Você está na página 1de 111

Journal of Hydrology 305 (2005) 4062

www.elsevier.com/locate/jhydrol

Using geochemical data and modelling to enhance the


understanding of groundwater flow in a regional deep aquifer,
Aquitaine Basin, south-west of France
*

L. Andre , M. Franceschi, P. Pouchan, O. Atteia


Institut EGID Bordeaux 3, 1 Allee Daguin, 33607 Pessac, France

Received 15 July 2003; revised 9 August 2004; accepted 19 August 2004

Abstract

In deep aquifers the complex flow pattern originating from the geological structure often leads to difficult predictions of water origin,
determination of the main flow paths, potential mixing of waters. All these uncertainties prevent an efficient management of the

resource. In the context of the Aquitaine basin an original modelling approach suggests that geochemical data can be used to identify
flow directions where geological and hydrogeological data are too scarce to provide sufficient information.

In the Eocene sands aquifer, the major patterns of groundwater geochemistry suggest the presence of two distinct areas within
the aquifer. In the north and the east, waters exhibit sodium bicarbonate or sodium sulphate facies, and moderate total
dissolved solids related to high sulphate concentrations. In the south, waters are characterised by calcium bicarbonate facies
and low total dissolved solids.

Sulphur isotopic ratios provided key information on the origin of sulphur in solution (meteoric, gypsum dissolution, pyrite
oxidation) and also on the intensity of the geochemical processes involved in the dissolution of minerals and the concentration
evolution.

A geochemical model was developed to analyse the processes generating the chemical composition of each sampled water. At
the aquifer scale, four main geochemical processesdissolution, redox, acidbase reaction, exchangeof varying intensity
could explain most of the observed spatial variability in groundwater composition.

Among several potential reaction schemes at each point, only one allowed to reproduce the independent variables (pH and
13

C). The developed model was used to select the most probable water pathways at the aquifer scale. In this context,
geochemistry clearly demonstrates the role played by subsurface structures on water flow velocities and residence time in
their vicinity. In addition, the concentrations of several ions could only be justified by the aquitardaquifer interactions.

q 2004 Elsevier B.V. All rights reserved.

Keywords: Water geochemistry; Groundwater hydrogeology; Isotopes; Modelling; Ion exchange; Leakage; Aquitaine

* Corresponding author. Address: Institut EGID Bordeaux 3, 1 Allee Duguin, Pessac 33607, France. Tel.:C33 5 57 12 10 17; fax: C33 5 57
12 10 01.

E-mail address: laurent.andre@egid.u-bordeaux.fr (L. Andre).

0022-1694/$ - see front matter q 2004 Elsevier B.V. All rights reserved. doi:10.1016/j.jhydrol.2004.08.027

L.
41

1. Introduction

The chemical composition of groundwater is


controlled by many factors that include composition
of precipitation, geological structure and mineralogy
of the watersheds and aquifers, and geochemical
processes within the aquifer. The interaction of all
factors leads to various water facies. Usually, major
ions studies are used to define hydrochemical facies
of waters and the spatial variability can provide
insight into aquifer heterogeneity and connectivity
(Murray, 1996; Rosen and Jones, 1998). With the
development of geochemical modelling, trace, major
and isotopic elements are used to infer the physical
and chemical processes controlling the water
chemistry and to delineate flow paths in aquifer
(Eberts and George, 2000; Plummer and Sprinckle,
2001; Guler and Thyne, 2004; Andre, 2002).

wells and exploratory borings were used to draw


structural maps and to define the hydrodynamic
properties of the aquifer. More recently, the conflicts for
water use have led to detailed studies of the
hydrogeology of the aquifer (Labat, 1998). The
hydrodynamic characterisation allowed identification of
some main flow lines and the approximate behaviour of
the aquifer near geologic structures.

The water chemistry has been first approached by a


description of the spatial variation of some ions and
by the calculation of the residence time (Blavoux et
al., 1993). The approach presented here uses
geochemical indicators and modelling, both linked to
the structural or deep sedimentologic features in
order to locate potential mass transfers. After a
detailed analysis of the regional geology and the
sampling locations used for observations, the
physico-chemical, chemical and isotopic data are
presented. Specific attention will be paid to isotopic
34

data, particularly S.
The Eocene sands aquifer, located in the south
Aquitaine Basin, constitutes an important water
resource used for various purposes (thermal water,
potable water.). This aquifer has been mostly studied
since the end of the 50s along with the development of
oil exploration in this part of France. The production

Geochemical modelling is needed to deal with the


interactions of several elements and dissolution
processes. A pre-requisite step is devoted to a detailed
inspection of the reactions that can occur, or not, in

2. Hydrogeologic and geologic settings


the studied aquifer. The modelling results are then
validated using independent variables. Particular
attention is paid to the points having an original
behaviour or water composition that could not be
predicted using the first group of assumptions.

The results of the geochemical modelling step are


then interpreted in terms of potential flow at the
aquifer scale. Similar regional approaches were used
by several authors to interpret the geochemical
evolution in regional aquifers (e.g. Hendry and
Schwartz, 1990; Weaver and Bahr, 1991; Gerla,
1992; Sracek and Hirata, 2002). The quantitative
approach of the reaction linked with the information
on the aquifer solid and isotope data allowed to
enhance the confidence in geochemical models. The
modelling results were able to locate flow barriers at
the regional scale and to elucidate the circulations in
complex geological structure. The developed
approach, involving the analysis of geochemical
reactions, validation variables and aquitardaquifer
interactions, seems to be applicable to other basins.

The Aquitaine Basin is limited in the east by the


foothills of Montagne Noire, in the south by the
North Pyrenean Piedmont, in the west by the Atlantic
Ocean and in the north by the Poitou Plateau. We will
be interested here in the southern sector of the basin
located to the south of the Garonne river.

The Eocene sands aquifer, located in this part of the


Aquitaine Basin, constitutes a major aquifer used for
drinking water, agriculture, gas storage and as a
thermal resource. This aquifer extends over 150 km
from east to west and 200 km from south to north and
constitutes a part of a multi-layer system. To manage
this resource, the interactions with the under- and
over-lying aquifers are fairly important because the
Eocene sands aquifer has very restricted outcrops.
The average thickness of the quartz sand deposit is
estimated as 50 m, with a high porosity of 2035%,
and average permeability, estimated from aquifer
testing and modelling results to approximately 3!
K5

10 m/s (Labat, 1998). The average interstitial


velocity, using a gradient of 0.001 and an effective
porosity of 20%, is close to 5 m/y. Groundwater flow
is mainly oriented from SE towards NW but outflow

42

L. Andre et al. / Journal of Hydrology 305 (2005) 4062

Fig. 1. The study region and the piezometric map of the Eocene sands aquifer. (Institut EGID, 1999)

from the aquifer is not completely identified


(Fig. 1). In the centre of the basin, the

exploration, have affected this surface. They


include local structures (e.g. the domes of
Garlin or Saint-Medard) as well as larger
structures (e.g. the Audignon anticline, located
at the western border of

14

estimated age of the groundwater (using C


data) is close to 2025 ky. This is consistent
with the effective advection calculated above.

Sands deposits mineralogy is very poor,


containing mainly quartz, augmented with
calcite, and, at places, dolomite and K-feldspars
(Andre, 2002). Detrital sediments were eroded
from the Massif Central, the Montagne Noire,
and the emerging Pyrenees moun-tains. These
eroded sediments were deposited in vast marshy
plains. Geological structures, identified by the

the aquifer, and the CeltAquitaine flexure


which seems to divide the aquifer in two
quite distinct zones) (Fig. 2). It seems that
most of the structures are related to the
deformation of deep Triassic sands (Rey,
1995; Serrano, 2001).

The DanoPaleocene calcareous and


dolomitic aquifer, lying beneath the Eocene
sands, is separated from them by clay
deposits of variable thickness, from 10 to 100
m. The Eocene sands aquifer is covered by
several hundreds of meters of Tertiary
molasses. Aquitard mineralogy, although less

well investigated, is richer, with quartz,


feldspars, mica and several clay types
(detrital limestone levels with sandy-clay
deposits). At the bottom of the molasse
deposits, crystallised gypsum has been
observed in cuttings at many places.

L. Andre et al. / Journal of Hydrology 305 (2005) 4062


43

Fig. 2. Location of the sampling points and major faults. The numbers with symbols correspond to sampling points having used to the
geochemical modelling.

measurements were made, shielded from the air, in a


flow through cell during sampling.
3. Water chemistry

3.1. Sampling and analyses

Electrodes were calibrated in the laboratory and in


the field. For titration of the total alkalinity

Waters were sampled during various field campaigns distributed over two years. This is due to the
irregular use of some wells (irrigation, geothermal
power heat). Owing to the residence time of the
water in the aquifer, the time lags may not play a
dominant role on the chemical composition of the
waters and consecutive samplings at some wells
showed a constant chemical composition.

and the total sulphides, the temperature of the


measuring cell was hold at the sample temperature.
Alkalinity, determined by using a fixed endpoint
methodology (pH 4.3), was titrated by 0.1 M hydrochloric acid with a precision of 2%. For titration of
the species of sulphur, the method developed by
Boulegue and Popoff (1979) was used (OhayonCourtes, 1992). The sulphur species were measured
in strongly basic environment by a solution of HgCl2.
The potentio-metric follow-up, between a specific

Several parameters were measured in the field:


temperature, pH, alkalinity and redox potential

electrode Ag/Ag2S and a double junction reference


electrode (AgCl) allowed the determination of an

inflexion point corresponding to the sum H2SCHS .


This titration allows the determination of
concentrations with a precision of 56%.

