Hydrochemistry of a complex volcano-sedimentary aquifer using major ions and environmental isotopes data Dalha basalts aquifer, southwest of Republic of Djibouti


Aboubaker, M.; Jalludin, M.; Razack, M.

Environmental Earth Sciences 70(7): 3335-3349

2013


In the Republic of Djibouti (Horn of Africa), fractured volcanic aquifers are the main water resources. The country undergoes an arid climate. Alluvial aquifers exist in the wadis (intermittent streams) valleys and, in relation with volcanic aquifers, form complex volcano-sedimentary systems. Due to increasing water demands, groundwater resources are overexploited and require a rigorous management. This paper is focused on the Dalha basalts aquifer, located in the Dikhil area (Southwest of Djibouti). This aquifer is of vital importance for this area. Hydrochemical data and isotopic tracers (<sup>18</sup>O and <sup>2</sup>H) were used to identify factors and phenomena governing the groundwater s mineralization. The Piper diagram shows complex water types. from multivariate statistical analyses highlight three water families according to their locations: (1) groundwater characterized by low ionic concentrations located at the wadis zones; groundwater characterized by moderate salinity and highly mineralized waters mainly flowing in the eastern and central part of the study area, in volcanic aquifers. from scatter plots, especially Na versus Cl and Br versus Cl, suggest that the origin of more saline waters is not from dissolution of halite. The ?<sup>18</sup>O and ?<sup>2</sup>H data indicate that the groundwater flowing in the alluvial aquifer is of meteoric origin and fast percolation of rainwater occurs in the volcanic aquifers. These findings provide a preliminary understanding of the overall functioning of this complex volcano-sedimentary system. Additional investigations (pumping tests, numerical modeling) are in progress to achieve a more comprehensive understanding of this system.

Environ
Earth
Sci
DOI
10.1007/s12665-013-2398-8
ORIGINAL
ARTICLE
Hydrochemistry
of
a
complex
volcano-sedimentary
aquifer
using
major
ions
and
environmental
isotopes
data:
Dalha
basalts
aquifer,
southwest
of
Republic
of
Djibouti
Mohamed
Aboubaker
Mohamed
Jalludin
Moumtaz
Razack
Received:
19
June
2012/Accepted:
6
March
2013
©
Springer-Verlag
Berlin
Heidelberg
2013
Abstract
In
the
Republic
of
Djibouti
(Horn
of
Africa),
fractured
volcanic
aquifers
are
the
main
water
resources.
The
country
undergoes
an
arid
climate.
Alluvial
aquifers
exist
in
the
wadis
(intermittent
streams)
valleys
and,
in
relation
with
volcanic
aquifers,
form
complex
volcano-
sedimentary
systems.
Due
to
increasing
water
demands,
groundwater
resources
are
overexploited
and
require
a
rigorous
management.
This
paper
is
focused
on
the
Dalha
basalts
aquifer,
located
in
the
Dikhil
area
(Southwest
of
Djibouti).
This
aquifer
is
of
vital
importance
for
this
area.
Hydrochemical
data
and
isotopic
tracers
(
18
0
and
2
H)
were
used
to
identify
factors
and
phenomena
governing
the
groundwater's
mineralization.
The
Piper
diagram
shows
complex
water
types.
Results
from
multivariate
statistical
analyses
highlight
three
water
families
according
to
their
locations:
(1)
groundwater
characterized
by
low
ionic
concentrations
located
at
the
wadis
zones;
(2)
groundwater
characterized
by
moderate
salinity
and
(3)
highly mineral-
ized
waters
mainly
flowing
in
the
eastern
and
central
part
of
the
study
area,
in
volcanic
aquifers.
Results
from
scatter
plots,
especially
Na
versus
Cl
and
Br
versus
Cl,
suggest
that
the
origin
of
more
saline
waters
is
not
from
dissolution
of
halite.
The
8
18
0
and
82H
data
indicate
that
the
groundwater
flowing
in
the
alluvial
aquifer
is
of
meteoric
origin
and
fast
percolation
of
rainwater
occurs
in
the
volcanic
aquifers.
These
findings
provide
a
preliminary
understanding
of
the
overall
functioning
of
this
complex
volcano-sedimentary
M.
Aboubaker
M.
Razack
(El)
Department
of
Hydrogeology
UMR
7285,
University
of
Poitiers,
Rue
Albert
Turpain
Bat.B8,
86022
Poitiers
Cedex,
France
e-mail:
moumtaz.razack@univ-poitiers.fr
M.
Jalludin
CERD,
PB
486
Djibouti,
Republic
of
Djibouti
system.
Additional
investigations
(pumping
tests,
numerical
modeling)
are
in
progress
to
achieve
a
more
comprehensive
understanding
of
this
system.
Keywords
Dikhil
Dalha
basalts
Hydrochemistry
Stable
isotopes
Volcanic
aquifer
Hierarchical
cluster
analysis
Principal
components
analysis
Introduction
The
Republic
of
Djibouti,
located
in
the
Horn
of
Africa,
undergoes
a
very
arid
climate.
The
country
is
particularly
marked
by
very
low
precipitation
(less
than
200
mm
of
precipitation
per
year)
which
results
in
the
absence
of
perennial
streams.
The
country
is
located
in
a
particular
geodynamic
context,
related
to
the
separation
of
African
and
Arabian
plates
since
about
thirty
million
years.
Therefore,
the
volcanic
formations
resulting
from
plate
tectonics
outcrop
over
major
part
of
the
territory
(80
%).
Consequently,
the
only
available
water
resources
are
rep-
resented
by
the
groundwater
in
the
volcanic
aquifers.
Given
the
increasing
water
demands
and
the
implications
gener-
ated
by
climate
change,
these
groundwater
resources
are
very
fragile.
Understanding
the
functioning
of
these
vol-
canic
aquifer
systems
has
become
a
vital
necessity
for
the
sustainability
of
these
water
resources.
This
paper
focuses
on
the
Dalha
fractured
basaltic
aquifer
in
the
Dikhil
basin,
Southwest
of
Djibouti
(Fig.
1).
Groundwater
resources
in
this
basaltic
aquifer
show
a
qualitative
and
quantitative
deterioration
developing
in
time
due
both
to
natural
constraints
(semi-arid
climate
with
low
amount
of
rainfall)
and
to
anthropic
activities
(over-
exploitation).
Hydrodynamic,
hydrochemical
and
isotopic
methods
were
used
in
this
study.
Published
online:
04
April
2013
4Springer
Water
samples
Stratiform
basalts
El
Fluvial
deposits
Dalha
basalts
Clayey
silts
Wadis
Mabla
rhyolites
Hambocto
Dbol2
..Dbol1
Awr1
rnbk
Galm2
Grand
Bara
M9
Awr5
Ab'ay
tou
Awr3
Mouloud
M5
Awr2
Dad1
Dade
MDk
Dikhil
DjG
Dadin
Dik9
Odin+
D
idahalou
Dik14
Dab2
N
de
Ethiopia
A
9000
4500
0
9000
meters
Di
k6
Kont1
a
Chk2
AbY
t.
ErytEMa
Ethiopia
A°0
AV
eV
4
0
Djibouti
Somalia
42°25'0"E
42°35'0"E
11°10'0"N
11°0'0"N
V
600
m
-
71-
.
500
m
V
V
V
V
V
V
V
V
V
V
V
1
Dad1
Dad3
Dad4
2
1
-
1
3
Mouloud
Yar
Dad6
SW
NE
B
7
-
77E-74
?
V
'
1
4
?
Environ
Earth
Sci
altitude
m
(as1)
Fig.
1
a
Simplified
geological
map
of
the
study
area
with
sampling
points.
b
NE-SW
structural
cross-section
at
Mouloud
Yar
basin
in
the
study
area
(in
Jalludin
1993).
1
alluvium,
2
basalt,
3
clay
4h
Springer
Environ
Earth
Sci
Precipitations
are
very
low
(average
of
150
mm/year)
and
irregular
in
time.
Recharge
occurs
during
rainfall
events
of
short
duration
and
high
intensity
through
the
wadis
valleys
which
are
preferential
recharge
areas.
In
Mouloud
area
for
instance
(Fig.
1),
overexploitation
of
the
aquifer
during
1972-2002
period
resulted
in
a
deterioration
of
water
quality
with
an
increase
in
salinity.
In
parallel,
the
groundwater
head,
decreasing
continuously,
lost
about
30
m
in
30
years
in
the
same
area
(Jalludin
1989).
There-
fore,
the
aim
of
the
present
study
is
to
assess
groundwater
dynamics
and
to
determine
the
origin
and
the
mechanisms
governing
its
salinization.
