1 Introduction

L'agglomération de la ville de Ho-Chi-Minh (HCM), anciennement ville de Saigon, au Sud-Vietnam, avec une population de plus de 5 millions d'habitants, est confrontée à un énorme problème d'alimentation en eau potable, à la fois pour les besoins domestiques et pour les besoins industriels. Le plus gros problème provient de la pollution des nappes aquifères superficielles. La pollution est d'abord d'origine naturelle, par invasion de l'eau de mer dans les régions côtières, la ville de HCM étant située seulement à 45 km à vol d'oiseau de la mer de l'Est et communiquant avec elle par plusieurs fleuves (Fig. 1) et par lessivage des dépôts de sel laissés durant les différentes transgressions du Tertiaire et surtout du Quaternaire. La pollution est ensuite d'origine humaine, à cause des énormes concentrations de population le long des cours d'eau qui sillonnent l'agglomération.

Les travaux de recherche hydrogéologique au Sud-Vietnam ont débuté dès 1930, sous la colonisation française [6], ont été poursuivis pendant la guerre par les Américains [1] et, depuis 1975, par différents organismes publics vietnamiens. De nombreux forages ont été effectués dans l'agglomération de HCM et ont permis de trouver des aquifères superficiels dans les sédiments quaternaires (Holocène, Pléistocène), et d'autres plus profonds, jusqu'à 200 m environ, dans les sédiments tertiaires (Pliocène, Miocène). Malheureusement, certains aquifères sont pollués par un taux de minéralisation excessif, dépassant une concentration équivalente à 1 g de chlorures par litre, norme admise pour l'eau potable.

Il est donc important de définir une méthodologie géophysique capable de délimiter les zones polluées, aussi bien près de la surface qu'en profondeur, afin de mieux comprendre le processus de contamination des nappes aquifères et de guider l'implantation des forages dans la recherche des nappes d'eau douce.

Parmi les méthodes géophysiques appliquées aux recherches hydrogéologiques, c'est la méthode de sondage électrique (SE) qui est de loin la plus utilisée [7,8]. En effet, il existe un bon contraste électrique entre les aquifères d'eau douce (d'une résistivité de quelques centaines d'$\Omega$ m) et les aquifères d'eau salée (quelques $\Omega$ m). Malheureusement, la profondeur d'investigation des SE est très limitée, de l'ordre de 1/4–1/10 de la longueur de la ligne d'injection de courant dans le dispositif Schlumberger. Pour des raisons d'ordre pratique, cette longueur est généralement limitée à 1000 m maximum. L'originalité de la méthodologie adoptée durant la présente étude consiste en l'utilisation conjointe de deux méthodes complémentaires : le SE pour l'exploration superficielle, jusqu'à une centaine de mètres de profondeur, et le sondage magnéto-tellurique (SMT) pour l'exploration profonde jusqu'à plusieurs centaines de mètres de profondeur [2,5]. La méthode combinée SE/SMT a été appliquée avec succès le long de deux profils DH-TB et HM-CC, situés au nord-ouest de HCM (cf. Fig. 1) [4,5]. Nous présentons dans cet article les résultats complets comportant cinq profils supplémentaires, LC, HB, BH, AA et CG (Fig. 1), et un total de 60 stations réparties sur les sept profils, couvrant une large région autour de l'agglomération de HCM (délimitée par le rectangle sur la Fig. 1). Au niveau de chaque station, les données SE et SMT sont obtenues sous forme de courbes de sondage, qui représentent respectivement la courbe de résistivité apparente en fonction de l'écartement des électrodes d'injection (2–1000 m) pour les données SE et la courbe de résistivité apparente en fonction des fréquences (3–2500 Hz) pour les données SMT. Un exemple de ces courbes de sondage a été donné dans l'article publié précédemment [5].

Après inversion des données SE et SMT, il est possible de réaliser une tomographie électrique de la région étudiée, en dressant les cartes de l'image électrique du sous-sol à différentes profondeurs, et en établissant leur relation avec la qualité des nappes d'eau souterraine.

2 Cadres tectonique et géologique

3 Relation entre la résistivité des nappes aquifères et la qualité de l'eau

La qualité de l'eau souterraine dépend principalement de sa salinité, c'est-à-dire de la concentration C. Le Tableau 1 montre la correspondance entre cinq « qualités » usuelles d'eau, caractérisées par leur concentration C et leur résistivité ρ_e, obtenue à partir des courbes de Keller et Frischknecht à la température de 20 °C. À partir de la loi simplifiée d'Archie, nous pouvons tracer les courbes de variation de la résistivité du réservoir en fonction de sa porosité pour les cinq qualités considérées de l'eau (Fig. 3). En adoptant une certaine hypothèse sur la porosité, il est possible d'en déduire la qualité de l'eau connaissant la résistivité du réservoir.

