Journal o f Hydrology, 135 (1992) 237-257
Elsevier Science Publishers B.V., Amsterdam
237
[21
Dissolved major elements exported by the Congo
and the Ubangi rivers during the period 1987-1989
Jean-Luc Probst a, Renard-Roger NKounkou ~', G6rard Krempp ~, Jean-Pierre
Bricquet b, Jean-Pierre Thi6baux c and Jean-Claude Olivry d
~Centre de G~ochimie de ia Surface (CNRS), I. rue Blessig, 67084 Strasbourg Cedex, France
bCentre ORSTOM, B.P. 181, Brazzaville, Congo
~Centre ORSTOM. B.P. 893, Bangui. Central dfrican Republic
OCentre ORSTOM, B.P. 2528. Bamako. Mali
(Received 20 September 1991; accepted 6 October 1991)
ABSTRACT
Probst, J.-L., NKounkou, R.-R., Krempp, G., Bricquet, J.-P., Thi6baux, J.-P. and Olivry, J.-C., 1992.
Dissolved major elements exported by the Congo and the Ubangi rivers during the period 1987-1989.
J. Hydrol., 135: 237-257.
On the basis of monthly sampling during the period 1987-1989, the geochemis:ry of ~he Congo and the
Ubangi (second largest tributary of the Congo) rivers was studied in order (I) to understand the seasonal
variations of the physico-chemical parameters of the waters and (2) to estimate the annual dissolved fluxes
exported by the two rivers. The results presented here correspond to the first three years of measurements
carried out for a scientific programme (Interdisciplinary Research Programme on Geodynamics of PeriAtlantic lntertropical Environments, Operation 'Large River Basins' (PIRAT-GBF) undertaken jointly by
lnstitut National des Sciences de l'Univers (INSU) and Institut Fran~ais de Recherche Scientifique pour
ie D6veloppement en Coop6ration (ORSTOM)) planned to run for at least ten years.
The Congo River is more dPuted than the Ubangi (34mgl -I vs. 42mgl-I). For both rivers, the
inorganic dissolved load is composed mainly of HCOf and SiO,. The chemical composition of the water
does not change with time. in the Ubangi River, because of the presence of Precambrian carbonate rocks
in its catchment, the proportions of HCOf and Ca2÷ are higher. On a seasonal scale, the concentration
of dissolved cations and anions varies inversely with discharge, except silica. The comparison of the
discharge-concentration relationship with a theoretical "zero dilution' shows that the evolution of the
concentration of dissolved substances is a simple dilution by the surface waters, with, in the case of the
Ubangi, a small supply of dissolved substances by the surface waters.
Using three different methods of calculation, the estimated annual inorganic dissolved flux of the Congo
ranges from 39 x 106 to 44 × l0t` tons (according to lhe year), with about 10% of this coming from the
Ubangi drainage basin.
Correspondence to: J.-L. Probst, Centre de G~ochimie de la Surface (CNRS), 1, rue Blessig,
67084 Strasbourg Cedex, France.
0022-1694/92/$05.00
© 1992 - - Elsevier Science Publishers B.V. All rights reserved
238
J.-L. PROBST ET AL.
INTRODUCTION
Until recently, the geochemistry of the Congo River has been poorly
documented. Most of the data on dissolved solids quoted by different authors
are based on one or few measurements, not taking into account seasonal
variations. Clerfayt (1955) was probably the first to sample the Congo River
water in January 1949 and to measure its chemical composition. Since then,
some other data have been published by various authors (Symoens, 1968;
Meybeck, 1978; Figu~res et al., 1978; Van Bennekom et al., 1978; Molinier,
1979). A longer series of measurements was carried out by Deronde and
Symoens (1980) from December 1976 to November 1977 and allowed them to
calculate an annual flux of dissolved solids of about 35.5 x 106tons.