Conventional
methods
such
as
Piper
diagram
(Piper
1944)
and
scatter
plots
of
chemical
parameters,
coupled
with
multivariate
methods
(hierarchi-
cal
cluster
analysis
HCA
and
principal
components
anal-
ysis
PCA),
are
used
to
interpret
geochemical
data
(Davis
1986).
Location
of
the
study
area
The
study
area
is
located
in
the
southwestern
part
of
Djibouti
(Fig.
1).
It
lies
between
11°0
/
0"N
to
11°15'0"N
latitude
and
42°20'0"E
to
42°41'0"E
longitude,
with
an
altitude
between
400
and
800
m
above
sea
level
and
covers
an
average
surface
of
1200
km
2
.
There
are
many
inter-
mittent
streams
called
wadis
(Awrawsa,
Dadin,
Mouloud,
Batoul,
Ab'aytou,
Cheykeyti)
crossing
the
study
area.
It
is
bordered
to
the
North—East
by
Hambocto
wadi,
to
the
South
by
Dabadere
wadi,
to
the
East
by
Mabla
rhyolites,
to
the
North
by
stratiform
basalts
series
and
Grand
Bara
depression,
and
to
the
West
by
the
sedimentary
formations
of
Plio-Pleistocene
to
Holocene.
Geological
and
hydrogeological
settings
The
geological
formations
are
mainly
represented
by
Dalha
basalts
series
(9-3.4
Ma),
and
stratiform
basalts
series
(3.4-1
Ma)
of
the
Afar.
The
stratiform
basalts
series
occupy
two-thirds
of
the
Republic
of
Djibouti
and
uncon-
formably
overlie
the
Dalha
basalts
series
and
the
sedi-
mentary
formations.
The
series
consists
of
a
regular
pile
of
basalt
flows
of
some
meters
to
more
than
10
m
in
thick-
ness,
with
intercalations
of
acid
lavas,
ignimbrite
mainly,
but
also
of
pumice
and
rhyolite
flows
inserted
with
sedi-
mentary
lake
formations
and
detrital
limestones
(Demange
and
Stieljes
1975).
Between
the
traps
are
inserted
scorias
and
paleosol
which
can
exceed
a
few
meters,
marking
the
cessation
of
volcanic
activity.
The
basaltic
flows
are,
in
their
lower
part,
often
altered
and
sometimes
rich
in
phenocrysts
of
pyroxene
and
olivine
(BGR
1982).
Indeed,
the
chemical
analyses
carried
out
on
samples
of
Dalha
basalts
(Gasse
et
al.
1986)
revealed
an
aphyric
(sometimes
porphyric)
texture
with
plagioclase,
pyroxene
and
olivine.
And
according
to
the
rocks
analysis,
the
proportion
of
feldspar
(alkali
and
plagioclase)
reaches
60
%
of
silicate
minerals
in
Dalha
basalts,
with
a
predominance
of
albite
(35
%)
and
anorthite
(20
%).
The
Dikhil
formation
is
a
thick
lacustrine
accumulation
consisting
of
clays
and
diatomite,
and
covered
at
the
top
by
gypsum
deposits.
Lacustrine
sediments
are
based
on
highly
weathered
basalts
through
a
reddish
sandstone
and
conglomerate
at
a
level
of
large
blocks
of
basalt
(Gasse
1975).
The
climate
is
arid
to
semi-arid
characterized
by
the
weakness
and
the
irregularity
of
the
precipitation,
high
temperature
and
intense
evaporation
(more
than
2,000
mm/
year).
Overall,
two
seasons
predominate:
a
cool
season
from
October
to
April
and
a
hot
season
from
May
to
September.
The
average
temperature
is
34
°C
and
mean
annual
rainfall
is
150
mm.
The
characteristics
of
such
climate
allow
only
limited
infiltration
through
the
wadis
valleys
during
flood
period
(Jalludin
1993).
The
Dalha
basalts
underwent
intense
tectonics
and
acquired
thereby
a
certain
fracture
permeability
(Jalludin
1993).
Furthermore,
the
basaltic
formations
are
often
ver-
tically
displaced
by
tectonic
accidents
presently
masked
by
the
sedimentary
deposits.
For
instance
at
the
Mouloud
catchment,
located
between
Dalha
basalts
and
the
plain
of
Grand
Bara
(Fig.
la),
the
groundwater
in
basalts
(strati-
form
basalts)
was
revealed
under
the
sedimentary
forma-
tions.
Another
example
is
found
at
Dadin
catchment,
where
two
aquifers
were
identified
(Fig.
lb).
The
Dalha
basalts
aquifer
is
an
unconfined
aquifer
which
characterizes
the
whole
outcrop
of
these
basalts.
According
to
the
Mouloud
catchment
geology,
the
basaltic
aquifer
continues
under
the
sedimentary
cover
of
the
Grand
Bara
plain.
Another
local
groundwater
was
recognized
in
the
wadis
valleys
alluvia.
In
these
local
alluvial
aquifers,
occurs
a
groundwater
underflow
directly
related
with
the
wadis
runoff
during
floods.
Thus,
the
basaltic
aquifer
recharge
would
be
provided
by
these
alluvial
groundwaters
that
may
play
a
role
of
transfer.
This
recharge
process
could
take
place
through
major
faults.
According
to
pumping
tests,
the
transmissivity
of
strat-
iform
basalts
aquifers
is
higher
than
that
of
the
older
Dalha
basalts.
Indeed,
the
transmissivity
values
of
stratiform
basalts
range
between
0.5
and
1,130
m
2
/h.
The
transmis-
sivity
values
of
the
Dalha
basalts
are
lower
than
those
of
stratiform
basalts,
and
range
between
0.5
and
36
m
2
/h.
The
transmissivity
of
Quaternary
sedimentary
rocks
aquifers
ranges
between
0.4
and
163
m
2
/h
(Jalludin
and
Razack
2004).
The
piezometric
map
of
the
Dalha
basalts
aquifer,
established
during
dry
season
(May
2010)
using
data
from
1Springer
Hmbk
Dbol2
•s
Isoline
Wadis
Flow
Dad3
Debacle,.
Dikhil
Dik8
-.413it14
Dab2
.
Dik9
Dik6
Dsihw
.
Ala
1240000
-
1235000
1230000
-
1225000
1220000
Environ
Earth
Sci
Fig.
2
Piezometric
map
of
Dalha
basalt aquifer
(main
aquifer
in
the
study
area)
1215000
1
1
200000
205000
210000
215000
220000
225000
230000
235000 240000 245000
drilled
wells
and
hand-dug
wells
unevenly
distributed
across
the
study
area,
shows
two
main
flow
directions
oriented
S—NW
and
S—NE,
respectively
(Fig.
2),
reflecting
a
significant
permeability
as
demonstrated
by
the
above
mentioned
pumping
tests.
High
hydraulic
gradients
observed
in
Dadin
and
Dikhil
areas
are
due
to
intense
exploitation
of
groundwater
in
these
locations.
As
the
piezometric
map
is
related
to
the
dry
season,
the
relation-
ships
between
the
wadis
(preferential recharge
areas)
and
the
groundwater
could
not
be
pointed
out.
Approach
and
methodology
The
hydrogeochemical
characterization
of
the
volcano-
sedimentary
aquifer
system
in
the
study
area
involved
groundwater
sampling
at
14
sites
distributed
over
the
study
area
and
groundwater
analysis
for
12
parameters
including
in
situ
measurements.
The
groundwater
samples
were
collected
during
the
sampling
campaign
in
October
2010.
Depth
of
the
drilled
wells
ranges
between
103
m
and
141
m,
and
of
the
hand-
dug
wells
between
3
m
and
11
m.
All
samples
were
filtered
using
a
membrane
filter
of
0.22
micron
diameter.
All
major
ions
were
analyzed
at
the
University
of
Poitiers
(France).
Major
cations
and
silica
were
analyzed
by
atomic
absorp-
tion
spectrometry
using
flame
of
a
double
beam
spec-
trometer
VARIAN
AA240FS.
The
major
anions
were
determined
by
ion
chromatography
using
a
chromatograph
DIONEX
ICS-1000.
Bicarbonates
were
analysed
by
the
titration
method.
Stable
isotopes
(8
18
0
and
8
2
H)
determi-
nations
were
undertaken
with
the
use
of
the
mass
spec-
trometer
at
the
University
of
Addis
Ababa.
All
stable
isotopic
compositions
(
18
0
and
2
H)
are
reported
in
standard
8
notation
as
follows:
=
[(Rsample/Rstandard)
1]
X
1,
000
where
R
sample
and
Rstandard
represent
the
ratio
of
heavy
to
light
isotopes
of
the
samples
and
standard,
respectively
(Craig
1961).
Measurements
of
electrical
conductivity
(EC),
pH
and
temperature
(°C)
were
made
in
the
field.
The
values
of
total
dissolved
solids
(TDS)
were
obtained
by
multiplying
EC
values
by
0.64
(Todd
1980).