Table 1

Correspondence between the quality of the groundwater characterised by its sodium chloride concentration C, and its resistivity ρ_e at temperature 20 °C (from [3]).

Correspondance entre la qualité de l'eau souterraine caractérisée par sa concentration C en chlorure de sodium, et sa résistivité ρ_e, à la température de 20 °C (d'après [3]).

Groundwater quality	Concentration C (g l−1)	Resistivity ${\rho }_{\mathrm{e}}(\Omega$ m)
Salty (seawater)	30	0.2
Brackish	7.5	0.8
Mineralised	2	3
Potable	0.4	15
Fresh	0.1	60

4 Coupes géoélectriques SE et SMT le long du profil BH

Sur la Fig. 2 sont donnés respectivement en (a) et en (b) les coupes géoélectriques SE et SMT du profil BH. Comme il a déjà été observé sur les anciens profils [5], la coupe géoélectrique SE donne plus de détails sur la structure superficielle jusqu'à 200 m de profondeur environ (Fig. 2a), tandis que la coupe SMT permet d'obtenir des informations à plus grande profondeur, jusqu'à 800 m environ (Fig. 2b). L'existence de trois forages carottés à proximité du profil BH permet de corréler les coupes géoélectriques avec la stratigraphie. On peut constater ainsi que, dans la partie sud-ouest du profil, les sédiments quaternaires sont plus conducteurs ($\rho <10\Omega$ m), probablement à cause de la présence d'eau salée ou saumâtre (Fig. 2a). L'existence de nappes saturées d'eau salée très conductrice (couleur rouge : $\rho \approx 1\Omega$ m) indique une forte porosité de ces nappes de l'ordre de 40 %, d'après les graphes de la Fig. 3. En adoptant, en première approximation, cette valeur de porosité pour tous les aquifères (axe vertical correspondant à 40 % sur la Fig. 3), il est possible d'établir une échelle de la qualité de l'eau en fonction de la résistivité des aquifères (échelle indiquée sur la partie droite des Figs. 2 et 4). On observe ainsi l'existence probable de nappes d'eau douce potable (couleur bleue) très locales dans les aquifères néogènes, superficiels à l'extrémité nord-est du profil, et plus profonds au sud-ouest. Au-dessous, on trouve les schistes argileux imperméables du Mésozoı̈que (ρ≈100–$1000\Omega$ m) et le socle très résistant ($\rho >1000\Omega$ m) de roches ignées (Fig. 2b), ces deux éléments étant tous deux dépourvus de nappes aquifères.

5 Tomographie électrique : évaluation de la qualité des ressources d'eau souterraine

6 Discussion et conclusion

The agglomeration of Ho-Chi-Minh (HCM) City (formerly Saigon City) and several surrounding rural districts in South Vietnam, with more than 5 million inhabitants, are confronted with a dramatic shortage of fresh water supply because of the contamination of several aquifers. Due to its location at about 45 km from the East Sea (Fig. 1), superficial aquifers along the coastal area are contaminated by the seepage of seawater and by the inland incursion of salty tidal water, especially during the dry season (November–April). In the other areas, the contamination originates from the leaching of saline sediments deposited during different Tertiary and Quaternary transgressions. The human activities contribute to the pollution of many rivers that run through the agglomeration. During high water season (August–November), about half the Delta is inundated by overflow of the Mekong River. As a result of the seasonal flooding and drying, the soils have become saturated with aluminium and iron salts, which are toxic to most crops. Moreover, a closely integrated manmade network of canals and inland waterways is subject to tidal incursion of brackish or salty water from the East Sea, which may extend far inland during the low water season (February–April).

Many wells were drilled in Saigon City by the French–Asian Hydraulic Company [6]. Since 1959, the Americans concentrated groundwater investigations in the delta of South Vietnam [1]. After the end of the war, in 1975, the demand of groundwater for domestic and industrial purposes in both old and newly developing areas required immediate attention. Many government agencies have been created to investigate and exploit groundwater. As a result of many years of investigation, especially by deep drilling (>200 m) in the agglomeration of HCM City, it has been possible to identify deeper fresh-water aquifers in several favourable areas. However, some deep aquifers as well as some superficial ones have brackish or salty water with variable Total Dissolved Solids (TDS) contents. The chemical characteristics of groundwater vary with depth in a very complicated way and are known only in some deep wells on limited areas. Therefore, it is very important to use an efficient geophysical methodology in order to delineate the extent of groundwater contamination, not only near the surface but also at larger depth.