NKounkou and Probst (1987) have reviewed the available data on the Congo
basin in a synthesis presenting hydrological features, dissolved and suspended
materials transported by rivers, the carbon cycle, rock weathering and the
erosion balance of the Congo basin. Since 1987, a scientific programme
(Interdisciplinary Research Programme on Geodynamics of Peri-Atlantic
Intertropical Environments, Operation 'Large River Basins' (PIRAT-GBF)
undertaken jointly by Institut National des Sciences de l'Univers (INSU) and
Institut Fran~ais de Recherche Scientifique pour le D6veloppement en
Cooperation (ORSTOM)), has been carried out in the Congo River and its
second largest tributary, the Ubangi River. The main scientific objective of
this programme is to determine the interannual fluctuations of dissolved and
suspended material carried by the river into the ocean, in relation to the
hydroclimatic fluctuations (Olivry et al., 1988; NKounkou, 1989; Bricquet,
1990; Probst, 1990a,b: Giresse et al., 1990; Jouanneau et al., 1990; Moukolo
et al., 1990; NKounkou et al., 1990; Mariotti et al., 1991). The project also
aims to estimate the interannual fluxes of atmospheric CO2 consumed by rock
weathering, which are drained into the ocean in bicarbonate form or in all
other dissolved and particulate carbon forms (NKounkou and Probst, 1987;
Probst, 1990b; Probst et al., 1992).
The purpose of this paper is to present the initial results obtained
concerning the water chemistry and the dissolved mineral fluxes exported by
the Congo and the Ubangi rivers during the first three years (1987-1989) of
the PIRAT programme.
SAMPLING PROCEDURES AND ANALYTICAL TECHNIQUES
River water samples were collected at Bangui for the Ubangi and 40 km
upstream from Brazzaville for the Congo (Fig. 1). At Brazzaville, the drainage
area of the Congo is 3.475 x 106knl2 (the total drainage area of the Congo
is 3.073 x 106km 2), with an interannual water discharge of 41 000 m 3s- ~.The
239
DISSOLVED MAJOR ELEMENT EXPORT: CONGO AND UBANGI RIVERS
,
i
tllg
|
u
15"' IZ
210°
25 °
liT'
,°-1
.
5 °
0°
0o
.,
13l't.A7+7_J,V I l .L I".
~,)
KINSIIASA
/
. . . . .
)~
Io °
H
0 o
*
E1u
Sampling station
%
15 °
mR
5GO km
0
Limits of the Ubang| basin
Scale
20 °
mn
E
,,,,
Fig. 1. G e n e r a l m a p o f the C o n g o and Ubangi river basins.
250
30 °
!
t
E
.
240
J.-L. PROBST ET AL.
drainage area and the water discharge of the Ubangi river at Bangui are
0.5 x 106km2 and 4300m3s -~ respectively.
Water samples were collected monthly by the ORSTOM teams, taken from
the river surface with a 101 tin, decanted into 200ml polyethylene bottles and
then dispatched to the laboratory for analysis. Filtration through a 0.45/~m
millipore filter was made before analysis and the following techniques were
used to analyse the different physico-chemical parameters (Krempp, 1988).
(1) pH was measured only in the laboratory with the Mettler DL 40 RC,
with an accuracy of + 0.02 pH units. This equipment was also used for
alkalinity (HCOF) measurement, by titration with HCI, with an accuracy of
0.001 mmol l -~ .
(2) Water conductivity was measured with the Hanna HI 8633 conductimeter and expressed at 20°C.
(3) The colorimetric method was used for the analysis of NH~ and H4SiO4
on an automatic Technicon colorimeter. The method is based on the photospectroscopy of the NH~-sodium salicylate-chloride and H4SiO4-ammonium
molybdate complexes. The sensitivity of the equipment is 0.001 mmol 1(4) The major cations (Ca 2~, Mg 2+, Na + and K + ) were analysed by atomic
absorption, with an air-C2H2 gaseous mixture. La was added (0.5%) to the
sample for Ca 2+ and Mg 2+ analysis, and Cs20 (0.2%) for K +. The measurement is made with an accuracy of 0.001 mmol 1-~.
(5) For the major anions (CI-, SO42- and NO;), liquid chromatography
was used on a Dionex chromatograph. The detection limit is 0.001 mmoll -~.
WATER CHEMICAL COMPOSITION
In spite of its high discharge, the Congo River is relatively diluted, if
compared, for example, to the Amazon River (Stallard and Edmond, 1983).
The total inorganic dissolved solids load averaged 34 mg 1-I for the period
1987-1989 (Table 1). The previous study of Deronde and Symoens (1980)
based on monthly sampling during a full year gave a lower value of 28 mg 1- t.
The Ubangi waters are more concentrated than the Congo, with an average
dissolved load of 42 mg 1-t
The ionic composition of the Congo and Ubangi waters is shown in Fig. 2.