Analyses of sulphur isotopes within reduced and


oxidised forms of sulphur in solution was undertaken.
The reduced forms were precipitated in the form of

44

L. Andre et al. / Journal of Hydrology 305 (2005) 4062

zinc sulphide by addition of zinc


acetate. This procedure precludes
any later oxidation of sulphides to
sulphates and allows a single sample
34

to be used for S analyses of both


sulphides and sulphates.

2 K
4 ,
2C

Anions (Cl , SO
2C

F, Br ) and
C

cations (Ca , Mg , Na , K )
concentrations in filtered samples
(acidified for cations) were
measured by ionic chromatography
on a Dionex DX-120. Trace
elements (Al, Mn, Fe, Ba, Sr, Rb,)
were measured in filtered and
acidified samples by graphitefurnace atomic absorption (Perkin
Elmer SIMAA 6000).

Isotope analyses were conducted


during two sampling campaigns in
20002001 as part of this study.
Analyses were performed by the
Environment Isotopes laboratory of
Waterloo, in Canada. Isotopes
analyses from another study
(Blavoux et al., 1993) were also
used for the interpretation.

3.2. Results

The analytical results are given in


Tables 1 and 2. Waters from the
Eocene sands aquifer mostly show a

calcium bicarbonate facies, a


common facies for waters stemming
from deep aquifers (Marini et al.,
2000). Sodium bicarbonate waters
remain excep-tional and sodium
sulphate waters occur at a few wells.
Most of the waters have a total
dissolved solid concentration lower
than 1000 mg/L. This is uncom-mon
considering the depth of the aquifer
and the residence times. The
chemical composition interpre-tation
clearly evidences three main
geochemical facies (Fig. 3), a
calcium bicarbonate facies, a sodium
bicarbonate facies with low total
dissolved solids (e.g. sample from
well 13) or average mineralised (e.g.
sample from well 17) and a sodium
sulphate facies (e.g. sample from
well 4).

The chloride concentrations vary


over the studied area from 0.20 to
more than 14 meq/L. Two zones can
be identified in this aquifer. One
zone is located in the south of the
basin, with chloride concentrations
on the order of 0.150.30 meq/L
with a few samples reaching
concentrations of about 1.50 meq/L
(Wells 19, 21, 22). Concentrations
show no clear trend, with values
close to the concentrations observed
for waters from the recharge area.
Schoeller (1956) quotes, as an
average chloride content of rain
water, a value close to 0.08 meq/L.
With a factor of concentration close
to 2.5, corresponding to local
evaporation rates,

concentrations lie between 2 and 23


meq/L.

we would obtain an average


concentration of 0.20 meq/L. The
other zone includes all the northern
part of the aquifer. The chloride
concentrations range from 1.4 to 16
meq/L, which seems to indicate
another processes occurring in
solution.

Sulphur isotope analyses (Andre et


al., 2002) were made on dissolved
sulphate. Gypsum samples, collected
from the top of the aquifer, were
also analysed. The mean values of

Sulphur is present under various


forms, the most common being
dissolved sulphate. Sulphide can
also

d S and d O measured on this


gypsum are respectively
12.72G1.20 CDT and 14.9
SMOW. These values agree with
those quoted in the literature for
sulphate evaporates (Claypool et al.,
1980).

be present but the concentrations


of this species
are always relatively low (lower

34

18

34

The spatial distribution of the d S


values observed in sulphates from
waters indicates four distinct zones
(Fig. 4):

than 1 mg/L),

and represent at most only 10% of


the quantity of total sulphur.
Sulphide arises, mostly, from local
phenomena of bioreduction.

Zone A, along the edge of the


Pyrenees, in the zone of outcrop of the
34

aquifer, where the d S values are

close to values found for sulphates


34

The distribution of the sulphate


concentrations shows the same
regional zonation as the one observed
for chloride. In the south, weakly
mineralised waters have
concentrations lower than 0.50 meq/L,
with, contrasting with chloride, an
increase from south to north, the
concentration going from 0.20 meq/L
near the outcrops to 1.50 meq/L at
well 11. In the north, the

from precipi-tation (C3.2!d S!


C8.2 CDT) (Pearson and
Rightmire, 1980). The value of
18

d O(SO4) is

slightly lower than the expected


precipitation values (C9!
18

d O(SO4)!C13.34 with respect to


SMOW) (Trembaczowski and
Halas, 1984) and results probably

from mixing of various


contributions in this zone of
outcrops (Pelissier-Hermitte et al.,
2000).

Zone B, the largest zone, which


extends from the Montagne Noire,
in the East, towards the centre of the
34

18

basin. Values of d S and O(SO4)


are close to those measured in the
gypsum crystals sampled in the
molasse cover, at the top of the
aquifer.

Table 1

Chemical composition of waters from the Eocene sands aquifer


No.
Wells names

(
m
S
/
c
m
)

T (8C)
Cond.

c
o
r
r
e
c
t
e
d

pH
Eh

Ca2C
Mg2C
Na

(
m
g
/
L
)

KC
HCO3
Cl

NO3

2K

SO4

[H2S]

(
m
g
/
L
)

(mg/L)
(mg/L)
(mg/L
(mg/L)
(mg/L
(mg/L
(!10

K6

(mV)

1
B
a
r
b
o
t
a
n

HCO3)

NO3)
SO4)
mol/L)

1
0
2
3
7
.
2
3
9
0
7
.
0
7

66.9
50.60

7
.
1
0

6.88
12.16
5.83
170.30
9.48
0
40.01
NM
2
Barbotan 103
36.9
372
7.25
73.1
48.29
6.70
11.74
5.99
176.00
8.71
0
31.62
NM
3
Barbotan Lotus 2

6
.
8
1
5
4
.
4
8
8
.
1
0
1
1
.
3
7
5
.
8
0
1
8
3
.
7
0
8
.
3
8
0
5
1
.
3
9

31

N
M

410

Lectoure

42.5

I
z
a
u
t
e

3010
7.20
72.6
44.64
13.49
548.20
20.30
367.70
321.20
0
651.80
0.00
5

5
3
5
.
5
3
1
9
7
.
3
5
K
7
.
8

Lussagnet 57
45.4
346

4
1
.
8
5

7.29
K12
43.71
7.12
9.37
5.52
178.54
6.59
0
20.32
2.23

5
.
2
3
9
.
3
3
4
.
8
7
1
6
3
.
6
9

7.95
0
15.22

1
6
3
.
0
0

15.57

Nogaro 2

7
.
1
0

51.3

295

2
7
.
6
7

7.35
K59
38.46
4.94
14.30
5.02
155.06
8.16
0
28.19
12.25
8
Gondrin
42.4
349
7.45
35.7
31.06
7.44
23.19
5.76

0
.
0
0
9
D
e
m
u
5
2
.
3
2
9
6
7
.
3
2
K
1
3
0
3
2
.
8
7

4.64
17.11
6.24
159.70
7.41
0
7.93
9.54
10
Beaucaire
34.3
2280
7.44
9.18
134.3
50.6

4
4
.
3
2
1
4
.
5
5
2
1
.
3
8
6
.
8
5
1
7
8
.
6
0

18

7
.
0
9

205.44

49.93

7
9
.
2
0

316

0
1069
1.20
11
Castera-Verduzan
30.4
460
7.48
76.3

0
.
7
6
1
2

Plehaut
27.9
391
7.45
30.3

(
E
L
B
2
)

42.48
10.62
18.58
8.98
167.11
6.98
0
49.11
0.00
13
Eugenie les Bains 2
19.0
357
7.94
K47
22.78
6.26
42.88
7.16
201.52
7.44
0
10.98
26.30

1
4
G
e
a
u
n
e
3
0
.
9

336
7.43

3
2
.
5

K8.1
53.08

3
3
4

5.09
9.20
4.32
181.63
5.83
0
13.53
0.00
15
Geaune Pecorade 101
32.4
330
7.41
38.7
49.84
4.13
6.08
3.52
185.30
5.65
0
12.77
0.00
16
Geaune Bats

7
.
3
1
7
.
6
1
5
0
.
6
0
4
.
2
7
5
.
2
7
3
.
6
9
1
8
4
.
3
5
5
.
2
7
0

11.32

0.83

5
8
.
3
8

17
Blagnac (piscine)
49.6
1377
7.20
K24
12.22
3.23
284.60
10.28
425.46
126.30
0
131.00
3.06
18
Lalbarede
25.7
857
7.77
K18
12.22
6.82
165.20
5.40
281.40
80.15

0
.
0
0
1
9
G
a
r
l
i
n
2
7
.
2
3
5
1
7
.
5
0
2
8
.
8
3
4
.
3
9
8
.
5
8

25.50
3.63

9
.
9
7

188.16
15.03
0
4.63
1.87
20
Lespielle 1
26.5
261
7.39
NM
42.19
5.10
9.93
2.82
173.55
7.46
0
10.88
14.82
21
Lamaze`re
50.1
494
7.31
341
32.18

5
7
.
1
5
1
0
.
3
3
2
9
2
.
3
8
7
.
2
6
0
2
5
.
9
0
0
.
0
0
2
2

Saint Medard

2
4

20.6
305
8.42
K45
22.30
11.91
15.60
5.26
143.06
11.71
0
19.16
1.04
23
Bordes 3
13.4
484
7.46
362
87.16
4.08
5.02
1.40
223.74
11.92
41.69
14.43
0.00

G
r
i
g
n
o
l
s
2
1
.
4
1
6
5
9
7
.
0
8
1
7
4
1
2
3
.
9
4
5
.
6
1
4
1
.
9
0
1
4
.
7

202.52
167.40
0
423.5
0.00

L.