The
accuracy
of
the
analyses
was
estimated
from
the
ionic
balance
error
(Freeze
and
Cheery
1979),
which
is
within
5
%
for
a
majority
of
samples.
Descriptive
and
multivariate
statistical
analysis
were
done
using
the
STATISTICA
data
analysis
software
version
7
(StatSoft
Inc.
2008)
on
12
hydrochemical
variables
(in
this
case
EC,
pH,
T
C,
Na
t
,
K
t
,
Ca
2t
,
Mg
2t
,
SiO
2
,
Cl
,
HCO3
NO
3
and
504).
A
factor
analysis
has
been
performed
on
the
data
sets
so
as
to
reduce
the
number
of
variables
into
more
important
variable
groups
called
factors
which
may
explain
the
underlying
hydrochemical
processes.
Results
and
discussion
Descriptive
analysis
Chemical
analysis
for
major
cations
and
anions
is
listed
in
Table
1
and
descriptive
statistics
for
major
hydrochemical
variables
was
also
calculated
(Table
2)
to
gain
information
on
the
ranges
of
variables
in
samples
in
the
study
area.
As
shown
in
Table
2,
hydrochemical
variables
show
large
variation
among
samples.
pH
is
moderately
alkaline
while
4t
1
Springer
Table
1
Hydrochemical
data
of
water
samples
in
the
study
area
Sample
ID
Date
of
sampling
Alt.
(m)
Depth
(m)
T
(°C)
pH
EC
(µS/cm)
TDS
(me)
Na
(me)
K
(me)
Ca
(me)
Mg
(me)
Cl
(me)
HCO
3
(me)
NO
3
(me)
SO
4
(me)
Ratios
(concentrations
in
meq/1)
Br/Cl
Na/
Cl
Ca/
SO
4
Ca/
Mg
SO
4
/
Cl
1
Awl
13.10.2010
579
195
33.1
7.60
3,160
1,868
325.5
12.7
91.0
59.8
554.3
156.2
195.5
266.4
0.0051
0.91
0.82 0.92
0.35
2
AbY
13.10.2010
402
10.78
35.7
7.37
820
503
95.7
1.7
36.4
9.3
20.4
252.3
10.0
112.7
0
7.21 0.77
2.36
4.08
3
M5
13.10.2010
578
115
39.1
7.08
3,730
2,400
361.8
8.4
92.8
92.84
622.5
255.3
85.0
344.4
0.0048
0.90
0.65
0.61
0.41
4
Dad
6
13.10.2010
626
129
34.7
7.34
1,567 1,027
227.5
5.4
29.1
25.0
170.0
403.9
34.5
107.1
0.0062
2.06
0.65
0.71
0.465
5
Dbd
2
13.10.2010
474
123
35.1
7.96
1,756
1,074
209.3
7.0
31.5
39.3
217.0
294.5
66.0
133.0
0.0062
1.49
0.57
0.49
0.45
6
Galm
2
13.10.2010
341
80
44
7.58
1,087
699
148.2
5.2
7.9
15.2
43.7
315.3
31.0
63.0
0
5.23
0.30
0.31
1.07
7
ChK2
13.10.2010
389
7.83
31.3
8.07
2,530
1,668
329.2
11.5
122.0
23.1
86.1
58.5
66.0
1,084.4
0
5.89
0.27
3.20
9.30
8
Kontl
13.10.2010
375
6.70
35
8.44
936
596
155.3
0.8
11.9
13.4
33.1
264.3
9.5
133.2
0
7.23
0.54 0.54
2.98
9
Ab'a
13.10.2010
372
2.68
38.5
7.56
1,800
1,087
228.5
4.5
27.9
35.3
247.9
253.7
49.0
125.2
0.0050
1.42
0.54
0.48
0.37
10
DjG
13.10.2010
683
11
34.2
7.06
1,552
960
146.6
9.1
59.1
34.1
184.8
292.8
76.5
102.6
0.0057
1.22
1.38
1.05
0.41
11
DB
oll
13.10.2010
566
93
33.1
7.54
2,750
1,720
322.2
8.5
57.2
59.1
510.2
158.6
64.0
236.3
0.0047
0.97
0.58
0.59
0.34
12
DBol
2
13.10.2010
561
90
38
7.24
5,320
3,410
522.7
11.5
119.0
109.5
830.7
128.7
63.5
348.5
0.0045
0.97
0.82
0.66
0.31
13
Hmbk
13.10.2010
615
120
33.9
7.52
820
525
102.3
4.1
57.0
42.5
151.9
202.4
129.0
108.8
0.0086
1.04
1.25
0.81
0.53
14
Dik
9
13.10.2010
510
141
38.7
7.30
1,130
656
111.8
2.3
33.9
26.4
89.4
280.7
34.5
65.6
0
1.93
1.24
0.78
0.54
Values
of
the
bi-elements
ratios
higher
than
1
are
indicated
in
bold
Environ
Earth
Sci
Table
2
Descriptive
statistics
of
the
hydrochemical
variables
of
groundwater
samples
Variables
Units
Mean
Median
Minimum
Maximum
Standard
deviation
WHO
guide
Cl
-
mg/1
268.7
177.4
20.4
830.7
255.3
250
HCO
3
-
mg/1
236.9
254.5
58.3
403.9
88.3
250
S0
4
2-
mg/1
230.8
129.1
63.0
1,084.4
263.5
250
NO
3
-
mg/1
65.2
63.7
9.5
195.5
48.8
50
Na
±
mg/1
234.8
218.4
95.7
522.6
123.4
150
IC
E
mg/1
6.6
6.2
0.7
12.7
3.8
12
Ca
mg/1
55.5
46.7
7.9
122.0
37.3
100
m
e-
mg/1
41.8
34.6
9.3
109.5
29.5
50
pH
7.6
7.5
7.2
8.4
0.3
6.5
<
pH
<
9.5
EC
µS/cm
2,068
1,661
820
5,320
1,304
2,500
Temp
C
36.2
35.4
31.4
43.6
3.0
temperature
and
electric
conductivity
vary
from
31
to
44
°C
and
820
to
5,320
µS/cm,
respectively.
The
total
dissolved
solids
(TDS)
range
from
503
(AbY)
to
3,410
mg/
1
(DBol2).
Indeed,
the
map
of
spatial
distribution
of
groundwater's
mineralization
in
the
study
area
(Fig.
3)
shows
that
the
electrical
conductivity
increases
globally
from
South
to
North,
and
this
increase
of
salinity
toward
discharge
areas
constituted
by
Grand
Bara
and
Checkeyti
basins
is
in
good
agreement
with
the
two
main
flow
directions
inferred
from
the
piezometric
map
previously
described.
Mean
anion
concentrations
occur
in
the
order
HCO
3
>
SO
4
>
Cl
>
NO
3
and
cation
in
the
order
Na
>
Mg
>
Ca
>
K.
Chloride
concentrations
show
wide
spatial
distri-
bution
in
the
area,
between
20.5
mg/1
(AbY
sample)
and
830.7
mg/1
(DBol2 sample)
with
standard
deviation
of
255.4
mg/l.
Physico-chemical
parameters
like
EC,
Cl,
in
some
samples
such
as
M5,
Awrl,
DBoll
and
DBol2
samples
located
in
the
northern
part
of
the
study
area,
are
higher
than
the
maximum
permissible
values
(Table
2)
prescribed
by
the
WHO
World
Health
Organization
(1993)
standards
set
for
drinking
water.
The
samples
presenting
TDS
<700
mg/1
are
wells
located
around
the
axis
of
Chekeyti
wadi
and
its
affluents
(AbY,
Kontl
and
Galm2
wells)
or
wells
located
near
Hambocto
wadi
(Hmbk
well)
which
are
preferential
recharge
areas.
Indeed,
as
discussed
previously
in
the
hydrogeology
setting
of
the
area,
the
basaltic
aquifer
recharge
is
performed
through
the
alluvial
aquifer
captured
par
these
wells
(AbY,
Kontl
wells,
Fig.
la).
Correlation
matrix
To
investigate
the
relationship
between
various
parameters
for
all
hydrochemical
samples
in
this
study,
a
statistical
correlation
matrix
of
the
twelve
variables
was
established
(Table
3).
The
correlation
matrix
allows
us
to
distinguish
several
relevant
hydrochemical
relationships
indicated
by
the
values
highlighted
in
bold.
As
shown
in
Table
3,
Na,
K
and
Cl
are
positively
cor-
related
showing
that
they
are
gradually
growing
toward
flow
direction.
These
variables
are
also
highly
correlated
with
electrical
conductivity
which
shows
that
the
increase
in
salinity
is
due
to
the
enrichment
of
these
variables.