For both rivers, Ca 2+ and Mg 2+ are the dominant cations whereas HCOf
represents the dominant anion. The Ubangi shows a greater proportion of
Ca 2+ and HCO3. This can be related to a greater dissolution of calcium
carbonate in its basin, in which the geological substratum is mainly composed
of crystalline and metamorphic rocks which have been weathered in a thick
ferricrete mantle. Furthermore, the presence of paleocryptokarsts in the
Precambrian carbonate formations has been pointed out by Boulvert and
TABLE I
Mean annual values of the physico-chemical parameters for the Congo and the Ubangi rivers
Rivers
Year
Qa
pH
Cond
Si02
NUt
Na+
K+
Ca 2+
Mg2+
HCO;
SO~-
NO;
Cl-
TDS
Congo
1987
1988
1989
38708
39116
37741
38522
6.61
6.92
7.11
6.87
36.15
31.24
31.61
33.00
11.19
9.67
10.21
10.36
0.12
0.01
0.00
0.04
2.08
1.86
2.03
1.99
1.54
1.26
1.40
1.40
2.40
2.31
2.40
2.37
1.38
1.33
1.45
1.38
12.64
12.97
14.70
13.43
1.44
1.07
1.01
1.17
1.59
0.12
0.07
0.59
1.63
1.34
1.23
1.40
36.05
31.97
34.54
34.18
1988
1989
3152
2528
2840
7.22
7.36
7.29
39.02
36.94
37.99
13.21
13.18
13.19
0.00
0.00
0.00
2.57
1.58
2.08
1.78
1.38
1.58
3.23
3.38
3.30
1.42
1.47
1.44
19.47
18.58
19.03
0.80
0.78
0.79
0.66
1.27
0.96
0.86
0.76
0.81
44.02
42.41
43.22
= total inorganic dissolved solids
(mg l");
Mean
Ubangi"
Mean
For the Ubangi, the samples of September 1988 and March 1989 are lacking.
Qa = mean annual water discharge (m's"); Cond = conductivity at 20°C (pScm- I ) ; TDS
SiO z• cations and anions (mg l").
a
242
J.-L PROBST ET AL.
a - Cations
b - Anions
Ha'. K"
HC0i
on0o
/,\
Ca**
50
°oo
oo,
Mg+*
Cl-
50
Fig. 2. Ternary diagrams t-epresenting the ionic chemical composition of the Congo and the Ubangi waters.
The hatched and darkened zones represent the domains of monthly variations of (a) cationic and (b)
anionic composition.
Salomon (1988), which allows an explanation of the relatively high proportion
of Ca 2÷ and HCO; in the Ubangi waters, particularly during the low water
period. Considering the variations of ionic composition during the study
period, one can observe for both rivers that the cation composition is
relatively stable whereas the proportion of each anion varies remarkably for
the Congo (Fig. 2).
Taking all ions and silica together, the discharge-weighted mean of the
chemical composition expressed as weight percentages of the total inorganic
dissolved load is represented in Fig. 3. Two-thirds or more of the dissolved
load of the two rivers is composed of HCO.~ and SiO.,. The water composition
does not change from one year to another. There is also no change in dissolved
a - Congo river
b - l.lbangi river
~
Na"
D Mg=*
4%
Q
5%
2%
2%
2%
B SiO=
[] HCO="
~ NO3"
~ CI"
0
39%
Ca=.
SO-="
44%
Fig. 3. Chemical compositions of (a) the Congo (mean values of the years 1987-1989) and (b) the Ubangi
(mean values of the years 1988-1989), expressed as weigh / percentages of the total inorganic dissolved
solids.
DISSOLVED MAJOR ELEMENT EXPORT: CONGO AND UBANGI RIVERS
243
solids concentratien. For the Ubangi River, the chemical composition
includes only monthly samples for the years 1988 and 1989 because of lack of
data in 1987. However, the water compositions of the two rivers are not
significantly different (except for HCO3 ~ and Ca 2+ ).
The analytical results show an imbalance between cationic (Z ~ ) and
anionic ( Z - ) charges. The ionic balance (Bi, in per cent) is calculated by
Si=
and shows a deficit of anionic charges. The B~ averages 22% for the Congo
and 9.7% for the Ubangi. These missing anionic charges can be attributed to
the presence of organic anions to which our interest will be devoted in the
future. Indeed, the dissolved organic matter represents 30-40% of the total
dissolved solids of the Congo water.