45

Andre et al. / Journal of Hydrology 305 (2005) 4062

46

L. Andre et al. / Journal of Hydrology 305 (2005) 4062

Table 2

Isotopic composition of waters from the Eocene sands aquifer

Wells
18

d O(H2O)

H (U.T.)
13

d C
14

A C
18

d O(SO4)
34

d S(SO4)
34

d S(H2S)
number
( SMOW)
( SMOW)

( PDB)
(p.c.m.)
( SMOW)
( CDT)
( CDT)

1
K8.90
K55.70
!0.8
K12.0
5.5G0.3
4.41
K20.12
4.28
2
K8.90
K57.40
0.2G0.3
K12.0
5.1G0.5
5.33
K16.18
NM
3
K8.60
K57.00
1.3

K11.9
24.2G0.4
3.95
K15.80
K32.25
4
K7.20
K45.00
0G0.4
K9.98
2.5G0.2
16.13
12.23
/
5
K9.40
K61.50
!0.8
K11.2
2.3G0.3
9.19
K3.21
K15.48
6
K8.78
K57.42
!0.8G0.4
K10.96
8.93G0.11

9.11
K1.63
K24.05
7
K8.70
K49.30
!0.8
K13.5
3.3G0.2
12.61
8.76
K20.17
8
K8.00
K50.40
0G0.3
K12.9
2.1G0.3
15.4
11.56
/
9
K7.69
K51.30
!0.7
K14.4
4.5G0.4
12.84
18.93

K16.33
10
K7.22
K49.44
!0.8G0.5
K6.81
4.94G0.08
13.66
15.05
NM
11
K8.18
K53.37
!0.8G0.4
K10.05
4.75G0.13
15.15
10.32
NM
12
K8.20
K50.20
!0.8
K13.8
4.5G0.5
15.28
9.76
/
13

K9.56
K63.67
!0.8G0.4
K12.58
5.24G0.12
14.14
33.18
K10.55
14
K8.60
K54.20
0.5G0.3
K11.4
6.9G0.5
8.30
K3.20
/
15
K7.29
K53.03
! 0.8G0.5
K11.29
9.72G0.11
5.98
K4.01
NM
16
K7.56
K52.54

!0.8G0.4
K11.5
12.81G0.13
6.63
K2.98
NM
17
K7.35
K50.37
!0.8G0.4
K8.41
1.44G0.07
15.23
13.23
7.86
18
K7.32
K48.86
!0.8G0.4
K11.98
2.89G0.06
17.23
13.06
10.78
19
K8.10
K53.90
0.80G0.3
K14.5

!0.7
12.85
13.01
K28.05
20
K8.00
K50.80
1.5G0.5
K14.3
3.0G0.8
11.96
5.79
NM
21
K8.50
K52.40
!0.8
K6.0
3.6G0.2
14.93
17.50
/
22
K7.31
K51.72
0.9G0.5
K8.48
13.09G0.2
14.27

24.37
NM
23
K7.30
K46.10
16.0G0.8
K11.9
77.3G1.0
5.13
4.87
/
24
K7.53
K51.10
!0.8G0.5
K4.15
0.96G0.07
12.83
15.70
/

NM, not measured.

A geochemical model is needed to quantify the


relative importance of each chemical reaction at
Zone C, a stretched shape oriented SWNE, shows
negative to very negative values (from K0.4 to

K20.1 CDT), with a decrease from the south


18

northward. The values of d O(SO4) follow the same


trend with a decrease along the same axis. These
values seem to reflect a different origin of dissolved
sulphates than in zone B, a plausible explanation
being the oxidation of sulphide minerals (Dazy et al.,
1980).

the sampled wells. During this first geochemical


modelling step, only the system Na, Ca, Mg, Cl, S,
CO3, H is considered. Si and K concentrations are not
included due to the lack of information on the
composition of the clay fraction of the aquifer, not
accessible from cuttings samples. Moreover, the
silicate equilibrium does not significantly modify the
acidbase and redox equilibrium, and can thus be
neglected. Trace elements such as Sr, Br, F and Ba
might be integrated in future works after modelling
the behaviour of the major species in solution.

One particular point in the west (e.g. Well 13Zone


D), close to the aquifer boundary, shows significant
34

enrichments in sulphur 34 (d SO33 CDT). A


proposed explanation for this is the action of
bioreduction processes.

4. Modelling strategy

4.1. Choice of mineral phases

The Eocene sands appeared, in some drillings, clean,


with translucent grains, pyrite and lignite traces. At
places the sandy series include argillaceous or
argillaceous-sandstones interbeds. The few min-erals
identified are quartz, calcite, feldspars, kaolinite,
mixture of iron oxides/hydroxides, pyrite in some

L. Andre et al. / Journal of Hydrology 305 (2005) 4062


47

Fig. 3. Trilinear Piper diagram of water chemistry of the Eocene sands aquifer.

samples while gypsum and anhydrite are


present at only one place in the extreme
east of the basin. The presence of few
minerals agrees with the fairly dilute
water samples. Although mineralogical
data are not sufficient for such a large
area, they are unable to explain the
widespread existence of sulphate in
groundwaters.

In the water samples, the saturation


indices have been calculated with the
geochemical code PHREEQC (Parkhurst
and Appelo, 1999). Taking into account
the accuracy of the field measurements
of pH and alkalinity and the
uncertainties of other analytical data,
Nordstrom and Ball (1989) or Busby et
al. (1991) estimate that waters are in
equilibrium with respect to calcite if the
saturation index ranges from K0.1 to
C0.1. For dolomite a saturation index
ranging from K0.2 to C0.2 may indicate
equilibrium. According to this, our
results show that the waters are generally

at equilibrium with respect to calcium


carbonate mineral and slightly
undersaturated com-pared to dolomite
(Table 3). All the waters are

undersaturated with respect to sulphur


bearing minerals (gypsum, anhydrite and
pyrite) and near saturation with respect
to siderite. All others minerals phases
which could control NaCl in a
sedimentary system are largely
undersaturated.

According to these results, calcite,


dolomite, gypsum, halite, siderite and
pyrite are the mineral phases taken into
account to represent our system. Water
interactions with these phases, including
redox reactions, can provide a basis for
the reactions that may occur in the
aquifer.

4.2. Infiltration waters

one well (Fig. 2, Well 23) situated near


the outcrops is influenced by anthropic
It is difficult to identify a composition
that might be typical of the average
infiltrating groundwater during the past
25,000 years. In the study area, the only

activities (NO 3 close to 50 mg/L) and


the water does not represent the original
water. The simplest assumption to obtain

48

L. Andre et al. / Journal of Hydrology 305 (2005) 4062

Fig. 4. Map of sulphur-34 values in dissolved sulphates (Andre et al., 2002).

the original infiltrating water


composition is gener-ally to take the
lowest observed concentration of each
ion within all groundwater samples. This
relies on the fact that most of the waterrock interactions processes tend to
increase ion concentrations. So, the
average concentrations for the
infiltration water, is deduced from both
these lowest observed concentrations
and elemental concentrations in
precipitation through historical times
(Schoeller, 1956): water infiltration
C

temperature of 10 8C, [Na ]Z[Cl ]Z0.2


meq/L, [SO
2C

2 K
4 ]Z0.04

[Mg ]Z0.4 meq/L.

meq/L,

The initial equilibrium pCO2 must be


addressed with more details. In soils, two
extreme conditions can prevail: calcite is
either in contact with a fixed CO2 partial
pressure, the open system, or it is
dissolved by water containing a given
amount of dissolved CO2, the closed
system. Two arguments suggest that water
infiltrating Eocene sands equilibrated with

a fixed CO2 pressure in an open system:


(i) water infiltrates mostly in alluvial
plains and mountain piedmont with
colluvial soils often saturated with
calcite and (ii) equilibrium with calcite
at depth, i.e. with a finite source of CO2,
would have led to alkalinity 510 times
lower than the one observed in
groundwater samples. Considering an
open system, soil pCO2 largely varies

K1

K3

from 10 to 10 atm in active to cold


soils (Clark and Fritz, 1997); it is thus
difficult to estimate an average

infiltration water having a temperature of


10 8C. First, this water reaches

equilibrium pCO2 for the infiltration


water over geologic time.

open system (for different initial pCO2),


and then, the temperature increases from
10 to 60 8C and the water is maintained in

Fig. 5 shows the theoretical evolution of


pH versus temperature for an initial

equilibrium with CaCO3, at 10 8C in an

equilibrium with CaCO3, in a closed


system.