Calcium
is
positively
correlated
with
electrical
con-
ductivity,
with
a
correlation
coefficient
of
0.79,
showing
that
the
calcium
concentration
is
one
of
the
factors
that
control
water
quality,
thus
emphasizing
the
calcium
as
an
indicator
of
the
intensity
of
calcic
minerals
alteration.
The
magnesium
contents
in
some
samples
(DBoll,
DBo12,
Awrl
and
M5)
are
high
(>50
mg/1,
Table
1)
and
globally
are
relatively
correlated
with
those
of
calcium
(R
=
0.68,
Table
3).
As
shown
in
Table
3,
HCO
3
is
fairly
negatively
correlated
with
sulfate
(-0.73)
and
calcium
(-0.78).
In
addition,
bicarbonates
are
weakly
correlated
with
Na
(-0.56),
K
(-0.58)
and
Mg
(-0.40).
Multivariate
statistical
analysis
Multivariate
statistical
analysis
is
a
quantitative
approach,
independent
of
the
classification
of
groundwater
allowing
aggregation
of
groundwater
samples
and
the
identification
of
correlations
between
chemical
parameters
and
ground-
water
samples.
In
this
study,
two
multivariate
statistical
analysis
methods
were
performed
to
classify
the
water
samples,
based
on
their
geochemical
characteristics,
according
to
the
methods
used
in
previous
studies
by
many
authors
(Razack
and
Dazy
1990;
Davis
1986;
Suk
and
Lee
1999).
The
following
methods
were
used:
Hierarchical
Cluster
Analysis
(HCA)
and
Principal
Components
Anal-
ysis
(PCA),
implemented
in
the
STATISTICA
software
version
7.0
(StatSoft
Inc.
2008).
1
Springer
Fig.
3
Electrical
conductivity
1245000
distribution
in
the
study
area
(units
in
µS/cm).
Irregular
straight
line
represents
Dalha
basalt
aquifer
1240000
Kont
1
1225000
Dab
2
ETHIOPIA
1220000
1215000
N
0
5000
10000
15000
20000m
Im
2
DBoI
2
Grand
Bara
Aw..1
Ab'a
M5
D
••
1235000
Chk2
D
k
Dad
6
Dk
DJG
1230000
Dik
9
Environ
Earth
Sci
Legend
3000
2500
2000
1500
1000
i i i i
200000 205000
210000 215000 220000 225000 230000 235000 240000 245000
Table
3
Correlation
matrix
of
the
variables
for
all
samples
of
the
study
area
Variables
Cl
-
HCO
3
-
S0
4
2-
NO3
Na
±
IC
E
Ca
m
g
2-1-
Si0
2
pH
EC
Temp
Cl
-
1
HCO
3
-
-0.41
1
S0
4
2-
0.14
-0.73
1
NO
3
0.48
-0.42
0.16
1
Na
±
0.87
-0.56
0.53
0.3
1
K
±
0.68
-0.58
0.58
0.64
0.77
1
Ca
0.65
-0.78
0.75
0.52
0.77
0.82
1
m
e-
0.97
-0.4
0.16
0.45
0.82
0.62
0.68
1
Si0
2
0.43 0.23
-0.24
0.53
0.2
0.4
0.17
0.44
1
pH
-0.35
-0.13
0.31
-0.13
-0.13
-0.15
-0.14
-0.33
-0.39
1
EC
0.93
-0.55
0.43
0.36
0.97
0.77
0.79
0.91
0.25
-0.23
1
Temp
0.15
0.32
-0.4
-0.05
0.01
-0.11
-0.24
0.16
-0.11
-0.26
0.1
1
Relevant
hydrochemical
relationships
are
indicated
by
the
values
highlighted
in
bold
Hierarchical
cluster
analysis
(HCA)
The
hierarchical
classification
is
widely
applied
in
the
Earth
Sciences
(Davis
1986)
and
is
often
used
in
hydrogeochem-
ical
data
classification
(Steinhorst
and
Williams
1985;
Schott
and
Van
Der
Wal
1992;
Ribeiro
and
Macedo
1995;
Oiler
et
al.
2002;
Fun
et
al.
2011;
Suk
and
Lee
1999;
Ab-
deramane
et
al.
2012).
In
this
study,
the
hierarchical
cluster
analysis
(HCA)
was
applied
to
the
raw
data.
Hydrochemical
data
of
the
12
variables
(temperature,
pH,
EC,
Na,
K,
Mg,
Ca,
Cl,
HCO
3
,
SO
4
,
NO
3
,
and
Si0
2
)
were
classified
into
distinct
groups
based
on
common
variables
they
share.
The
result
is
presented
in
the
dendrogram
(Fig.
4).
The
hierarchical
analysis
allowed
to
distinguish
three
clusters
of
water
(Fig.
4),
depending
on
their
chemistry.
Cluster
1
represents
four
samples
(Ab'a,
DAD2,
DJG,
Dad6)
which
are
characterized
by
moderate
salinity
with
TDS
between
800
and
900
mg/l.
Of
these
four
samples,
three
capture
Dalha
basalts
aquifer
(Dab2,
DjG,
Dad6)
contrary
to
the
sample
taken
at
Ab'aytou
(Ab'a),
in
the
western
part
of
the
study
area,
which
captures
the
alluvial
aquifer.
Cluster
2
represents
five
samples
(Dik9,
Galm2,
Hmbk,
Kontl,
Aby),
which
are
characterized
by
low
mineral
content
water
(TDS
<
500
mg/1)
located
in
recharge
areas.
Finally,
the
cluster
3
also
shows
five
samples
(DBo12,
Chk2,
M5,
DBol
1,
Awrl)
that
capture
the
basaltic
aquifers
(Dalha
and
strati-
form),
with
the
exception
of
Chk2
and
Awrl
water
points
located
in
sedimentary
formations
(Fig.
1).
The
samples
of
cluster
3
are
more
mineralized
(TDS
>
1,000
mg/1)
compared
to
the
other
two
groups.
1
Springer
Fig.
4
Hierarchical
cluster
analysis.
Dendrogram
of
water
samples
120
100
80
60
Cluster
3
Cluster
1
Cluster
2
ti
(
Dlien
IDmax)
*100
40
20
Environ
Earth
Sci
Ab'a
Dabd2
EIG
Dade
Dik9
Glikb2
arnbk
Kenn
AbY
Decitz
chk2
145
06011
Awn
These
highly
mineralized
waters
of
group
3
are
due
to
water-rock
interaction.
In
this
cluster,
major
ions
are
Na
+
and
Cl
-
ions,
with
S0
4
2-
,
Ca
2+
and
Mg
2+
ions
in
signif-
icant
amounts,
participating
thus to
the
water
mineralization.
Principal
components
analysis
(PCA)
Principal
components
analysis
(PCA)
is
a
multivariate
statistical
technique
used
to
identify
important
components
or
factors
explaining
most
of
the
variances
of
a
system
(Jollife
1985;
Meng
and
Maynard
2001;
Ouyang
2005;
Sarbu
and
Pop
2005).
In
PCA,
the
number
of
components
to
keep
was
based
on
the
Kaiser
criterion
(Kaiser
1958),
for
which
only
the
components
with
eigenvalues
greater
than
1
are
retained.
To
maximize
the
variance
of
the
principal
axes
retained,
the
Varimax
normalized
rotation
was
applied
(Usunoff
and
Guzman-Guzman
1989;
Melloul
and
Collin
1992;
Schott
and
Van
Der
Wal
1992;
Jayakumar
and
Siraz
1997;
Adams
et
al.
2001;
Aiuppa
et
al.
2003).
Principal
components
analysis
of
groundwater
chemical
variables
produced
three
components
(Table
4)
accounting
for
84.4
%
of
the
total
variance
of
the
dataset.
Table
4
presents
the
principal
components
loadings,
as
well
as
their
respective
explained
variance.
Loadings,
that
represent
the
importance
of
the
variables
for
the
components,
are
in
bold
for
values
greater
than
0.7
(Table
4).
Each
component
is
characterized
by
few
high
loadings,
except
PC1
(first
component)
which
has
six
high
loadings
(Table
4).
PC1
accounts
for
51.3
%
of
the
total
variance
and
has
a
strong
loading
for
EC,
Na,
K,
Ca,
Mg
and
Cl
and
negative
loadings
in
HCO
3
and
pH
(Table
4)
and
can
be
ascribed
to
natural
hydrogeochemical
evolution
of
groundwater
by
Table
4
Principal
components
loadings
and
explained
variance
for
the
three
components
with
Varimax
normalized
rotation
Parameters
PC-1
PC-2
PC-3
Cl
-
0.90
0.19
0.31
HCO
3
-
-0.61
0.65
0.09
S0
4
2-
0.47
-0.79
-0.19
NO
3
-
0.35
-0.22
0.71
Na
±
0.95
-0.1
0.05
IC
E
0.74
-0.33
0.41
Ca
0.8
-0.46
0.21
mg
z-F
0.88
0.19
0.30
Si0
2
0.12
0.21
0.92
pH
-0.23
-0.54
-0.37
EC
0.98
0.0
0.11
T
(°C)
0.19
0.76
-0.27
Var.