TEMPORAL VARIATIONS OF WATER DISCHARGF AND PHYSICO-CHEMICAL
PARAMETERS
Variations of water discharge
The variations of the Congo discharge on seasonal and secular scales
(1902-1983) have been considered by some authors, including Molinier and
MBemba (1979), NKounkou and Probst (1987), Probst and Tardy (1987),
NKounkou (1989) and Martins and Probst (1991). The general trend for the
Congo discharge fluctuations is a significant increase during this century
(Pearson's correlation coefficient between annual discharge and time,
r = 0.27, whereas the Ubangi discharge variations show a decrease which is
not significant during the same period (r = - 0.12). The humidification of the
Congo basin noted in its discharge increase duririg this century has probably
affected only areas located south of the equator, whereas those located north
of the equator, mainly the Ubangi region, were drying out. On a seasonal scale
the Congo annual hydrograph shows two peaks resulting from the mixing of
equatorial and tropical hydrological regimes. The main peak (in December)
occurs two months after the unique peak (in October) of the Ubangi (see thin
curves on Figs. 4 and 5).
Variations of pH, conductivity, total dissolved solids ( TDS)
and silwa concentrations
Considering the pH and SiO2 variation curves for the Congo River (Fig. 4),
one can observe no significant evolution with respect to discharge variations.
However, the pH curve (laboratory measurement) shows a significant increase
244
J.-L. PROBSTET AL,
7.4.,H
h
I
^,.,_
7.0
3.0
"
N.+
"
.
'~4 'i I'ICO3"
2oi
AA
2.0
12,
8
1.0
0.5j
[
°
4
iY.
.~.
"ll"
,,$,,,,j8,
.....
0.4
-tAA
I
A
t
35
4~
°''
eol ~ s
--
/~
~..-~,; '25
o
J I I I I J Ii II a JIM J II II J MM J II N
3.o 1 I c ~ +
A
0.4.............................
2~
J I I i J | II J MM J S N J I I I I J | W
3.s l cr
,.
3.0
2.0
/
.
2.2
O
20 ~ T i . i . i . i . i . ~ i . i
_co,t,,cti,i~
I
.
so,
(~
u
i.e ,villivIjllilij,
2.1. MI2+
'
..~
1. §
VI~IV V V
V1
21
1.2
"i'i'i'i'i'i'i'i'i'i'i'i'i'i'iTi
1987 1988 1989
1987
1986 1909
435
5
l, • i i
m i
i,
a
o. i i .
a
i
m
1987 1988 1989
Fig, 4, Monthly variations of discharge (Ihin curves) and physico-chemical parameters (thick curves) for
the Congo River during the period 1987-1989.
with time (r = 0.65; significance level, S = 0.01%; number of observations,
N = 36) during the study period. The reasons for this increase still have to
be determined. For the Ubangi (Fig. 5), the pH (laboratory measurement)
also shows an increase with time (r = 0.61; S = 0.07%; N = 26) and varies
inversely with water discharge, whereas the variations of dissolved silica
concentration follow that of the discharge, but with a time lag of about three
months. For both rivers, water conductivity and the total dissolved solids
(TDS) vary inversely with discharge, that is to say when discharge increases,
TDS and conductivity decrease and vice versa. This phenomenon, common to
most of the world rivers, expresses a dilution ce *he river waters by rain and
surface waters and consequently a decrease ol" conductivity. So, conductivity
(Cond) is a linear function of TDS (TDS = SiO2 + cations + anions) and
DISSOLVEDMAJOR ELEMENTEXPORT: CONGOAND UBANGI RIVERS
245
381
7.8 p]B[
.9
4
.6
7.4
pH 7.0:
-3
2
-0
..........................
8.6 t'i'i'i'iTi'i'i'i'i"
141 sto2 ~
z
'i'i'i'i'i'i'i'i'i'i'i
1 i~ ~
i'i
A
zs. K+
,
12
"
1.8.
11
i.
1.4:
--
1.0
"i';'"
j
1.81
-.4
m
1.4
1"0 1
0.6
0.2
.........................
Slid
liJ
|ll
^.
"6
=0
"3
O
Lo ~'
JIIHJS
9
-1"
m
3.5
°.._ .....................