L. Andre et al. / Journal of Hydrology 305 (2005) 4062

49
Table 3

Computed saturation indices of waters (PHREEQC)

Wells
Aragonite Calcite Dolomite Anhydrite Gypsum Fluorite
Halite Quartz
Amorphous

Siderite

CO2

number

silica

3
1
K0.41
K0.27
K0.96
K2.23
K2.07
2
K0.43
K0.30
K1.00
K2.34
K2.18

K0.38
K0.24
K0.90
K2.14
K1.94
4
K0.24
K0.10
K0.29
K1.44
K1.31
5

K0.04

10

0.09

0.08

K0.08

0.21

K2.48

0.40

K2.38

K0.88

K0.70

K0.15

11

K0.01

K0.16

K0.50

K0.02

K2.72

K0.13

K2.53

K2.07

K1.86

K0.05

12

0.08

K0.27

K0.23

K0.12

K2.79

K0.47

K2.72

K2.26

K2.05

K0.18

13

K0.04

K0.20

K0.22

K0.05

K2.52

K0.37

K2.39

K3.18

K2.94

K0.06

14

0.06

K0.16

K0.21

K0.02

K2.94

K0.66

K2.89

K2.62

K2.42

K3.35

15

K3.13

K0.09

20

0.05

K0.28

K0.56

K0.14

K2.73

K0.84

K2.54

K2.88

16

K2.66

K0.18

21

K0.04

0.02

K0.73

0.14

K2.77

0.29

K2.59

K2.53

17

K2.46

K0.27

22

K0.15

0.27

K0.34

0.42

K2.18

0.86

K2.17

K2.92

18

K2.68

K0.33

23

K0.18

K0.01

K0.27

0.14

K2.79

K0.84

K2.56

K2.53

19

K2.28

K0.22

24

K0.08

K0.36

K0.39

K0.22

K0.58

K3.12

K1.20

K8.81

K0.96

0.18
K0.95
0.52

K3.08
K8.52
0.26
K0.92
0.14
K1.76
K3.02
K8.57
0.27
K0.92
0.25
K1.73
K3.00
K8.60
0.33
K0.89
0.07
K1.83
K0.69
K5.44
K0.09
K1.26
K0.03
K1.56

K1.96
K3.00
K8.75
0.37
K0.84
0.58
K2.18
K3.04
K8.60
0.16
K0.95
0.56
K2.04
K2.72
K8.37
0.19
K0.96
0.29
K2.12
K2.84
K8.43
0.15
K0.94
0.18

K2.00

K0.65

K0.64

K2.06

K6.47

K3.08

0.29

K9.03

K0.92

0.32

0.32

K0.89

K2.12

0.58

K1.46

K2.10

K8.40

K3.25

0.28

K9.13

K0.94

0.31

0.01

K0.90

K2.20

0.52

K1.97

K2.00

K8.46

K0.74

0.32

K5.82

K0.93

0.08

0.25

K0.99

K2.22

0.96

K2.65

K1.32

K7.97

K0.69

0.40

K6.46

K0.90

0.12

K0.84

K1.14

K2.60

0.25

K3.23

K2.33

K8.95

K2.41

0.27

K7.97

K0.95

0.26

K0.98

0.15

0.26

K1.14

K2.21

1.07

K2.52

K3.31

K8.68

K3.31

0.31

K8.65

K0.94

0.46

K4.88

K0.88

K2.14

K3.54

K1.18

K2.18

K8.00

0.01

0.15

K6.23

K0.96

0.34

K4.89

K0.95

K1.68

0.20

K1.51

K1.83

K8.28

4.3.1. CaHCO3 system


Plotting pH versus temperature for all the sampling
points (Fig. 5) reveals that most of the points lies on
the same theoretical line; they correspond to the same
model and thus to the same original pCO2 of
approximately 10

K2.5

atm. This unique value of

pCO2 will be considered as the one of the infiltrating


water for the whole basin.

4.3. Elementary processes controlling water


composition

As described in part 3.2, CaHCO3 water is the


predominant type in the Eocene sands aquifer,
produced by dissolution of carbonate minerals such
as calcite. The CO2 produced by the oxidation of
organic matter and root respiration in the unsaturated
zone and dissolved by the recharge water is at the
origin of such dissolution. On Fig. 6, plotting pH
measured in the field, versus pCO2, we can note a
good correlation with the calcite saturation line. So,
equilibrium with calcite and potential dissolution

of dolomite will be used in the geochemical


modelling to set the CaMgCO3 system.

4.3.2. Sulphur case

The evolution of water composition towards CaSO4


type can be explained by using isotope

Fig. 5. Variation of pH versus temperature of emergence waters.


Lines represent pH computed from evolution of infiltration water
at 10 8C in equilibrium with calcite and with different initial pCO2
versus temperature and points represent waters of the Eocene
sands aquifer.

50

L. Andre et al. / Journal of Hydrology 305 (2005) 4062

Fig. 6. Variation of pH versus pCO2. Full line, pH evolution of a solution at equilibrium with calcite. Dashed line, pH evolution of a solution
at equilibrium with dolomite.

values of sulphur. Data allowed us to distinguish two


reactions:

Gypsum or anhydrite dissolution. Where it occurs, this


process has been applied in a closed system consisting
of a confined aquifer with calcite remaining at
saturation. During the gypsum (or anhydrite)
dissolution, delivering sulphate and

2C

Fe

CHCO 34H CFeCO3

The overall reaction is written as


calcium to the water, calcite becomes over-saturated
K
3

and, as it precipitates, the HCO


decreases and the pH increases.

concentration

Pyrite oxidation. This process must be detailed in


order to model it. Waters from the Eocene sands
aquifer revealed small quantities of oxygen (00.15
mg/L) and redox potentials ranging between K50 mV
and 0 mV, signifying slightly reduced media, and
suggesting that pyrite oxi-dation is not associated with
oxygen reduction. Among several other possibilities,
the following reactions involving iron oxides and
pyrite (Holmes

and Crundwell, 2000) were the only combination able


to reproduce the concentrations of Fe, H2S, SO4 in
solution

FeOH3 C3H 4Fe

3C

FeS2 C14Fe

3C

C3H2O

C8H2O415Fe

2C

2 K
4

FeS2 C14FeOH3 C11H C15HCO 342SO

2 K
4

C2SO

C16H

C34H2O C15FeCO3

This reaction shows that carbonate equilibrium is


modified by this process. Two arguments are in
18

agreement with this process: in zone C, the d O(SO4)


values express an origin of dissolved sulphates from an
oxidation of sulphurized compounds and a Eh-pH
diagram of the sampled waters (Fig. 7) clearly indicates
that this reaction should be the prominent one for these
samples. In our approach, pyrite and siderite being at
equilibrium, the reaction process is limited by the
availability or dissolution rate of iron hydroxides. The
dissolution of these solids is quite slow and they can
thus remain in the solid phase while slowly dissolving.
One may argue that the presence of pyrite and iron
hydroxides in the same sediment can be surprising as
one species occurs in reduced media while the second
one is present in oxidised media. However, in
sediments, reduced and oxidised periods can occur
(Dubreuilh, 1987; Rey, 1995) and cuttings from wells
drilled in zone C (Cazal et al., 1967; Rechiniac, 1962)
report important quantities of

L. Andre et al. / Journal of Hydrology 305 (2005) 4062


51

Fig. 7. Stability fieldsof oxides, carbonates


and iron sulphur, at 25 8C,
K

[H2S]Z[HS ]Z10
3C

2C

K6

[Fe ]Z[Fe ]Z10

mol/L,

K5

mol/L, [Ctotal]Z3.10

K3

2 K
K4
[SO 4 ]Z2!10

mol/L,
mol/L Diagram
obtained with equilibrium constants from
Garrels and Christ (1965). Points represent
waters in which sulphates come from pyrite
oxidation and gypsum dissolution and crosses
symbolise waters in which sulphates come
from gypsum dissolution exclusively.

pyrite in the aquifer as well as at the


top and bottom of the unit. These
investigators reported also the
presence of mixtures of iron
hydroxides. The choice of iron
hydroxides instead of others iron
oxides will be discuss below.

4.3.3. Cation exchange

This process explains the observed


C

increase in Na concentration
without an associated increase in

Cl concentration. Many studies


have shown this kind of process
(Back, 1966; Freeze and Cherry,
1979; Thorstenton et al., 1979;
Chapelle and Knobel, 1983; Appelo
and Postma, 1999). In an aquifer
where carbonates minerals are
present, cation exchange may be
accompanied by calcite dissolution
2C

2C

(and dolomite), as Ca (Mg ) is


removed from solution and replaced

have a unique solution, provided the


data are of good quality. Although the
modelling system by itself might be
consistent, the assumptions might be
wrong; for instance we might be able
to solve a system by adding a small
amount of dolomite, while Mg may
come from another source. To
overcome this difficulty it is important
to have validation variables that may
confirm or reject our hypotheses.

by Na . The carbonate mineral


4.4. Model variables validation
dissolves and provides more Ca

2C

2C

(and Mg ) in solution to exchange


C

K
HCO 3

with Na and causing


concentration to increase. In the
Eocene sands, the series sometimes
include argillaceous or argillaceoussandstones interbeds, able to play a
role as ion exchangers. Depending
on location, these layers show
variable thickness. Under these
conditions, ion exchangers will be
considered in the modelling
approach and will be used to explain
Na-HCO3 type of some waters.

Under the chemical conditions


considered here, excluding redox
reactions in a first approach, seven
conditions are imposed: equilibrium
with calcite, equilibrium or addition of
dolomite, addition of gypsum and
halite, ion exchange, starting pCO2 of
infiltration water and electro
neutrality. Seven ion concentrations
are considered, i.e. Ca, Mg, Na, CO3,
SO4, Cl and H. We are therefore in a
fully determined system which should

The pCO2 is different from the other


variables as an initial value is
specified over the whole area. In
fact all other processes or added
minerals are adjusted at each point
to reproduce the local concentration
of the ions of interest, as it is done
in any modelling study (Plummer
and Sprinckle, 2001; Guler and
Thyne, 2004). As all the processes
may change carbonates equilibrium,
the value of pCO2 may vary along
the water pathway and can be used
to validate the differences between
the points. But this value is difficult
to measure. So, in this condition, we
decide to use an other variable
which depends of all processes: the
pH. Moreover, the attention brought
on the measure of pH in the field
allows us to consider this variable as
a good validation variable.