Traditionally, Vertical Electrical Sounding (VES) has been widely used and has played a significant role in groundwater geophysics (e.g., [7,8]). Owing to the strong contrast of resistivity between a freshwater aquifer (a few hundred $\Omega$ m) and a saline water aquifer (a few $\Omega$ m), the efficiency of the VES method lies in the direct correlation, in simple cases, of the interpreted data to groundwater quality. Unfortunately, the investigation depth of this method is relatively low, about 1/4–1/10 of the distance between the current electrodes with the Schlumberger array. For deeper exploration, the Magneto-Telluric Sounding (MTS) method is more suitable (e.g., [2,5]). In fact, these two methods are quite complementary, and the use of the combined VES/MTS method can provide a more complete electrical image of the sedimentary series from the surface to the deep basement. This combined method was first successfully applied along two profiles, DH-TB and HC-CC, located to the northwest of HCM City (see Fig. 1). The results showed a close correlation of the groundwater pollution of the groundwater with the palaeo-geomorphology of laterites, and with the subsequent palaeo-relief during the Tertiary and Quaternary periods [4]. A simple guide for hydrogeological investigation was proposed in order to localise fresh aquifers both in shallow and deep zones connected respectively with Quaternary and Tertiary tectonics [5]. Based on this guide, we have extended the studied area with five more profiles, LC, HB, BH, AA and CG, i.e. a total of 60 stations including the two former profiles, covering a large region around HCM City along the East border of the Mekong Delta (rectangle in Fig. 1). This large number of stations allows performing, after inversion of both VES and MTS data, an electric tomography of the studied area, showing electrical maps at different depths, which may be connected with the quality of groundwater.

The oldest known rocks of the region are granite, gneiss, quartzite and other crystalline rocks, considered to be of Precambrian to Early Palaeozoic age [1]. In South Vietnam, hills of rhyolite and dacite as well as Hercynian granite are aligned along a northwest trend from Vung-Tau (VT) to Tay-Ninh (TN) (see Fig. 1). This trend coincides with the tectonic trough in which the alluvial complex of the Mekong Delta was later deposited. From Permian to Eocene, interbedded continental sediments were deposited in the subsiding basin. A very thick Mesozoic shale was encountered in deep drill-holes. The northwest-trending tectonic trough began to subside in Late Tertiary period (Neogene). During Plio-Pleistocene, the Old Alluvium was deposited by the Mekong and its tributaries in a vast deltaic fill that extends landwards up to the bedrock outcrops, and seawards up to the present coastline. Along the northeastern margin of the trough, the Old Alluvium is in places interbedded with and in part overlain by basalt flows provided by local eruptions of small volcanoes in the early stages of the alluviation. To the southwest, the delta is entirely covered by the recent Quaternary Holocene sediments. The Holocene alluvium contains relatively poor aquifers, largely because of the fine texture of the deposits. Non-potable brackish or salty groundwater is also widely present. Aquifers in Old Alluvium are the principal sources of fresh water supply [1].

In the studied area, several drill-holes tap water-bearing zones in the Old Alluvium (see Fig. 2), especially in the Neogene sediments that have high yield and economical importance because they spread widely with a large thickness and down to 200 m depth. However, the Neogene aquifers have variable TDS and are also contaminated by brackish or salty water in some places.

Numerous investigations have established the efficiency of electrical resistivity methods as a tool for evaluating groundwater quality [7]. Many contaminants contain an ionic concentration considerably higher than the native groundwater. Increased ionic concentration, or TDS, results in higher electrical conductivity (or lower resistivity) of the water. Thus, in an aquifer containing a contaminant, one would expect the pore water resistivity to be lower than in a non-contaminated area. The resistivity of a water-bearing rocks decreases with increasing water content. In a fully saturated rock, which is the general case for the aquifers in the Mekong Delta, water content may be taken as equal with porosity. For these aquifers the rock-reservoir of which is made principally of porous sand or gravel, it has been observed that the resistivity varies approximately as the inverse square of the porosity. This observation has led to the widespread use of an empirical function relating resistivity and porosity which is known as Archie's law [3]:

For a first approximation, a value of 1 may be assumed for a and a value of 2 for m:

In practice, the uncertainty in the knowledge of the parameters a and m is less serious than the uncertainty in the choice of the proper value of ρ_e.

The resistivity ρ_e of the electrolyte depends on the salinity of the water. Since groundwater may have a variety of salts in solution, the concept of an ‘equivalent salinity’ is frequently used. It is defined as the salinity of a sodium chloride solution that would have the same resistivity as that of the particular solution. Keller and Frischknecht [3] give a set of curves showing the relationship between the resistivity and the salinity of a sodium chloride solution at various temperatures. The curves indicate a decrease of the resistivity ρ_e with increasing temperature, but the effect has little influence in normal condition of temperature between 20 and 40 °C. The decrease of ρ_e, as the inverse of the salinity (expressed as the concentration C of sodium chloride in grams per litre (g l⁻¹)), for C<100 g l⁻¹ [3] is more important.