Expl.
5.45
2.41
2.03
%
Total
variance
51.33
20.09
11.00
Cumulative
(%)
51.33
71.42
82.41
High
loadings
are
indicated
in
bold
groundwater-geological
medium
interaction.
The
factorial
axis
1
defines
the
mineralization
of
water.
It
corresponds
to
the
waters
of
high
mineralization
represented
by
cluster
3
in
the
hierarchical
classification
previously
established.
PC2
accounts
for
20.1
%
of
the
total
variance
and
is
characterized
by
highly
positive
loading
in
HCO
3
(and
T
C)
and
highly
negative
loading
in
SO
4
.
The
factorial
axis
2
is
the
axis
expressing
the
evaporites.
Moreover,
low
loadings
of
the
variables
expressing
the
salinity,
namely
Cl
(0.19)
and
EC
(0.0)
imply
that
this
axis
corresponds
to
the
waters
belonging
to
the
group
with
lower
mineralization
defined
by
the
cluster
2.
PC3
accounts
for
11
%
of
the
total
1Springer
........
.....
0
\
KonI1
0
-
1
-
iingehr
N
o
0
Du
0
hmrs11
0
145
0
0
...............
-
1
-2
-3
7
6
5
4
4
-a
-7
-6
-5
-4
-3
-2
-1
0
Fact.
1
:
51,33%
2
3
4
5
Environ
Earth
Sci
variance
and
is
characterized
by
highly
positive
loading
in
NO
3
and
SiO
2
.
Taking
account
of
factor
loadings
for
the
parameters
governing
the
salinity,
the
principal
axis
3
corresponds
to
moderate
mineralization
waters
(cluster
1).
The
plot
of
variables
(Fig.
5)
on
the
plane
associated
with
the
two
principal
components
(PCland
PC2)
indicates
that
the
mineralization
of
water
in
the
study
area
is
more
affected
by
the
chemical
variables
such
as
Na,
Mg
and
Ca
for
cations, Cl
and
SO
4
for
anions
(Fig.
6).
On
the
other
1
0.
,
014
...'
:
i
.
t
....:
,
$'
1%l9A
3
t
il
Ci-
t
r
.
‘'..:P
mg2+
.
1
...
r
,,..
-
--
--........
2
i
...
i
\
--
-
_
,'
SiCi
..o
Hcct
0
..
0
y'
N
-----
-:::"
-
1,0
-
0,5
0,0
0,5
1,0
Fact.
I
:
51.33%
Fig.
5
Projection
of
the
hydrochemical
variables
on
the
plane
associated
with
principal
components
1
(Fact.
1)
and
2
(Fact.
2)
hand,
chemical
variables
plotted
in
groups
1
and
3
showed
in
Fig.
5
are
the
dominant
chemical
descriptors
of
groundwater
flowing
in
the
study
area.
The
concentration
of
HCO
3
variable
(group
2)
is
affected
by
the
concentration
of
variables
in
group
1
and
3
(i.e.
the
level
of
Na,
Cl,
Ca,
Mg).
Figure
6
shows
the
projections
of
samples
on
the
plane
associated
with
the
first
two
principal
components
(PC1
and
PC2)
which
clearly
distinguish
the
three
families
of
water
identified
in
the
hierarchical
clustering
above
mentioned.
Water
types
classification
and
origin
of
salinity
Piper
diagram
The
Piper
diagram
(Piper
1944)
allows
representing
several
water
samples
simultaneously.
It
is
particularly
suitable
for
studying
the
evolution
of
water
types
where
the
minerali-
zation
increases,
or
to
compare
groups
of
samples
with
each
other
and
indicates
the
types
of
dominant
cations
and
anions.
Thus,
to
better
illustrate
the
different
water
types
in
the
study
area,
all
hydrochemical
data
were
plotted
on
a
Piper
diagram.
Several
water
facies
can
be
depicted
in
this
diagram
(Fig.
7):
Na-HCO
3
-50
4
facies
characterizes
waters
with
low
mineralization
(Aby,
Kontl,
Galm2)
cir-
culating
mainly
in
the
sedimentary
aquifer
located
in
western
part
of
the
study
area,
Na—Mg-Cl-SO
4
facies
which
represents
the
highly
mineralized
water
exploited
in
Dou-
doub-Bolole
(DBoll,
DBol2)
and
Mouloud
(M5)
areas.
1.0
0.5
Cr)
0
,
0
-0.5
-1,0
Fig.
6
Projection
of
the
water
samples
on
the
plane
associated
with
principal
components
1
(Fact.
1)
and
2
(Fact.
2).
1
moderately
mineralized
water,
2
lowly
mineralized
waters,
3
highly
mineralized
waters
1Springer
25
-
1:1
line
20
-
Na
excess
-7
-
cr
.
15-
0
5
10
15
(meq/I)
v
E
10
-
z
5
-
0
0
o
Dalha
basalt
L.
Alluvial
aquifer
strati
form
basalt
Line
1
a
2
'
0
25
Fig.
8
Plot
of
Na
±
concentrations
versus
Cl
concentrations
Environ
Earth
Sci
Fig.
7
Piper
plot
of
the
100
hydrochemical
samples
Sedimentary
aquifer
Straliform
B.
Aquifer
Delha
EL
Aquifer
-Ps
Flow
direction
O
nt1
A
CChk2
Climb
-
"iv.~741.1
bY
anti
Chk2
100
100
CI
100
These
waters
flow
in
the
volcanic
aquifers
of
Dalha
basalts
(DBoll
and
DBol2)
and
stratiform
basalts
(M5).
These
two
dominant
facies
characterize
the
majority
of
samples.
However,
other
facies
are
also
pointed
out
such
as
Na-Ca-
50
4
(Chk2),
Na-Cl-SO
4
(Awrl),
Na-Mg-HCO
3
-Cl
(Dik9),
Na-HCO
3
-Cl
(Dad6)
and
Na-Mg-Ca-Cl-HCO
3
(Hmbk).
These
facies
evolve
differently
in
the
two
main
flow
directions
identified
by
the
piezometric
map
(Fig.
2).
Indeed,
within
the
Dalha
basalts
aquifer,
three
water
types
are
found
depending
on
whether
the
flow
is
directed
toward
the
northeast
or
northwest.
Thus,
the
Na-HCO
3
-Cl
facies
(Dad6)
evolves
into
Na—Mg-Cl-SO
4
facies
(characterizing
the
waters
with
high
mineralization)
in
the
flow
direction.
Following
the
flow
direction
oriented
north—west,
the
Na-
Mg-Cl-HCO
3
facies
is
encountered
characterizing
water
exploited
in
Dikhil
area
(Fig.
7).
Bi-elements
diagrams
A
series
of
plots
illustrate
the
key
hydrochemical
trends
along
the
groundwater
flow
path
(Yuce
2007;
Zhu
et
al.
2007;
Jalali
2009;
Nandimandalam
2011;
Diaw
et
al.
2012)
showed
in
Figs.
8,
9,
10, 11,
12,
13.
Chloride
is
used
as
a
conservative
reference
element
for
studying
the
water—rock
interaction.
A
main
feature
of
the
groundwater
is
the
enrichment
of
Na
relative
to
Cl
(Fig.
8).
Also,
the
Na/Cl
ratios
of
a
majority
of
samples
(62
%
of
samples)
are
greater
than
1
except
for
the
more
saline
water
(Table
1).
The
high
Na/Cl
ratios
are
indicative
of
strong
water—rock
interaction.
Analytical
data
in
Fig.
8
deviate
from
the
expected
1:1
line
relation
indicating
that
a
large
fraction
of
sodium
is
1Springer
Fig.
9
Plot
of
HCO
3
concentrations
versus
Ca
concentrations
7
6
5
V;
4
V
o
o
3
2
1
25
A
1:1
line
20
-
0
5-
0n
0
0
5
10
15
20
25
Cl
-
(meq/1)
10
9
8
7
6
5
Z
a
n
4
3-
2-
0
0
0
0 0
a
2
3
4
Cat*
(meq/I)
1
0
0
5
0.051
0.046
-
0.041-
0.036
-
0.031-
0.026
-
0.021-
0.016
-
0.011-
0.006
-
0.001
1
2
0
O
CO
10
CI
-
(meq/I)
0
0
A
0
100
Environ
Earth
Sci
O
O
O
1:1
line
00
O
O
2
3
4
5
6
7
HCO3
(meq/I)
10
9
0
8
7
5•
6
E
5
1:1
line
0)
4
1,1
2
3
0
8
2
0
1
0
0
1
2
3
4
5
6
7
HCO3
(meq/I)
o
0
O
Fig.