13

d C can also be used as a validation


variable because its relative
concentration in groundwater will

depend only on the geochemical


processes cited above (Clark and

Fritz, 1997). Thus if the calculated


amounts

52

L. Andre et al. / Journal of Hydrology 305 (2005) 4062

of calcite and dolomite dissolved are


13

the right ones, the C values should


match. The underlying assump-tion is
13

that C values come from the two end


members: calcite and soil water.
Assuming here the classical literature
13

values, i.e. 0 for CO3 in calcite


and dolomite and K23 for soil
13

CO2 gas, we consider that all the


waters infiltrate under the same
K2.5

conditions (soil pCO2Z10


atm)
and that calcite dissolution at
equilibrium with CO2 of the soil gives
13

a value of d C of the dissolved CO2


equal or near to K12 (Clark and
Fritz, 1997). The further evolution of
this value depends only on the
different processes (dissolution or
precipitation) affecting carbonate
minerals during the groundwater flow
in a closed system. The precipitating
13

calcite has a d C value starting from


the isotopic equilibrium with the
groundwater and slightly enriched by
C2 (Clark and Fritz, 1997); the
same hypothesis will be used for
siderite.

4.5. Comparison of modelling


approaches

All the chemical processes detailed


above are well known and are used in
studies in literature (Murray, 1996;
Rosen and Jones, 1998; Edmunds et
al., 2000; Marini et al., 2000). As the
chemical composition of groundwater
in such systems mainly results from

salt dissolution and carbonate


equilibrium, it is possible to fit most
of the groundwater composition by
adjusting the dissolved amount of the
various minerals. This approach
generally gives results in good
agreement with observed data for
major chemical parameters. Modelling
results are thus used to suggest
hypotheses on the water pathway
within the aquifer. Although other
geochemical processes occurring
within the aquifer (exchanges with
aquitards or mixing with others
aquifers) are assumed to explain
anomalous chemical composition,
these hypothesis are often not
confirmed by additional data. A more
detailed approach, incoming recently,
implies the use of inverse models
including mixing, dissolution/precipitation salts, cation exchange that may
explain changes of groundwater
chemistry between points lying on the
same flow line. These models often
concerned major ions present in
solution (Gosselin et al., 2001; Sracek
and Hirata, 2002; Martinez and
Bocanegra, 2002). Inverse modelling,
although including more variables, can
still consist to fit the variable amounts
of minerals added to the solution, if
validation variables are not

used. A typical adjustment can consist


in using a variable equilibrium pCO2
for each modelled point. Other authors
validate their models using additional
data for model calibration. They use
13

34

isotope data, like C or S, or


elements in solution like total
dissolved carbon (Walvoord et al.,
1999; Plummer and Sprinckle, 2001;
Guler and Thyne, 2004). Our model

was inspired from this approach but,


in our case, the system is much more
constrained. The origin of sulphate by
the pyrite dissolution may be used to
illustrate this purpose: the reactions
must satisfy conditions on the
concentrations of sulphur and iron, the
redox potential, the availability of O2,
sulphur isotope values and finally pH.
All these parameters could not be
matched by simple dissolution of
pyrite; the complex reaction scheme,
involving two dis-solving and one
precipitating solid, seems to be the
only one to fulfil all conditions. The
few available informations on the
sediment solid phase tend to confirm
our hypotheses. We also constrain the
carbonates equilibrium by setting a
unique pCO2 for all infiltrating waters.
An inverse modelling of this value
would have led to an easy fit of pH.
But, in turn, this might have hidden
the effects of redox reactions on pH or

CO2 satisfies carbonate equilibrium,


13

pH and d C.

During water-rock interactions


numerous equili-bria are involved
but they can be ordered according to
the dependency among the different
reactions, in the considered case it
is: redox reaction, gypsum
dissolution, carbonate equilibrium,
halite dissolution. At each step, a
complete validation of all
parameters influenced by the
geochemical processes is done with
a justification by the composition of
the solid phase. The major objective
is to have as less as possible variable
phase amounts.

other origins of CO2. In addition,


13

being a good tool for validation, d C


also allowed to identify potential
sources of additional CO2. In that case
again, several hypothesis were tested
and, for the points where anomalous
CO2 occur, only a direct addition of

5. Modelling results

All groundwater samples were


modelled using the approach
described above (Fig. 8). It was
possible to reproduce most of the
water compositions (Tables 46).

L. Andre et al. / Journal of Hydrology 305 (2005) 4062


53

Fig. 8. Geochemical processes governing the chemical composition of waters from the Eocene sands aquifer.

Table 4

Comparison between measured and modelled chemical composition in south-western zone

Geaune

Lussagnet
Well 13

Barbotan

Initial PCO2 (atm)


10K2.5

10K2.5

10K2.5

10K2.5

NaX
5!10

K2

5!10

K2

1.7

4!10

K1

Halite
/

Dolomite
/

Gypsum
K6!10

K2

K1!10

K1

K6!10

K2

Pyrite
K2.86!10

K2

K4.29!10

K2

K2.86!10

K2

K1.43!10

K1

Fe(OH)3
K4!10

K1

K6!10

K1

K4!10

K1

K1

Calcite
C2.73!10

K1

C3.92!10

K1

C1.77!10

K1

C4.36!10

K1

Siderite
C4.23!10

K1

C6.38!10

K1

C4.25!10

K1

C1.14

2(g)

2.5!10

K1

Measured
Computed
Measured
Computed
Measured
Computed
Measured
Computed

values
values

values
values
values
values
values
values

PH
7.35
7.54
7.37
7.63
7.86
8.12
7.10
7.85
HCO3
2.99
2.25

2.84
1.96
3.21
2.43
2.77
1.47
Ca2C

1.24
1.01
1.09
0.93
0.52
0.38
0.87
0.63
Cl

0.16
0.20
0.20
0.20
0.24
0.20
0.25
0.20
K

0.10
0.00
0.14
0.00

0.17
0.14
0.16
0.04
2C

Mg

0.18
0.19
0.26
0.19
0.26
0.09
0.28
0.16
Na

0.24
0.25
0.26
0.25
1.85
1.83
0.57
0.60
2K

SO4

0.14
0.12
0.19
0.18
0.12
0.13

0.31
0.27
Al

total

! LD
8.3!10

K5

2.8!10

K4

1!10

K4

! LD
7.1!10

K5

! LD
2.7!10
Si

K4

total

0.25
0.13
0.30
0.13
0.22
0.08
0.29
0.15
Log(PCO2)
K2.03
K2.35
K2.08
K2.50
K2.60
K2.99
K1.80

K2.81
13

C ( PDB)

K11.4
K12.5
K11.2
K13.0
K10.05
K12.5
K12.0
K14.0

LD, limit of detection (C); mineral precipitation from solution; (K), mineral dissolution from solid to solution. All data for minerals or exchangers
quoted in the upper part of the table indicate the amounts of these minerals dissolved/precipitated in solution to reach the measured concentration.
They are expressed as mmol/L of solution. In the low part of table, pH are expressed in standard units whereas measured and computed values are
expressed in mmol/L. Some wells are very near and the chemical compositions of waters are similar. In these conditions, Geaune represents waters
from wells 1416, Lussagnet represents waters from wells 5 and 6 and Barbotan represent waters from wells 13.

T
a
b
l
e
5

Comparison between measured and modelled chemical composition in south-eastern zone

Well 9

I
n
i
t
i
a
l
P

Well 12

(
a
t
m
)

Well 8

Well 11
Well 7

0
K
2.
5

Well 20

Well 19

0
K
2.
5

1
0
K
2.
5

1
0
K
2.
5

1
0
K
2.
5

1
0
K
2.
5

8
!
1
0

10K2.5

K
1
CO2

2
.
5
!
1
0
K
1

6
!
1
0
K
1

H
a
l
i
t
e

/
NaX
/

5.5!10

7.5!10

K1

K1

4.50!10
5.50!10

K1

K1

K2.25!10

K1

Dolomite

K
2
.
8
!
1
0
K
2

K6!10

K1

K
7
!
1
0

K3!10
K2!10

K1

3
K1

K2!10

K1

Gypsum

K
2
.
1
4
!
1
0
K
2

/
K8!10

K2

K1
K9!10

K2

K
1
!
1
0
K

K5.5!10
K3!10
K1!10
K3!10
Pyrite

K1

K1

K1

K2

C
1
.
0
1
!
1
0

Fe(OH)3

K4!10
K1!10
K3!10

K1

K1

C
3
.
1
5
!
1
0

K1

K1.4!10

K1

Calcite

C
1
.
4
4
!
1
0

C1.10!10

K1

C1.44

C1.55!10
C8.14!10
C4.12!10
C1.12!10
C3.82!10

K1

K1

K1

K1

K1

Siderite

C4.23!10

K1

M
e
a
s
u
r
e
d

CompuMeasured
CompuMeasured
CompuMeasured
CompuMeasured
CompuMeasured
CompuMeasured
Compu-

values
ted values
values

P
h

ted values
values
ted values
values
ted values
values
ted values
values
ted values
values
ted values

7
.
3
3
7
.
2
3
7
.
4
8
7
.
6
1
7
.
4
0

7.28
7.45
7.59
7.50
7.48
7.39
7.51
7.50
7.58
HCO

2.62
2.67
2.93
2.28
2.56
2.56
2.78
2.43
2.67
2.86
2.80
2.64
2.99
2.92
3

C
a
2
C

0
.
8
0
0
.
9
5
1
.
1
1
0
.
9
4
0
.
8
8
0
.
9
6

1.00
1.03

0
.
2
1

0.78
0.96
1.05
1.11
0.75

0
.
0
5
0
.
1
5

0.85
Cl

0.22
0.20
0.20
0.20
0.13
0.20
0.21
0.20
0.20
0.20
0.21
0.20
0.43

0
.
0
8
0
.
0
7
0
.
0
2
0
.
0
9
0
.
0
6
M
g
2
C

0.43
C

0.15
0.05
0.18
0.07
0.14
0.04

0
.
1
9
0
.
1
6
0
.
6
0

0.60
0.18

0
.
8
2

0.17
0.43
0.40
0.31
0.30

0
.
1
0
0
.
1
1

0.21
0.18
0.32
0.32
Na

0.74
0.74
0.92
0.94
0.58
0.64
0.75
0.74
1.00
0.99
0.43
0.45
1.02
1.01
SO

2K

0.08
0.08
0.83

0
.
5
4
0
.
5
2
0
.
2
9
0
.
2
8
0
.
1
1
0
.
1
2
0
.
0
4
0
.
0
4
4

!
1
0
K
5

!
1
0
K
5

!
1
0
K
3

!
1
0
K
5
Al

total

9.6
28
1.1

!
1
0
K
4

9
!LD
2.4
1.30
0.57
7.4
7.3

!
1
0
K
4

!
1
0
K
4

1.2
4.7
7.4
5.5

!
1
0
K
5

!10
!10
!10
!10
!10
Si

K5

K5

K
2
.
1
3

K5

K5

K5

total

0.36
0.25
0.25
0.13
0.34
0.23
0.24
0.11
0.29
0.13
0.23
0.11
0.21
0.11
Log(PCK2.00
K1.81
K2.21
K2.43
K2.05
K1.90
K2.22
K2.41