The quality of the groundwater depends principally on the concentration C. In practice, we consider five usual groundwater qualities as shown in Table 1 with the corresponding sodium chloride concentrations C and resistivities ρ_e obtained from Keller and Frischknecht's curves at 20 °C.

By using the simplified Archie's law (Eq. (2)), we obtain the curves of the bulk resistivity ρ versus the porosity (P%) for the five groundwater qualities in Table 1 (Fig. 3). With a hypothesis on the porosity of an aquifer, it is possible to evaluate its quality from its bulk resistivity, which can itself be obtained from the inversion of VES and MTS data.

The inversion results of two former profiles, DH-TB and HM-CC, were presented in a previous paper [5]. As an example, we present hereafter the new results obtained along the profile BH, about 60 km long, including 10 stations and passing through the agglomeration of HCM City (see Fig. 1). The VES data and the MTS data in audiofrequency range (3–2500 Hz) were obtained by the same procedure of measurement as previously described [5]. An example of VES curve and an example of MTS curve are shown in this paper [5].

The stratigraphy in the Mekong Delta is subhorizontal. Therefore, the assumption of a layered model (1D) for the data inversion is justified. Using interactive software and a multilayer 1D model, the resistivity and the thickness of each layer were successively modified until the theoretical VES/MTS curves have a best fit with the experimental VES/MTS curves at each station.

The inversion results of 10 stations of the profile BH are shown in Fig. 2 in the form of two geoelectrical cross-sections corresponding respectively to VES results (Fig. 2a) and to MTS results (Fig. 2b). As already observed in the previous paper, the VES geoelectrical cross-section gives more detail on shallow structures until about 200 m depths, whereas the MTS geoelectrical cross-section provides more information on deeper structures until about 800 m depths.

The profile BH encounters three drill-holes dug for hydrogeological exploration, DH812, 816 and 817, location and stratigraphy logging of which are reported in Fig. 2. It is therefore possible to correlate the geoelectrical cross-section with the stratigraphy. Red colour indicates a resistivity of about 1 $\Omega$ m, which corresponds to a porosity of about 40% on the salty-water curve of the graph in Fig. 3. The intersection of the 40% vertical line with the different water quality curves (Fig. 3), gives an approximate correspondence between the bulk resistivity ρ (colour scale) and the water quality, as shown on the right part of Figs. 2 and 4. A porosity of 40% may be somewhat high and relevant only to a shallow aquifer, a value of 20–30% is more probable for deeper aquifers. However, the corresponding groundwater quality is not significantly changed.

Quaternary sediments in the southwestern part of the profile are more conductive ($\rho <10\Omega$ m), corresponding probably to the presence of salty or brackish shallower aquifers (Fig. 2a). Fresh groundwater (blue colour) can be found locally in shallow Neogene aquifers to the northeastern extremity of the BH profile, and deeper to the southwest. However, there is some ambiguity between deep Neogene aquifers and impermeable Mesozoic shale layers that have a bulk resistivity of the same order (ρ=100–$1000\Omega$ m). Finally, the deepest resistive layer ($\rho >1000\Omega$ m) detected by MTS (Fig. 2b) can be attributed to the igneous bedrock that outcrops at the East border of the studied area (see Fig. 1). Like the Mesozoic layer, the bedrock does not contain any interesting aquifer.

Seven profiles in the studied area (Fig. 1) have been interpreted with the procedure presented above for the BH profile, providing an electrical image of the underground structure from the surface until a depth of about 800 m. It is then possible to realise an electric tomography by drawing the resistivity maps at different depths. Fig. 4 shows three such maps corresponding respectively to the depths 25, 100 and 300 m, and covering the studied area (rectangle of 117×75 km in Fig. 1).

5.1 25-m depth map

5.2 100-m depth map

5.3 300-m depth map

For two decades, a great deal of effort has been devoted to groundwater investigation in the Mekong Delta, the exploration technique used being principally the traditional Vertical Electrical Sounding (VES) method. Its investigation depth rarely reached more than 200 m, so that only Quaternary shallow aquifers could be detected. Unfortunately, the water of these aquifers is frequently salty or brackish. To reach deeper aquifers, some deep drill-holes (>200 m depth) were realised, with a variable success. These deep drill-holes are very expensive and their number was limited.

In the present paper, we show that a combined VES/MTS (Magneto-Telluric Sounding) methodology gives a good electric image of the ground structure from the surface until great depths (∼800 m). Geological information inferred from some drill-holes provided good constraints on the stratigraphic interpretation of the geoelectrical results. The resulting electric tomography is made of electrical images in horizontal planes of different depths. This methodology is less expensive than an extensive deep drill-holes program, especially in the case of the Mekong Delta, which covers a very vast alluvial plain and where most fresh groundwater resources can be found only at great depth [1]. The electric tomography constitutes an interesting and cheap guide for the evaluation of the quality of the groundwater resources of the Mekong Delta.