10
Plot
of
SO4
concentrations
versus
Cl
-
concentrations
(see
Fig.
8
caption)
Fig.
11
Plot
of
Br
concentrations
versus
CF
concentrations.
(see
Fig.
8
caption)
associated
with
an
anion
other
than
chloride.
Na
enrich-
ment
is
linked
with
processes
of
water—rock
interaction.
During
percolation
to
the
deeper
zones,
Na/Cl
ratios
increase
progressively
by
dissolution
of
sodium
in
volcanic
rocks
(Join
et
al.
1997).
Grunberger
(1989)
has
shown,
on
one
hand,
that
chloride
ions
stay
in
solution
in
soil
and
become
concentrated
by
evaporation,
and
that
chloride
in
volcanic
rocks
is
practically
unavailable
for
dissolution
during
percolation
and,
on
the
other
hand,
that
the
sodium
Fig.
12
Plot
of
Mg
concentrations
versus
Ca
concentrations.
(see
Fig.
8
caption)
4,0
3,5
3,0
-
2,5
r
2.0
.
1,5
-
1,0
-
0,5
-
0,0
0
A
0
o
0
odP
a
50
100
150
200 250
300
Nail(
350
400
Fig.
13
Relationships
between
Ca/Mg
and
Na/K
ratios
(see
Fig.
8
caption)
present
in
groundwater
is
a
result
of
the
hydrolysis
of
sodium-rich
minerals
(e.g.
albite),
which
are
significantly
present
according
to
the
chemical
composition
of
Dalha
basalts
previously
mentioned.
The
governing
equation
of
the
dissociation
is:
NaAlSi
3
0
8
+
4H
+
+
4H20
—>
Na
+
+
A1
3+
+
H
4
SiO
4
The
high
Na/Cl
ratios
in
the
low
and
moderate
salinity
groundwater
(Table
1)
may
be
explained
by
Na
being
derived
predominantly
from
weathering
of
plagioclase
in
the
study
area.
L
4
t1
Springer
Environ
Earth
Sci
Bicarbonates
are
found
both
in
alluvial
(Aby,
Kontl,)
and
sedimentary
(Galm2,
Awrl)
waters
as
well
as
in
water
flowing
in
Dalha
basalts
aquifer
(DjG,
Dad6,
Hmbk,
Dik9).
A
lack
of
positive
correlation
is
observed
among
HCO
3
and
Mg
(-0.40)
and
HCO
3
and
Ca
(-0.78)
(Fig.
9),
suggesting
that
dissolution
of
dolomite
or
calcite
may
not
be
the
main
source
of
bicarbonates,
calcium
and
magnesium.
However,
calcite
is
found
in
basalts
fractures
and
cracks
(Jalludin
1993).
Its
dissolution
may
thus
release
HCO
3
and
Ca.
A
lack
of
correlation
(R
2
=
0.14)
between
Cl
and
SO4
concentrations
(Fig.
10)
and
the
SO
4
/Cl
ratios
for
a
majority
of
samples
(SO
4
/C1
<
1,
Table
1),
except
for
four
wells,
militates
against
gypsum
dissolution
as
a
major
source
of
SO
4
.
These
four
wells
are
those
located
in
Checkeyti
waddi's
area
(Chk2,
AbY,
Kontl
and
Galm2)
whose
SO
4
/C1
ratio
is
greater
than
1
(Table
1).
The
origin
of
sulfates
in
these
samples
may
probably
result
from
evaporates
dissolution
such
as
gypsum
(ratio
above
0.5)
contained
in
alluvial
aquifer
sediments.
Table
1
also
shows
that
the
Ca/SO
4
ratio
is
less
than
1
in
most
samples,
with
the
exception
of
three
wells
(DjG,
and
Hmbk
Dik9)
where
this
ratio
is
greater
than
1.
Excess
calcium
in
these
samples,
from
water
flowing
through
the
Dalha
basalts
aquifer,
probably
results
from
the
alteration
of
calcic
plagioclase
such
as
anorthite
highlighted
in
the
chemical
composition
of
Dalha
basalts
(Gasse
et
al.
1986).
The
dissociation
equation
of
anorthite
is:
CaAl2Si208
+
4H
+
+
4H
2
0
—>
Ca
2
+
+
2A1
3
+
+
3H
4
SiO
4
In
general,
the
use
of
the
Br/Cl
ratio
is
the
most
appropriate
method
of
identifying
the
source
of
chloride
in
ground-
water
(Rittenhouse
1967).
And
also
the
Br/Cl
ratio
may
help
to
identify
the
flow
system
of
groundwater,
when
the
salinity
increases
due
to
the
water—rock
interaction.
According
to
Marjoua
et
al.
(1997),
the
Br/Cl
ratio
of
seawater
is
3.47
x
10
-3
and
that
of
the
dissolution
of
halite
is
0.183
x
10
-3
.
In
the
study
area,
most
samples
exhibit
a
greater
ratio
well
above
that
of
halite
(Table
1).
These
results
argue,
therefore,
against
the
hypothesis
of
dissolution
of
halite
in
the
matrix
of
the
aquifer.
This
is
confirmed
by
the
graph
of
Br
versus
Cl
(Fig.
11).
Another
feature
is
the
significant
enrichment
in
Mag-
nesium
relative
to
Calcium
(Fig.
12)
in
a
majority
of
samples.
The
source
of
magnesium
in
volcanic
aquifer
is
probably
the
weathering
of
mafic
minerals
such
as
biotite
by
reaction
such
as:
14KA1Mg
3
Si
3
0
10
(OH)
2
+
98H
2
CO
3
+
7H
2
0
=
7Al
2
Si
2
0
5
(OH)
4
+
42Mg
2
+
+
14k+
+
98HCO
+
28H
4
SiO
4
biotite
kaolinite
Moreover,
the
lack
of
correlation
between
Ca/Mg
and
Na/
K
ratios
(Fig.
13),
and
the
high
Na/Cl
ang
Mg/Ca
ratios
previously
described,
militates
against
cation
exchange
between
water
and
aquifer,
in
other
words
there
is
no
adsorption
and
release
of
Ca
2+
.
Isotope
geochemistry
The
environment
isotopes
of
oxygen
8
18
0
and
hydrogen
8
2
H
are
excellent
tracers
for
determining
the
origin
of
groundwater
and
are
widely
used
in
studying
the
natural
water
circulation
and
groundwater
movement
(Ali
2004;
Kamel
et
al.
2005;
Yitbarek
et
al.
2012).
The
absence
of
isotopic
data
reference
rainfall
in
Djibouti
and
in
the
study
area
has
led
us
to
take
as
reference
the
data
measured
by
the
IAEA
at
Addis
Ababa
(Ethiopia)
near
our
study
area
(IAEA
1992).
In
this
study,
we
used
as
local
meteoric
water
line
(LMWL)
the
line
defined
from
isotopic
refer-
ence
data
at
Addis
Ababa
station.
Groundwater
in
the
study
area
exhibits
isotopic
contents
varying
in
space.
Thus,
8
18
0
values
range
from
—3.55
to
1.15
%o
(vs.
V-SMOW)
with
an
average
of
—2
%o.
The
deuterium
contents
vary
between
—0.87
and
—21.33
%o
(vs.
V-SMOW)
with
an
average
of
—5.10
%o
(Table
5).
Concerning
the
deuterium,
a
few
wells
have
positive
values
of
8
2
H
(Table
5).
The
8
18
0
and
8
2
H
values
of
water
from
wells
and
boreholes
sampled
in
this
study
are
plotted
and
compared
to
Local
Meteoric
Water
Line
(LMWL)
and
Global
Meteoric
Water
Line
(GMWL),
whose
equations
are
respectively
(Fig.
14):
LMWL
:
6
2
H
=
7.2
60
18
+
12
GMWL
:
6
2
H
=
8
60
18
+
10.
The
main
feature
resulting
from
the
analysis
of
8
2
H
vs.
8
18
0
graph
is
that
groundwater
samples
in
the
study
area
are
plotted
mostly
close
to
the
GMWL,
indicating
that
the
groundwater
is
of
meteoric
origin
(Fig.
14).
The
samples
classified
in
group
1
represent
waters
enriched
in
stable
isotopes,
in
other
words
those
affected
by
evaporation.
Kebede
(2004)
reached
the
same
conclusion,
while
study-
ing
the
recharge
processes
in
the
Afar
region
in
Ethiopia.
The
points
located
above
the
global
meteoric
water
line
but
plotted
closer
to
the
local
meteoric
water
line
represent
mainly
water
from
Dalha
basalts
aquifer
(group
2)
which
is
the
main
aquifer
system
in
the
study
area.