K
2
.
1
9
K
2
.
1
5
K
2
.
2
8
K
2
.
2
2
K
2
.
3
1
O
2
)

13

C (

K14.4
K12.1
K12.6
K10.2
K13.5
K12.2
K13.8
K11.0
K12.9
K10.9
K14.3
K12.2
K14.5
K11.1
PDB)

5
4
(C): mineral precipitation from solution; (K): mineral dissolution from solid to solution. pH are expressed in standard units whereas amounts of
minerals (upper part of table) and measured and computed values (low part of table) are expressed in mmol/L.

L.

Andre et al. / Journal of Hydrology 305 (2005) 4062

L. Andre et al. / Journal of Hydrology 305 (2005) 4062

55
Table 6

Comparison between measured and modelled chemical composition in northern zone

Well 17

Well 4

Well 24
Well 18
Well 10

Initial PCO2 (atm)

10K2.5

10K2.5

10K2.5
10K2.5
10K2.5

NaX

11

20

15
5

14

Halite

K3.4

K8.9

K4.4
K2

K1.2

Dolomite

K9!10

K1

K2.62

K2.5
K5.6!10

K1

K5.7

Gypsum

K1.5

K7

K6
K6.5!10
K15

K1

Calcite

K2.11!10

K1

C3.6

C4.77
C9.05!10

K1

C11.27

CO2(g)

2.2

5!10

K1

5!10

K1

Measured
Computed
Measured
Computed
Measured

Computed
Measured
Computed
Measured
Computed

values
values
values
values
values
values
values
values
values
values

pH
7.20
7.47
7.20
7.25
7.08
7.17
7.77
8.50
7.44
7.06
HCO3
6.98
6.83
6.03
6.03
3.32
3.26
4.61
3.22
3.37
3.08
Ca

2C

0.31
0.30
1.12
0.94
3.1
3.45

0.31
0.11
3.36
3.72
Cl

3.56
3.60
9.05
9.10
4.72
4.60
2.26
2.20
1.40
1.40
K

0.26
0.53
0.52
0.96
0.38
0.14
0.14
0.28
0.46
0.96
2C

Mg

0.13
0.12

0.56
0.51
1.89
1.90
0.28
0.08
2.10
2.19
Na

12.32
11.94
23.73
23.67
6.14
5.96
7.15
6.21
13.68
13.86
2K

SO4

1.37
1.39
6.79
6.00
4.41
4.43
0.61
0.64
11.13

10.94
Al

total

2.2!10
4!10

K4

K4

5.6!10

K4

1.0!10

K4

2.6!10

K4

4.30!10

K4

! LD
3.8!10
Si

K5

total

0.3
0.23
0.14
0.17
0.20
0.09
0.14
0.11
0.29
0.14
Log(PCK1.32
K1.7
K1.56
K1.62
K1.84
K1.93

1.3!10
4!10

K5

K4

K2.33
K3.20
K2.12
K1.7
O2)

13

C (

K8.41
K6.6
K9.98
K5.7
K4.15
K7.1
K11.8
K9.4
K6.81
K8.69
PDB)

(C): mineral precipitation from solution; (K): mineral dissolution from solid to solution. pH are expressed in standard units whereas amounts
of minerals (upper part of table) and measured and computed values (low part of table) are expressed in mmol/L.

Among the processes, calcite equilibrium is the major


factor influencing pH evolution, the pH value being also

influenced by pyrite oxidation where it occurs. A good


agreement is found between calcu-lated and measured
in situ pH, the maximum variations reaching only 0.3
pH unit except for two points having variation near 0.7
pH unit (Fig. 9). This result is very convincing because
the implied processes are numerous and the pH is a
quite sensitive variable. Moreover, the range of
variation (between 7 and 8.3) is quite wide. Contrary to
other regions, the western zone is the only one where
the modelled pHs are approximately 0.2 pH units higher
than the analytical data. All proportions of sulphate
originating from pyrite dissolution were calculated
34

using the only available value, d SpyriteZK20,


obtained from literature (pyrite samples were too small
34

to measure S). A change in this isotopic composition


of pyrite modifies significantly the amount of dissolved
pyrite

Fig. 9. Comparison between pH measured and computed. Error


bars on measured pH correspond to measurements errors (5%).
Computed pH have been obtained from three different data bases
(WATEQ, MINTEQ and PHREEQC).

56

L. Andre et al. / Journal of Hydrology 305 (2005) 4062

which may, in turn, modify the pH


value by 0.2 to 0.3 units. This may
be the source of the observed small
discrepancy.

13

Measured and computed d C


values are similar (Tables 46). The
variations do not exceed 4. Owing
to the uncertainty in values
presented in literature (calcite and
dolomite: 0G2, CO2 gas in soil:
C23 G2) (Clark and Fritz,
1997), the results can be considered
to agree well. More precision could
13

be obtained in future studies if d C


of the calcite from the aquifer would
be measured.

The overall dissolved-solids


concentrations can be obtained mainly
from gypsum dissolution and slight
halite (NaCl) dissolution at some
locations. Although this process is in
agreement with the water composition, waters are largely
undersaturated with respect to both
minerals. This means that gypsum and
halite minerals are not present inside
the active part of aquifer, a conclusion
consistent with the deposit age of the
aquifer and its fairly fast renewal.
Then another process may deliver to
the aquifer water that has previously
dissolved gypsum (gypsum

Throughout the aquifer small amounts


of dissolved dolomite were needed to
model the measured concentrations of
Mg. Again, most of the waters being
undersaturated with respect to
dolomite, the Mg rich waters may
come from vertical leakage and not
from the local solid phase. However,
for Mg content, calcite can also play
an important role as non-negligible
amounts of Mg are often present in
calcite. With the existing data, we
cannot answer this question.

As expected, the variations in Na


concentrations are influenced by the
ion exchange process. Despite our
poor knowledge of the argillaceous
phase, the typical ion exchange
constant used seems to give results
in agreement with measured concentrations (Tables 46). Ion exchange
occurs at only few places, but
significantly modifies the
groundwater composition at such
places.

As said above, the measured alkalinity


could not be modelled at a few wells
(wells 4, 10, 17). Among other
possibilities the only process that has
been able to reproduce the measured
data is the addition of CO2. For these
13

wells d C was not used as a


validation

34

dissolution is needed to reach the d S


values of sulphate existing in
groundwater). The source of gypsum
must then be search in the aquitard.

variable but as an understanding tool.


13

In fact the d C

values of these waters were only


compatible with a deep origin of
13

CO2 with a mean d C of K5

(Clark and Fritz, 1997), which could


result from metamorphic gas
production. This process has been
observed elsewhere (Chiodini et al.,
1999) and may be explained by the
presence of important structures
facilitating upward gas flow. It is
interesting to notice that at two sites
(wells 4 and 10) anomalous radon
content has also been observed
(Franceschi, 2005). As radon is
known to come from deep granitic
bedrock, a deep origin of CO2 seems
to be plausible.

lead from one point to the other.


This is the basis of how we used the
geochemical model to delineate
groundwater flow paths: some water
paths are not possible and some
preferential flow directions can be
drawn if one compound is increasing
along the flow line. A flow barrier
constraint is present at several
places; it is not possible to go from
Demu (well 9) to Nogaro 2 (well 7),
nor from Beaucaire (well 10) to
Gondrin (well 8). Furthermore, the
major concentrations trends are from
east to west in the northern area and
from south to north in the southern
area. Fig. 10 depicts the major flow
directions derived from the
modelling results; the four major
flow lines with the associated
chemical composition evolution and
models are given in Tables 46.

6.1. Major flow directions


6. Geochemical arguments for flow
delineation

The developed model can also help


to understand the connections
between different sampled points. In
fact a similar composition of two
waters is not a proof of
hydrodynamic connection. However,
if an hydraulic connection is
assumed, one must be able to
explain the chemical processes that

One main conclusion from the


reconstruction of the groundwater
composition by modelling is that only
two major zones exist in the studied
area: the south and the north. Within
each zone the differences in
groundwater composition arise from
the local intensity of the geochemical
processes. These differences may be
used to assess the major flow
directions. The main assumption
underlying this approach is that,
except

L. Andre et al. / Journal of Hydrology 305 (2005) 4062


57

Fig. 10. Major flow direction derived from modelling results.

waters from the north, while the higher


head

gradient
where mixing of large volumes occurs,
or for the specific case of ion
exchange, the concentrations of
dissolved species may not decrease
along the direction groundwater flow.
In the northern part of the aquifer there
is a clear trend of salt content from east
to west along the line Lalbare`de (well
18) Blagnac (well 17)Beaucaire
(well 10). The main flow directions are
in agreement with the main gradient
directions. From Beaucaire and in the
western part, the flow pattern is not
clear, both head and salinity showing a
strong decrease in the west of
Beaucaire. The concentration decrease
might be explained by flow of dilute

may indicate a lower permeability


and
a limited
flow from east to west at this place.