.... 0,4(i. i.k.i-i, i-i.i.k-;,.i.i, i
,6
J 40:
1.8
08
,
~
3
1.4
30 i*i'J'i Yi'i'"i'i'iTi'
1887
1980
1989
1'0 ~"*'Yi'i" F ;'k" Fi'i'i Ti
1087
198B
1989
I
1987
JEliJIII
1998
l|lJlild
0
1989
Fig. 5. Monthly variations of discharge (thin curves) and physico-chemica! parameters (thick curves) for
the Ubangi River during the period 1987-1989.
can be expressed by the following equations: for the Congo River
TDS (mgl- ~) =
1.02 x Cond (#S cm- ~ at 20°C)
(2)
(r = 0.92; S = 0.01%; N = 36)
and for the Ubangi River
TDS (mgl-~) =
1.08 x Cond (/~Scm- t at 20°C)
(3)
(r = 0.94; S = 0.01%; N = 26)
These equations can be useful to determine approximately in situ, the
concentration of the total mineral dissolved solids by measuring water
conductivity.
24,6
J.-L. PROBST ET AL.
Variations of cation and anion concentrations
For both rivers, the evolution of the concentration of major cations (Ca 2+,
Mg 2+, Na +, K +) during the study period appears clearly to be cyclic,
following the hydrological cycle. This is contrary to Moukolo et al. (1990),
who concluded in their study that during the period 1987-1988 there was no
relationship between dissolved element concentrations and discharge. The
concentration is at a minimum during high water discharge, increases during
falling river stage and reaches the maximum value at low discharge. For the
Congo, the lower concentration often occurs one month after the peak
discharge of December. This behaviour expresses a simple dilution by rain
waters of the cations released by rock weathering, i.e. no other process seems
to affect their concentration variation. The dilution of concentration at high
discharge is also observed for the major anions (HCOf, Cl-, SO42- and
NO~-), especially for HCOj- and CI-. Note that NO3 concentration often
equals zero and indicates values below the detection limit.
RELATIONSHIPS BETWEEN DISCHARGE AND CONCENTRATION
The dilution of dissolved substances at high water discharge can be
understood by considering the relationship between water discharge and the
concentration of dissolved substances. This allows the determination of the
type of dilution and the calculation of the mathematical expressions of these
relationships. The theoretical approach to understanding these relationships
has been investigated by Hall (1970, 1971), who proposed various models to
explain the evolution of the concentration of dissolved solids with respect to
water discharge and river input of elements.
Represented on Figs. 6 and 7 are plots of relationships for the parameters
which show the best correlation, i.e. TDS, Cond, Na +, K +, Ca 2+, Mg2+,
HCOj-, SO~- and CI-. In general, it appears that for the Ubangi, the points
are less scattered than those for the Congo, e'.,en if some data are lacking for
the Ubangi. This shows that the law which links river discharge and dissolved
matter concentration is simple. For the Congo, one can observe a little
scattering of the points, which indicates a more complex feature of the
relationships. Indeed, the Congo water at Brazzaville results from the mixture
of waters coming from three different hydrological regions: the equatorial
region, and the north (Ubangi) and south tropical regions. When the north
tropical region discharges little concentrated water, the south tropical region
supplies a high discharge of diluted water. This explains the disturbance of the
discharge--concentration relationship observed at Brazzaville. It also appears
for both rivers that the scattering of points is less marked for the cations than
247
DISSOLVED MAJOR ELEMENT EXPORT: CONGO AND UBANG! RIVERS
3.0| ',~
\ ~ HCO3"
K+
2 51 ~ , , ~ * ~
• /
* "~J~
2.0] • . - ~ ~ _
•
( ~ 19 87
225: ~o',,0,
® ~9o8
17.5
01989
~®~
7.5.
~0 38 ~:5 54 62 7025:22 3~0
3~o 3"8
®
s~2.75
Na +
252~
2.0.
®
®
2.25.
5"4 -(~2
7"0
~
1.75-
t.5
i6
®®
1.25.
E
1.0
Z
0
0,5'
O
3o
22
m
:3e 4s
54
62
I-,
<
I,Z
UJ
U
Z
O
0.750.25.
70 22
®
3o
3.5.