These
samples
are
depleted
in
heavy
stable
isotopes,
relative
to
the
sam-
ples
identified
in
group
1.
Their
mean
composition
is
—2.60
%o
8
18
0
and
—5.26
%o
82H
(Table
5).
The
sample
from
M5
well
(Fig.
14),
representing
group
3,
is
identified
as
affected
by
slight
evaporation
(it
is
depleted
in
heavy
stable
isotopes)
as
it
is
located
below
and
near
the
global
meteoric
water
line
(GMWL).
This
well
1Springer
OH
(
%e
)
vs
V-
SMO
W
15
10
5
0
5
-19
-15
-20
-
-25
-4
a
2
H
=
86'
8
0
10
I
.5
2
H
=
7200
1
2
Pelf
Xy.>""
3
1
-3
-2
-1
0
5"
O
(960
vs
V-SMOW
1
0
°
LMW
[raffia
aquifer
MISS,
at.
aquifer
Alluvial
aquifer
GMWL
1
2
Environ
Earth
Sci
Table
5
Isotopic
composition
of
groundwater
in
the
study
area
Date
of
sampling
Sample
ID
Well
depth
(m)
Altitude
(m)
618
0
(%.)
62H
(%
o)
13-10-2010
P.
Harrou
10
520
-1.23
1.65
13-10-2010
F.
Moul5
115
588
-3.55
-21.33
13-10-2010
F.
Dik6
103
475
-2.09
-8.46
13-10-2010
P.
DjG
11
683
-1.81
-3.98
13-10-2010
P.
Ab'a
3
372
-2.35
-8.84
13-10-2010
F.
Dik9
141
510
-2.11
-6.87
13-10-2010
P.
MDk
11
678
-1.84
-0.87
13-10-2010
P.
AbY
11
402
-1.94
-4.39
13-10-2010
F.
Dikl
1
102
487
-2.87
-9.33
13-10-2010
P.
Chkl
3
387
-1.04
1.31
13-10-2010
F.
Galm2
73
389
-2.04
-5.15
13-10-2010
P.
Bond
4.5
461
-0.50
7.87
13-10-2010
F.
Dad6
129
650
-2.45
-1.20
13-10-2010
P.
Chk2
8
389
1.15
18.39
13-10-2010
F.
Batoul
60
480
-1.21
6.66
13-10-2010
P.
Kontl
7
375
-0.19
18.85
F
well,
P
hand-dug
well
Fig.
14
Relationships
between
6
2
H
and
6
18
0
captures
the
stratiform
basalts
aquifer.
Isotopic
composi-
tion
of
samples
plotted
closer
to
LMWL
suggests
fast
percolation
of
rainwater
(before
evaporation).
Results
from
environmental
isotopic
data
presented
in
this
study
are
in
perfect
agreement
with
the
classification
previously
established
in
cluster
analysis.
Conclusion
This
integrated
study
has
demonstrated
the
wide
spatial
variations
of
the
hydrochemistry
of
the
volcano-sedimen-
tary
complex
aquifer
system.
To
further
refine
geochemical
interpretation,
major
ion
chemistry
and
multivariate
sta-
tistical
methods
(HCA
and
PCA)
were
used.
The
multi-
variate
methods
proved
well-suited
because
of
highly
variable
groundwater
geochemistry
influenced
by
a
variety
of
geological factors.
Conventional
methods
using
Piper
diagram
showed
complex
water
types
dominated
by
Na-
HCO
3
-Cl
and
Na-SO
4
-Cl
facies.
Cluster
analysis
has
suc-
cessfully
extracted
three
clusters:
cluster
1
(4
observa-
tions),
cluster
2
(5
observations)
and
cluster
3
(5
observations).
These
methods
describe
differences
in
the
chemistry
of
the
groundwater
resulting
from
the
different
aquifer
materials
through
which
they
have
flown.
Sodium
and
magnesium
are
the
dominant
cations.
Bicarbonates
and
sulfates
ions
are
the
dominant
anions.
The
quality
of
the
waters
is
mainly
due
to
water-rock
interactions.
All
the
samples
are
mesothermal
waters
(temperature
vary from
31
to
44
C)
and
are
slightly
alkaline.
Also,
the
water
quality
problems
can
be
successfully
handled
by
the
use
of
prin-
cipal
component
analysis.
It
was
concluded
that
12
chemical
parameters
which
were
measured
in
the
collected
samples
can
be
substituted
by
three
components
which
represent
82.4
%
of
the
statistical
information,
the
hydro-
chemical
processes
as
well
as
their
geographic
distribution.
Results
from
multivariate
statistical
analysis
highlight
three
groundwater
groups,
each
restricted
to
specific
location
and
water
quality
property:
(1)
the
groundwater
characterized
by
low
ionic
concentrations
and
located
at
the
recharge
zones;
(2)
the
groundwater
characterized
by
moderate
salinity
and
(3)
the
highly
mineralized
waters
mainly
flowing
in
the
eastern
and
central
part
of
the
study
area,
mainly
in
volcanic
aquifers
(Dalha
and
stratiform
basalts
aquifers).
These
findings
provide
a
preliminary
understanding
of
the
overall
functioning
of
this
complex
volcano-sedimen-
tary
system.
Additional
investigations
(pumping
tests,
numerical
modeling)
are
in
progress
to
achieve
a
more
comprehensive
understanding
of
this
system
and
to
develop
a
sustainable
exploitation
of
its
groundwater
resources.
1
Springer
Environ
Earth
Sci
Acknowledgments
The
authors
gratefully
thank
Claude
Fontaine,
University
of
Poitiers,
for
his
help
in
performing
the
analyses.
They
are
grateful
to
the
staff
of
the
Hydrochemistry
Laboratory
of
the
Center
for
Studies
and
Research
of
Djibouti,
who
contributed
strongly
to
the
field
works.
References
Abderamane
H,
Razack
M,
Vassolo
S
(2012)
Hydrogeochemical
and
isotopic
characterization
of
the
groundwater
in
the
Chari-
Baguirmi
depression,
Republic
of
Chad.
Environ
Earth
Sci.
doi:
10.1007/s12665-012-2063-7
Adams
S,
Titus
R,
Pietersen
K,
Tredoux
G,
Harris
C
(2001)
Hydrochemical
characteristics
of
aquifers
near
Sutherland
in
the
Western
Karoo,
South
Africa.
J
Hydrology
241:91-103
Aiuppa
A,
Bellomo
S,
Brusca
L,
D'Alessandro
W,
Federico
C
(2003)
Natural
and
anthropogenic
factors
affecting
groundwater
quality
of
an
active
volcano
(Mt.
Etna,
Italy).
Appl
Geochem
18:863-882
Ali
MS
(2004)
Use
of
chloride
mass-balance
and
environmental
isotopes
for
evaluation
of
groundwater
recharge
in
the
alluvial
aquifer,
Wadi
Tharad,
West
Saudi
Arabia.
Environ
Geol
46:741-749
BGR
Bundesanstalt
fiir
Geowissenschaften
and
Rohstoffe
(1982).
Inventaire
et
mice
en
valeur
des
ressources
en
eau
de
la
Republique
de
Djibouti.
Inventory
and
development
of
water
resources
of
the
Republic
of
Djibouti.
In
French.
Unpublished
report.
Volume
1,
Hannover
Craig
H
(1961)
Standards
for
reporting
concentrations
of
deuterium
and
oxygen-18
in
natural
waters.
Science
133:1702-1703
Davis
JC
(1986)
Statistics
and
data
analysis
in
geology.
Wiley,
New
York
Demange
J,
Stieljes
L
(1975)
Geologie
de
la
region
sud-ouest
du
TFAI
(region
Lac-Abhe-Lac
Asal).
Geology
of
the
southwestern
TFAI
(Region
of
Lake
Asal-
Lake
ABHE).
In
French.
Bull
BRGM
2(IV):83-114
Diaw
M,
Faye
S,
Stichler
W,
Maloszewski
P
(2012)
Isotopic
and
geochemical
characteristics
of groundwater
in
the
Senegal
River
delta
aquifer:
implication
of
recharge
and
flow
regime.
Environ
Earth
Sci
66(4):1011-1020
Freeze
RA,
Cheery
JA
(1979)
Groundwater.
Prentice-Hall,
Engle-
wood
Cliffs
Fun
W,
Razack
M,
Abiye
TA,
Kebede
S,
Legesse
D
(2011)
Hydrochemical
characterization
of
complex
volcanic
aquifers
in
a
continental
rifted
zone:
the
Middle
Awash
basin,
Ethiopia.
Hydrogeol
J
20(2):385-400
Gasse
F
(1975)
L'evolution
des
lacs
de
l'Afar
central
(Ethiopie
et
TFAI)
du
plio-pleistocene
a
l'actuel.