We presently do not have enough


data to confirm or refute this
hypothesis.

In the southern area, the


potentiometric map tends to show a
major flow direction from south-east
to north-west. Most of the waters are
very dilute and mainly of bicarbonate
type. In the eastern area some trace
elements such as fluoride, strontium
or lithium are more concentrated. On
the other hand isotopic studies
showed that in the western part a
significant portion of the sulphates
arise from pyrite dissolution. Water
composition shows large bands of

similar composition oriented in the


southnorth direction. Modelling
results give several constraints on the
flow directions; it is not possible to
obtain Nogaro 2 water (well 7) from
evolution of the water at Demu (well
9). For this reason, we propose that a
geochemical watershed limit exists
between the two regions and a flow
towards the north occurs in this area
(Fig. 10). In the western part, the
evolution of

34

S isotopes

58

L. Andre et al. / Journal of Hydrology 305 (2005) 4062

layer may be the origin of the slow


flow and thus older water.
suggests a trend of pyrite dissolution
oriented in the same southnorth
direction. All geochemical arguments thus converge to indicate a
south-to-north flow direction.

6.2. Local flow patterns

6.2.1. Closed structures

6.2.2. Open structures

In contrast to the previous


structures, the waters sampled
around open structures show only
slight variations in the groundwater
composition. At several wells the
waters seem to be younger, owing to
14

Around several structures two


properties are observed: (i) the
estimated age of the waters are older
than in the surrounding aquifer, and
(ii) the groundwaters are more
mineralized and marked by ion
exchange. These phenomena occur
around the Garlin and Eugenieles-Bains structures (Fig. 2). The
waters sampled close to the Garlin
dome are among the oldest water
with an estimated age of 35,000
years while the waters from the very
close Lespielle borehole (well 19)
are 16,000 years old. Furthermore,
the modelling step showed that half
of the sodium comes from ion
exchange at this location. The same
phenomenon occurs at well 13
where waters are considerably older
(28,000 year) than the nearby well
of Geaune (well 14) (19,000 year).
At both sites the importance of ion
exchange can be related to the
presence of thick layers of
interbeded clays within the aquifer.
The presence of this non-conductive

their C estimated age. It appears


that these younger waters may arise
from mixing with superficial
aquifers at places where waters from
the Eocene sands aquifer discharge.
This process is found at Barbotan
(Wells 1 and 2) and CasteraVerduzan (Well 11). At Barbotan,
another argument may confirm this
process. At that place we showed
that all sulphates come from pyrite
oxidation. The use of the oxygen
isotopes of the sulphate molecule
also revealed that this oxidation may
partly come from oxygen (Andre et
al., 2002). The presence of oxygen
in the pyrite oxidation process
occurring only at that location might
be explained by mixing with oxygen
rich superficial waters.

6.2.3. More complex structures

The model developed suggests that the


waters from the southern part may
discharge to the north. This was

suggested particularly at Demu (well


9) where the geochemical arguments
preclude flow to the west (i.e. to well
7). The Celt-aquitaine fault zone is not
continuous and there might be a
potential outflow between faults (Fig.
11). It is quite interesting to notice that
the waters sampled at well 8, in the
northern area, could be obtained by
mixing equal proportions of well 9
and well 12 waters. It seems thus
plausible to have global outflow to the
north at this place resulting in closer
flowlines. This would also explain the
lower salt content of well 8 waters
compared to points located at the east
(wells 4 and 10) and the surprisingly
steep head gradients in this area.

6.3. Leakage

Leakage is used with increasing


frequency in regional modelling of
multi-layer aquifers (Toth, 1999).
The presence of such vertical fluxes
is fairly important for the
management of aquifer pumping at
the regional scale.

Eocene sands contain very few


minerals and groundwaters are very
dilute, in the southern part. In this

context, the presence of calcium


carbonate waters is expected but the
presence of sulphate is difficult to
explain except if mass transfers from
aquitards are considered.

Several processes can explain this


vertical sulphate transfer, the major
ones being leakage and diffusion.
Leakage consists in a vertical flow
through an aquitard, in this case from
the molasses, to the Eocene sands. As
interstitial molasse waters may be
saturated with respect to gypsum, i.e.
100 times more concentrated than the
aquifer waters, a small vertical water
flux can modify the composition of
the groundwater in the Eocene sands.
On the other hand, sulphate can
simply move by diffusion from a
concentrated medium (the molasse) to
a dilute one (the Eocene sands).

A detailed theory of the different


possibilities of vertical exchanges
between aquifers and aquitard has
been developed elsewhere (Atteia et
al., 2005). The authors showed that
(i) diffusion might explain low
concentrations of some ions and a
slight increase along flow of ions
present in aquitard pore waters and
(ii) leakage could generate fairly
important

L. Andre et al. / Journal of Hydrology 305 (2005) 4062


59

Fig. 11. Potential flux through the CeltAquitaine flexure.

concentrations of the same ions


where the difference between the
land surface and the potentiometric
surface of the Eocene sands aquifer
is quite sufficient to generate
leakage (according to Darcys law).
For our case, the diffusion of
sulphate from the molasse gypseous
layers could explain the observed
concen-trations and sulphur isotopic
content of the aquifer waters in the
southern zone. At contrary, the high
concentrations found in the north
could only be justified by leakage.

This difference between the south


and north of the basin can be
explained by geological conditions.
First, in the south, the molasse is
much thicker than in the north, and

second, the molasse lithology might


be coarser in the north than in the
south. The differences in molasse
lithologies could be explained by the
sedimentation history in both
geological compart-ments (Rey,
1995).

7. Conclusion

The results of this study show that


detailed hydrochemical data coupled
with geochemical mod-elling can
help to elucidate the hydrologic and
geologic factors controlling water
chemistry on a regional basin.

Waters from the Eocene sands


aquifer present mainly calcium
bicarbonate facies with an evolution
of the geochemical facies at several
places. To understand these
variations, detailed chemical and
isotopic data were used; changes in
water chemistry were interpreted
and three distinct hydrochemical

processes have been identified to be


responsible of these evolutions: (1)
Waters equilibrium with calcite from
recharge areas to discharge places;
(2) Cation exchange between waters
and the aquifer material; and (3)
Pyrite oxidation and gypsum
dissolution that

60

L. Andre et al. / Journal of Hydrology 305 (2005) 4062

can explain the regular increase of


sulphur concen-tration along
pathways.

These processes identified, a


modelling approach allowed us to
quantify those. This step by step
modelling used a limited number of
mineral phases that were identified
in the solid and successfully
reproduces the concentration of
major ions at each sampling points.
Numerous constraints that were
imposed on the model and its
13

validation by pH or C allowed us
to justify and quantify the
geochemical processes occurring in
solution.

Afterwards, based on the local


reconstitution of chemical
composition of waters, geochemical
model-ling was used to delineate
geochemical pathways. The origin
of local chemical composition
variations has been identified and
the role played by deep geological
structure has been outlined. It
appears that a concept of open and
closed structures seems to be
applicable in this region. This
concept would enlarge the classical
discussion of faults as barriers or
conducts for flow.

The study also shows that the


chemical compo-sition of waters

cannot be explain without evoking the


influence of aquitards. Considering the
poor miner-alogy of the sands, the
increase of total dissolved solids along
pathways seems to be associated to
mass transfers from the gypseous
molasses, present at the top of the
aquifer. This transfer of ions occurs
according two processes: leakage,
principally in the northern part of the
aquifer, and diffusion, mainly in the
south of the study area.

In the context of the Aquitaine Basin,


the geochemical modelling has proven
its ability to constrain efficiently the
groundwater flow in an aquifer were
few data are available. However, the
developed approach shows that a
detailed analysis of each reaction, the
use of validation variables and
information on mineral phases present
in the solid are required. At contrary,
the concentration increase for one
dissolved species can be attributed to
the wrong reaction. Along the paper,
we also showed that a geochemical
model was able to provide four types
of hydrodynamic information:
estimates of regional flow directions,
existing flow barriers, mixing between
aquifer waters and interactions
between aquifer and aquitard.
However, this information remains
qualitat-ive, the only solution to reach
more quantitative flow patterns is the
use of 3D multilayer models

including hydrodynamics, transport


and chemistry. The chemistry may
constrain flow but the opposite is

also true: the flow pattern governs


mixing and residence time.
Back W., 1966. Hydrochemical facies and
groundwater flow patterns in northern part of
Atlantic Coastal Plain. US Geological Survey
Professional Paper 498-A, p. 42.

Acknowledgements

The authors would like to thank


TotalFinaElf Stockage Gaz France
for supporting this work.

References

Blavoux, B., Dray, M., Fehri, A., Olive, P.,


Groning, M., et al., 1993. Paleoclimatic and
hydrodynamic approach to the Aquitaine
Basin deep aquifer (France) by means of
environmental isotopes and noble gases. Int.
Symp. App. Isotope Techn. 1993, 293305.

Boule`gue, J., Popoff, G., 1979. Nouvelles


methodes de determi-nation des principales
espe`ces ioniques du soufre dans les eaux
naturelles. J. Fr. Hydrol. 10 (2) 8390 n 29.

Andre L., 2002. Contribution de la ge


ochimie a` la connaissance des ecoulements
souterrains profonds. Application a`
laquife`re des SablesInfra-Molassiques du
BassinAquitain,The`se:Bordeaux3. p. 230.