3.75-
®
®
46
54
62
i,
62 7o
70
el"
c# +
2.5
2.75
35
®0
="*~-k*
®
1.5
1.75
0.75 ,___.__.,__.___~___ .
22
30 38 45
54
~2
2.1
@
-. 0.5
70 22
60
3.0 3'~ 46
®
TDS
52'
1.7
44'
36"
1.3
--~,7"-
0.9
.5
28'
~"- ~
O - ' O m..._
20"
~
22
.
30
35
,
,=6 54. 52
WATER
7.0 12 22
DISCHA RGE
3"0 35 46
5~
i2
70
(103 m3.s "1)
Fig. 6. Congo River. RelationsMps between discharge (Q,) and concentration (C,) of dissolved substances.
The continuous curves represent the relation C~ = f(Qi); the dotted curves are the theoretical dilutiot,
J.-L. PROBST ET AL.
248
3
•
~®
o
®
24
2
®
@--------® @ @
@0'3
;®'"
'~O0
\
12
\
\
%,
\
0
0
"
0
5
3
S
9
m
<
'•'•
1
\
•
~\
~o
\\ 0
0.4
S
IZ
LM
9
0
~e~¢~o~
\
" ~~ ~
@
O
9
Cl"
2.0
®
@
o
@
" 1.5
\
@~
%
2.5
Ca:Z+
\
o
~ ~ ' ~ ' - - ® ' - - $ ® - s - e - - - - . - _ 0.8
0
4
®O
®
1.2
0
Z
0
SO4 2-
1.6
2
,
6
3
¢l
!
m
E
0
2.0
Na +
.
z
O
O 1987
O ) 1988
HCO3"
36 ¸
K+
1,0
®®
\
~
-O"
0.5
0
3
o --~
S
o
"9
3
6
9
6
i
Mg 2+
\
\
0
3
S
W A T E R
9
0
0
\\
20
\
o
D !S C H A R G E
3
(103 m3.s -1)
Fig, 7, Ubangi River. Relationships between discharge (Qi) and concentration (C~) of dissolved substances,
The continuous curves represent the relation Ci = f(Qi); the dotted curves a~e the theoretical dilution.
249
DISSOLVED MAJOR ELEMENT EXPORT: CONGO AND UBANG1 RIVERS
TABLE 2
Mathematical expressions adjusted to the relationships between instantaneous discharge (Qi,
103m 3s - ' ) and concentration (C~, mgl -~) of the dissolved substances
Rivers
Species
Equations
r
N
S(%)
Excluded months
Congo
Na +
K÷
Ca 2+
Mg 2+
HCOf
SO 2CITDS
Ci = 50.24 Qi-°'g87
C~ = - 0 . 0 1 8 Qi + 2.15
Ci =-- 50.09 Qi-I + 1.07
Ci = 43 Qi-i + 0.27
C~ = 555.01 Qi--1"016
Ci = 83.51 Qi --1"178
C~ = - 0 . 0 2 5 Qi + 2.40
Ci = 359.24 Q-0.u3
-0.70
-0.76
0.79
0.87
-0.87
-0.69
-0.61
-0.79
35
34
36
36
35
36
35
36
0.01
0o01
0.01
0.01
0.01
0.01
0.01
0.01
Jan. 87
Sep. 87, Jan. 88
Jan. 87
Apr. 88
-
Ubangi a
Na ÷
K÷
Ca 2+
Mg 2+
HCOjSO~CITDS
Ci =
C~ =
Ca =
Ca =
Ca =
Ci =
C~ =
Ci =
-0,93
-0.50
-0.93
-0.94
-0.95
-0.43
0.87
-0.95
25
25
26
26
25
26
26
24
0.01
1.03
0.01
0.01
0.01
2.53
0.01
0.01
Oct. 88
Oct. 88
Oct. 88
Aug, Oct. 88
- 0 . 3 4 8 lnQi + 2.08
- 0 . 2 0 9 InQ~ + i.64
4.397 Qi--°2s3
2 Qi-°'29°
25.53 Qi--°'3°9
- 0 . 1 6 9 InQ~ + i.05
0.428 Q ~ + 0.66
51.11 Q-O.185
For the Ubangi, the samples of September 1988 and March 1989 are lacking.
r = Pearson's correlation coefficient; N = number of observations; S = significance level.
for the anions. On the plots, one can sometimes see that there are one or two
points far from the others. Therefore these outliers were not taken into
account in the regression procedure. This applies to Na +, K +, HCO~- in both
Congo and Ubangi, Cl- for the Congo, and TDS for the Ubangi. These
excluded points are listed in Table 2.
