey time, 32% of MRSsoundings were not able to reach targeted aquifers below20 to 70 m because of this screening effect. 83% of VEShad sufficient signal to noise ratio, and VES equivalencewas acceptable (±18% on average) because the depthand thickness of the targeted aquifer were fixed by MRSinversion.

To be economically acceptable, the cost of geophy-sics has to be compensated by an increase of the successrate of drilled boreholes. Economical analysis of AcFactivities for the period of 1995–2003 shows that theapplication of joint MRS and VES methods is savingmoney when the initial success rate is less than 30%.

Finally, the hydrogeophysical survey was efficientto characterize complex coastal aquifers. The prelimin-ary hydrogeological survey precises the questions thatneed to be answered integrating hydrogeology andgeophysics. On one hand geophysics can give a densedata coverage at a lower cost and in a shorter periodthan by hydrogeological methods, but on the otherhand it needs hydrogeological data to be calibrated.Hydrogeophysical characterization of aquifers has to bea continuous process: the geophysical characterizationis validated by drillings and pumping tests, and thecalibration of geophysics is improved with newhydrogeological data.

Acknowledgments

This work was carried out in the framework of AcF,IRD and BRGM “hydrogeophysical study” collabora-tion, with the support of IRIS Instruments company forMRS instrumentation. We thank Yves Albouy for hisencouragement and support, and Hubert Fabriol whomade the BRGM collaboration possible. We also thankthe whole “Action contre la Faim” field staff involved inthis study. Finally, we thank the reviewers for their

constructive comments, and Mary Joy for the languageediting.

References

Albouy, Y., Andrieux, P., Rakotondrasoa, G., Ritz, M., Descloitres, M.,Join, J.L., Rasolomanana, E., 2001. Mapping coastal aquifers byjoint inversion of DC and TEM soundings—three case histories.Ground Water 39, 87–89.

Archie, G.E., 1942. The electrical resistivity as an aid in determiningsome reservoir characteristics. Metallurgical and PetroleumEngineers 146, 54–62.

Chatenoux, B., Vouillamoz, J.M., Mathieu, F., 2004. GroundwaterInvestigation in North Rakhine State of Myanmar, Hydrogeophy-sical Survey 2004. AcF, Paris.

De Marsily, G., 1986. Quantitative Hydrogeology. Academic Press.439 pp.

Freeze, R.A., Cherry, J.C., 1979. Groundwater. Prentice Hall, Engle-wood Cliffs, NJ.

Goldman, M., Arad, A., Kafri, U., Gilad, D., Melloul, A., 1989.Detection of Freshwater/Sea-water interface by the Time DomainElectromagnetic (TDEM) method in Israel. 10th SWIM, Ghent,pp. 329–344.

Goldman, M., Rabinovich, B., Gilad, D., Gev, I., Schirov, M., 1994.Application of the integrated NMR–TDEM method in thegroundwater exploration in Israel. Journal of Applied Geophysics31, 27–52.

Goldman, M., Neubauer, F.M., 1994. Groundwater exploration usingintegrated geophysical techniques. Surveys in Geophysics 15,331–361.

Kafri, U., Goldman, M., 2005. The use of time domain electro-magnetic method to delineate saline groundwater in granular andcarbonate aquifers and to evaluate their porosity. Journal ofApplied Geophysics 57, 167–178.

Kenyon, W.E., 1997. Petrophysical principles of applications of NMRlogging. The Log Analyst, March–April 21–43.

Kruseman, G.P., de Ridder, N.A., 2000. Analysis and evaluation ofpumping test data. ILRI, Wageningen. 377 pp.

Legchenko, A., Shushakov, O.A., 1998. Inversion of surface NMRdata. Geophysics 63, 75–86.

Legchenko, A., Beauce, A., Guillen, A., Valla, P., Bernard, J., 1997.Natural variations in the magnetic resonance signal used in PMRgroundwater prospecting from surface. European Journal ofEnvironmental and Engineering Geophysics 2, 173–190.

Legchenko, A., Baltassat, J.M., Beauce, A., Bernard, J., 2002. Nuclearmagnetic resonance as a geophysical tool for hydrogeologists.Journal of Applied Geophysics 50 (1–2), 21–46.

Legchenko, A., Baltassat, J.M., Bobachev, A., Martin, C., Robain, H.,Vouillamoz, J.M., 2004. Magnetic resonance soundings applied tocharacterization of aquifers. Ground Water 42 (3), 436–473.

Lubczynski, M., Roy, J., 2004. Magnetic Resonance Sounding: newmethod for ground water assessment. Groundwater 42, 291–303.

Lubczynski, M., Roy, J., 2005. MRS contribution to hydrogeologicalsystem parametrization. Near Surface Geophysics 3 (3), 131–139.

Roy, J., Lubczynski, M., 2003. The case of an MRS elusive secondaquifer. Deuxième séminaire international sur la méthode deRésonance Magnétique des Protons appliquée à la recherche noninvasive d'eau souterraine, Orléans, France, pp. 105–108.

Schirov, M., Legchenko, A., Creer, G., 1991. New direct non-invasiveground water detection technology for Australia. ExplorationGeophysics 22, 333–338.

13J.M. Vouillamoz et al. / Journal of Applied Geophysics xx (2006) xxx–xxx

ARTICLE IN PRESS

Singha, K., Gorelick, S.M., 2005. Saline tracer visualized with three-dimensional electrical resistivity tomography: field-scale spatialmoment analysis. Water Resources Research 41, 1–17.

Vouillamoz, J.M., 2003. La caractérisation des aquifères par uneméthode non invasive: les sondages par Résonance MagnétiqueProtonique. Hydrogeophysical thesis, Paris XI, Orsay. 315 pp.

Vouillamoz, J.M., Descloitres, M., Bernard, J., Fourcassié, P.,Romagny, L., 2002. Application of integrated magnetic resonance

sounding and resistivity methods for borehole implementation, acase study in Cambodia. Journal of Applied Geophysics 50 (1–2),67–81.

Worthington, P., 1993. The uses and abuses of the Archie equations, 1 :the formation factor—porosity relationship. Journal of AppliedGeophysics 30, 215–228.