Busby J.F., Plummer L.N., Lee R.W.,


Hanshaw B.B., 1991. Geochemical evolution
of water in the Madison aquifer in parts of
Montana, South Dakota, and Wyoming. US
Geological Survey Professional Paper 1273-F,
p. 89.

Andre, L., Franceschi, M., Pouchan, P.,


Atteia, O., 2002. Origines et evolution du
soufre au sein de laquife`re des Sables InfraMolassiques du Bassin Aquitain. C.R. Ge
osciences 334, 749756.

Cazal, A., Moussie, B., Feneyrou, G., 1967.


Etude de la circulation dune eau minerale a`
travers un aquife`re complexe, Chronique
dhydrogeologie. BRGM 12, 91112.

Appelo C.A.J., Postma D., 1999.


Geochemistry, Groundwater and Pollution,
A.A. Balkema/Rotterdam/Brookfield/1999, p.
536.

Chapelle, F.H., Knobel, L.L., 1983. Aqueous


geochemistry and the exchangeable cation
composition of glauconite in the Aquia
aquifer, Maryland. Ground Water 21 (3), 343
352.

Atteia O., Andre L., Franceschi M., Dupuy


A., 2005. Contributions of diffusion,
dissolution, ion exchange and leakage from
low permeability layers to confined aquifers
(Accepted by Water Resour. Res.).

Chiodini, G., Frondini, F., Kerrick, D.M.,


Rogie, J., Parello, F., 1999. Quantification of
deep CO2 fluxes from Central Itaky.
Examples of carbon balance for regional

aquifers and of soil diffuse degassing. Chem.


Geol. 159, 205222.

L. Andre et al. / Journal of Hydrology 305 (2005) 4062


61

Clark, I., Fritz, P., 1997. Environmental


Isotopes in Hydrogeology. Lewis
Publishers, New York, p. 328.

Claypool, G.E., Holser, W.T., Kaplan, I.R.,


Zak, I., 1980. The age curves of sulphur and
oxygen isotopes in marine sulphate and
their mutual interpretation. Chem. Geol. 28,
199260.

Franceschi M., 2005. Radon 222 activities


in a deep sandy aquifer and geological
factor controlling its occurrence (Accepted
by Ground Water).

Freeze, R.A., Cherry, J.A., 1979.


Groundwater. Englewood Cliffs, PrenticeHall, NJ, p. 604.

Garrels, R.M., Christ, C.L., 1965.


Solutions, Minerals, and Equilibria. Harper
and Row, New York. Appendix 2, p. 450.
Dazy, J., Rochat, J., Olive, Ph., 1980.
Nouvelles donnees geochimiques et
isotopiques sur les eaux thermales
e`me

dUriage-les-Bains (Ise`re), 105


Congre`s National des Societes Savantes,
Caen. Sciences Fasc. II, 111122.

Dubreuilh, J. 1987. Synthe`se paleoge


ographique et structurale des depots
fluviatiles tertiaires du nord du Bassin
Aquitain. Passage aux formations palustres,
lacustres et marines, The`se de Doctorat de
tat, Bordeaux III.

Gerla, P.J., 1992. Pathline and geochemical


evolution of ground-water in a regional
discharge area, Red River Valley, North
Dakota. Ground Water 30, 743754.

Gosselin, D.C., Harvey, F.E., Frost, C.D.,


2001. Geochemical evolution of
groundwater in the great plains (Dakota)
aquifer of Nebraska: implications for the
management of a regional aquifer system.
Ground Water 39 (1), 98108.

Eberts S.M., George L.L., 2000. Regional


groundwater flow and geochemistry in the
Midwestern basins and Arches aquifer
system in parts of Indiana, Ohio, Michigan,
and IL. US Geological Survey Professional
Paper 1423-C, p. 103.

Guler, C., Thyne, G.D., 2004. Hydrologic


and geologic factors controlling surface and
groundwater chemistry in Indian WellsOwens Valley area, southeastern California,
USA. J. Hydrol. 285, 177198.

Edmunds, W.M., Carrillo-Rivera, J.J.,


Cardona, A., 2000. Geo-chemical evolution
of groundwater beneath Mexico City. J.
Hydrol. 258, 124.

Hendry, M.J., Schwartz, F.W., 1990. The


chemical evolution of groundwater in the
Milk River Aquifer, Canada. Ground Water
28, 253261.

Holmes, P.R., Crundwell, F.K., 2000. The


kinetics of the oxidation of pyrite by ferric
ions and dissolved oxygen: an
electrochemical study. Geochim.
Cosmochim. Acta 64 (2), 263274.

Institut EGID, 1999, Actualisation des


cartes piezometriques regionales du
bassin hydrogeologique de lAdour, 1998,
p. 12, 5 pl.

Labat N., 1998, Role de particularites se


dimentaires et structurales sur le
comportement des sables sous-molassiques
soumis aux fluctuations induites par les
stockages souterrains de gaz. Application a` le
tude de leur influence sur lhydrodynamisme
des emergences locales, The`se: Bordeaux 3.
p. 228.

Marini, M., Ottonello, G., Canepa, M.,


Cipolli, F., 2000. Water-Rock interaction in
the Bisagno valley (Genoa,

Italy): application of an inverse approach to


model spring water chemistry. Geochim.
Cosmochim. Acta. 64 (15), 26172635.

Martinez, D.E., Bocanegra, E.M., 2002.


Hydrogeochemistry and cation-exchange
processes in the coastal aquifer of Mar del
Plata, Argentina. Hydrogeol. J. 10, 393
408.

Murray, K.S., 1996. Hydrology and


geochemistry of thermal waters in the
Upper Napa Valley, California. Ground
Water 34 (6), 11151124.

Nordstrom, D.K., Ball, J.W., 1989. Mineral


saturation states in natural waters and their

sensitivity to thermodynamic and analytic


errors. Sci. Geol. Bull. 42 (4), 269280.

Ohayon-Courte`s C., 1992. Levolution


des espe`ces reduites du soufre dans les
eaux minerales, The`se: Bordeaux II, p.
319.

Parkhurst D.L., Appelo C.A.J., 1999. A


computer program for speciation, batchreaction, one dimensional transport and
inverse geochemical calculations. US Geol.
Surv. Water Resour. Invest. Rep. 99-4259,
p. 312.

Pearson, F.J., Rightmire, C.T., 1980.


Sulphur and oxygen isotopes in aqueous
sulphur compounds, in: Fritz, P., Fontes, J.
Ch. (Eds.), Handbook of Environmental
Isotopes. Geochemistry. Elsevier,
Amsterdam, pp. 227258. Chapter 6.

Pelissier-Hermitte, G., Franceschi, M., Che


ron, J., Dupuy, A. 2000. Contamination
azotee dune nappe captive par drainance
dune nappe libre-Mecanismes et e
volution. Colloque international ESRA1315 septembre. S2-11-S2-14

Plummer, L.N., Sprinckle, C.L., 2001.


Radiocarbon dating of dissolved inorganic
carbon in groundwater from confined parts
of the Upper Floridan Aquifer, Florida,
USA. Hydrogeol. J. 9, 127150.

Rechiniac A., 1962. Etude se


dimentologique des principales formations
detritiques du Paleoge`ne aquitain,
The`se: Bordeaux, p. 69.

Rey, J. 1995. Le Bassin dAquitaine:


composition, evolution et structure. In: Du
Lias Nord aquitain aux molasses mioce`nes
Generalites Livret Guide dexcursion,
STRATA, Actes du laboratoire de Geologie
Sedimentaire et Paleontologie de
lUniversite Paul SabatierToulouse, Se
rie. 2 Memoire, pp. 7140

Rosen, M., Jones, S., 1998. Controls of the


chemical composition of groundwater from
alluvial aquifers in the Wanaka and
Wakatipu basins, Central Otago, New
Zealand. Hydrogeol. J. 6, 264281.

en (Bassin dAquitaine): sedimento-logie,


stratigraphie et evolution geodynamique,
The`se: Rennes.

Sracek, O., Hirata, R., 2002. Geochemical


and stable isotopic evolution of the Guarani
Aquifer System in the state of Sao Paulo,
Brazil. Hydrogeol. J. 10 (6), 643655.

Thorstenton, D.C., Fisher, D.W., Croft,


M.G., 1979. The geochem-istry of the Fox
Hills-Basal Hell Creek aquifer in
southwestern North Dakota and
northwestern South Dakota. Water Resour.
Res. 15 (6), 14791498.

Schoeller, H., 1956. Geochimie des eaux


souterraines. Masson, Paris, p. 581.

Serrano, O. 2001. Le Cretace Superieur/Pale


oge`ne du bassin davant-pays Nord-Pyrene

Toth, J., 1999. Groundwater as a geologic


agent: an overview of the causes, processes
and manifestations. Hydrogeol. J. 7, 114.

62

L. Andre et al. / Journal of Hydrology 305 (2005) 4062

Fredrickson,
Trembaczowski, A., Halas, S. 1984. Oxygen
and sulphur isotope ratios in sulphate from
atmospheric precipitations. In: Proc. Int.
Symp. Isotope Hydrol. Water Resour. Dev.,
I.A.E.A., Vienna.

J.K.,
McKinley,
J.P.,
Swenson, J.B.,
1999. Groundwater flow

819820
and geochemistry

in the southeastern San Juan Basin:


implications for microbial transport and
activity. Water Resour. Res. 35 (5), 1409
1424.

Walvoord,
M.A.,
Pegram, P., Phillips, F.M., Person, M.,
Kieft,
T.L.,

Weaver, T.R., Bahr, J.M., 1991.


Geochemical evolution in the CambrianOrdovician Sandstome aquifer, Eastern
Wisconsin. 2. Correlation between flow
paths and groundwater chemistry. Ground
Water 29, 510515.

Você também pode gostar