The
evolution
of
lakes
in
central
Afar
(Ethiopia
and
TFAI)
of
Plio-Pleistocene
to
the
present.
In
French.
Doctorat
d'Etat
Thesis.
University
Paris
VI
Gasse
F,
Maze
G,
Recroix
F,
Ruegg
JC,
Varet
J
(1986)
Carte
geologique
de
la
Republique
de
Djibouti
a
1/100,000.
Notice
Feuille
Ali-Sabieh.
Geological
Map
of
the
Republic
of
Djibouti
at
1/100,000.
Notice
of
Ali
Sabieh
Sheet.
ORSTOM,
Paris,
p
104
Grunberger
0
(1989)
Etude
geochimique
et
isotopique
de
l'infiltration
sous
climat
tropical
contrasts,
Massif
du
Piton
des
Neiges.
Geochemical
and
isotopic
study
of
infiltration
under
contrased
tropical
climates,
Massive
of
Piton
des
Neiges.
In
French.
PhD
thesis,
University
Paris
XI
Gilder
C,
Thyne
GD,
McCray
JE,
Turner
AK
(2002)
Evaluation
of
graphical
and
multivariate
statistical
methods
for
classification
of
water
chemistry
data.
Hydrogeol
J
10:455-474
IAEA
(1992)
Stable
isotope
hydrology.
Technical
rapport
series,
International
Atomic
Energy
Agency,
Vienna
Jalali
M
(2009)
Geochemistry
characterization
of
groundwater
in
an
agricultural
area
of
Razan,
Hamadan,
Iran.
Environ
Geol
56:1479-1488
Jalludin
M
(1989)
Implantation
des
forages
de
reconnaissances
pour
l'
alimentation
en
eau
de
la
ville
d'Ali-Sabieh.
Implantation
of
reconnaissance
boreholes
for
water
supply
of
the
city
of
Ali-
Sabieh.
In
French.
ISERST,
Djibouti.
Unpublished
report
Jalludin
M
(1993)
Proprietes
geometriques
et
hydrodynamiques
des
aquiferes
en
milieux
volcaniques
fissures
sous
climat
aride.
Republique
de
Djibouti.
Geometric
and
hydrodynamic
properties
of
aquifers
in
fissured
volcanic
environments
under
arid
climate.
Republic
of
Djibouti.
In
French.
PhD
thesis,
University
of
Poitiers
Jalludin
M,
Razack
M
(2004)
Assessment
of
hydraulic
properties
of
sedimentary
and
volcanic
aquifer
systems
under
arid
conditions
in
the
Republic
of
Djibouti
(Horn
of
Africa).
Hydrogeol
J
12:159-170
Jayakumar
R,
Siraz
L
(1997)
Factor
analysis
in
hydrogeochemistry
of
coastal
aquifers—a
preliminary
study.
Environ
Geol
31:174-177
Join
JL,
Coudray
J,
Longworth
K
(1997)
Using
principal
components
analysis
and
Na/Cl
ratios
to
trace
groundwater
circulation
in
a
volcanic
island:
the
example
of
Reunion.
J
Hydrol
190:1-18
Jollife
TT
(1985)
Principal
component
analysis.
Verlag,
New
York
Kaiser
HF
(1958)
The
Varimax
criterion
for
analytical
rotation
in
factor
analysis.
Psychometrika
23b:187-200
Kamel
S,
Dassi
L,
Zouari
K
(2005)
Geochemical
and
isotopic
investigation
of
the
aquifer
system
in
the
Djerid-efzaoua
basin,
southern
Tunisia.
Environ
Geol
49:159-170
Kebede
S
(2004)
Methodes
isotopiques
et
geochimiques
des
eaux
souterraines
et
hydrologie
des
lacs:
Cas
du
bassin
du
Nil
Bleu
et
du
Rift
Ethiopien.
Isotopic
and
geochemical
methods
and
groundwater
hydrology
of
lakes:
Case
of
Blue
Nile
basin
and
the
Ethiopian
Rift.
In
French.
PhD
thesis.
University
of
Avignon
Marjoua
A,
Olive
Ph,
Jusserand
C
(1997)
Contribution
of
chemical
and
isotopic
tools
to
the
identification
of
water
salinization
origins:
the
case
of
coastal
Chaouia
aquifer
(Morocco).
Rev
Sci
Eau
4:489-505
Melloul
A,
Collin
M
(1992)
The
principal
Component
Statistical
Method
as
a
complementary
approach
to
geochemical
methods
in
water
quality
factor
identification,
application
to
the
coastal
plain
aquifer
of
Israel.
J
Hydrol
140:49-73
Meng
SX,
Maynard
JB
(2001)
Use
of
statistical
analysis
to
formulate
conceptual
models
of
geochemical
behavior:
water
chemical
data
from
Butucatu
aquifer
in
Sao
Paulo
State,
Brazil.
J
Hydrol
250:78-97
Nandimandalam
JR
(2011)
Evaluation
of
hydrogeochemical
pro-
cesses
in
the
Pleistocene
aquifers
of
Middle
Ganga
Plain,
Uttar
Pradesh,
India.
Environ
Earth
Sci
65(4):1291-1308
Ouyang
Y
(2005)
Evaluation
of
river
water
quality
monitoring
stations
by
principal
component
analysis.
Water
Res
3:2621-
2635
Piper
AM
(1944)
A
graphic
procedure
in
the
geochemical
interpre-
tation
of
water-analyses.
American
Geophysical
Union
Trans-
action,
Hydrology
pp
914-923
Razack
M,
Dazy
J
(1990)
Hydrochemical
characterization
of
groundwater
mixing
in
sedimentary
and
metamorphic
reservoirs
with
a
combined
use
of
Piper's
principle
and
factor
analysis.
J
Hydrol
114:371-393
Ribeiro
L,
Macedo
ME
(1995)
Application
of
multivariate
statistics,
trend
and
cluster
analysis
to
groundwater
quality
in
the
Tejo
and
Sado
aquifer.
In:
Groundwater
quality:
Remediation
and
Pro-
tection.
Proceedings
of
the
Prague
Conference,
May
1995.
IAHS
Publ.
225:39-47
Rittenhouse
G
(1967)
Bromid
in
oil
field
waters
and
use
in
determining
possibilities
of
origin
of
these
waters.
Bull
Am
Assoc
Pet
Geol
51:2430-2440
1
Springer
Environ
Earth
Sci
Sarbu
C,
Pop
HF
(2005)
Principal
component
analysis
versus
fuzzy
principal
component analysis:
a
case
study:
the
quality
of
Danube
water
(1985-1996).
Talanta
65:1215-1220
Schott
PP,
Van
Der
Wal
I
(1992)
Human
impact
of
regional
groundwater
composition
through
intervention
in
natural
flow
patterns
and
changes
in
land
use.
I
Hydrol
134:297-313
StatSoft
Inc.
(2008)
STATISTICA
(Data
Analysis
Software
System),
version
7.2300
East,
14
th
St,
Tulsa,
OK
74104
Steinhorst
RK,
Williams
RE
(1985)
Discrimination
of
groundwater
sources
using
analysis,
MANOVA,
Canonical
analysis
and
discriminant
analysis.
Water
Resour
Res
21:1149-1156
Suk
H,
Lee
KK
(1999)
Characterization
of
hydrochemical
system
through
multivariate
analysis:
clustering
into
groundwater
zones.
Groundwater
37(3):358-366
Todd
K
(1980)
Groundwater
hydrology,
2nd
edn.
Wiley,
New
York
Usunoff
EJ,
Guzman—Guzman
A
(1989)
Multivariate
analysis
in
hydrochemistry:
an
example
of
the
use
of
factor
and
correspon-
dence
analyses.
Ground
Water
27:27-34
WHO
World
Health
Organization
(1993).
Guidelines
for
Drinking-
Water
Quality,
2nd
edition.
Geneva.
http://www.who.int/water_
sanitation_health/dwq/2edvoll
i.pdf
Yitbarek
A,
Razack
M,
Ayenew
T,
Zemedagegnehu
E,
Azagegn
T
(2012)
Hydrogeological
and
hydrochemical
framework
of
Upper
Awash
River
basin,
Ethiopia:
with
special
emphasis
on
inter-
basins
groundwater
transfer
between
Blue
Nile
and
Awash
Rivers.
I
Mr
Earth
Sc
65:46-60
Yuce
G
(2007)
A
Geochemical
study
of
the
groundwater
in
the
Misli
basin
and
environmental
implications.
Environ
Geol
51:857-868
Thu
F,
Li
Z,
Su
YH,
MT,
Zhang
YY
(2007)
Hydrogeochemical
and
isotope
evidence
of
groundwater
evolution
and
recharge
in
Minqin
basin,
Northwest
China.
I
Hydrol
333:239-251
1
Springer