The mathematical expressions of these relationships are graphically represented by the thick curves on the plots. Various functions were fitted (Table
2) and the best Pearson's correlation coefficients (r) were found for the
following relations:
a " Q~-I/.
(4)
Ci =
a ' Q i -I/" + b
(5)
Ci =
-a'lnQi
(6)
Ci
=
+ b
where Ci is concentration in mg 1- ' ; Q~ is water discharge in 103m 3s- I; and a,
b and n are constants. These relationships are also used to calculate the annual
dissolved fluxes exported by each river (next section).
Many authors fitted similar equations to the relation C~ = f(Q,) f:_,,
250
J.-L. PROBSTET AL.
dissolved material (Ledbetter and Gloyna, 1964; Hart et al., 1964; Ineson and
Downing, 1964; Gunnerson, 1967; Steele, 1968; Probst, 1983; Etchanchu,
1988; Kattan, 1989; Orange. 1990). This theoretical approach is based on the
following considerations.
(1) The total discharge (Qt) of the river results from the sum of the
discharges of the various sources. As a simplification, these sources are surface
water (Qs) and ground water (Qg):
Qt =
Qs + Qg
(7)
(2) There is mass conservation:
Ct'Q, = Cs'Qs + C~.Qg
(8)
where Ct, Cs and C~ ;~.~'erespectively the concentration in river water, surface
water and ground water.
One can consider now the dilution of the dissolved solids. At low water
discharge (Q~ = Qm~,), the concentration of the river water is maximum
(C~ = Cmax) and the dissolved flux is
Fd
=
Cmax" Qrain
(9)
If one dilutes this flux by an increasing river discharge (Qi) with an assumed
null concentration, the resulting evolution of the concentration (Cth) in river
water can be expressed as
Cth = (Cm~x"Qmin)/Qi
(10)
This theoretical dilution or 'dilution zero' is represented by the dotted
curves on the plots (Figs. 6 and 7), first proposed by Kattan and Probst (1986).
The comparison between the two curves (observed and theoretical) shows that
for the Congo, the evolution of the concentrations follows more or less the
theoretical dilution for most elements, even if the clusters of points are
relatively scattered owing to the contributions of the different hydrological
regions. On the contrary, for the Ubangi, the concentrations measured deviate
markedly from the theoretical dilution when the discharge increases.
However, when the discharge is higher than about 3000 m 3s-~, the two curves
tend to be parallel. Except for sulphate, chloride and potassium concentrations, the clusters of points are less scattered for the Ubangi than for the
Congo. These two different behaviours mean that for the Congo, the river
concentration is simply diluted by surface waters which supply no significant
amount of dissolved substances, and this dilution is only disturbed by the
contribution inputs of the different drainage areas. In contrast, for the
Ubangi, the surface waters supply significant inputs of dissolved substances to
the river, even if its concentration decreases. However, the discharge
DISSOLVED MAJOR ELEMENT EXPORT: CONGO AND UBANG1 RIVERS
251
variations are so important that, when the discharge increases, only dilution
affects the river concentr.ation behaviour.
ANNUAL DISSOLVED RIVER FLUXES
According to the literature, the export of dissolved minerals by the Congo
River varies between 35 x 106 and 50 x 106 tons year-~: 37 × 106 tons
year -I (Livi:gstone, 1963); 46.5 x 106 tons year -I (Symoens, 1968);
35.4 x 106 tons year -~ (Deronde and Symoens, 1980); 36.6 × 106 tons
year -t (NKounkou and Probst, 1987); 72 x 106 tons year -i (including
organic matter, Olivry et al., 1988); and 50 x 106 tons year-~ (NKounkou,
1989). The variability of the above estimations can be explained by the
temporal and spatial representativeness of the measurements used for flux
calculation (point or long-term measurements), and by the analytical techniques used to measure the dissolved elements. The variability of the
estimations is due also to the variability of the annual discharge and, lastly,
to the method used in the calculation of the annual flux. The dissolved fluxes
presented below are based on monthly measurements during the years 1987,
1988 and 1989. Because of lack of data, annual fluxes for the Ubangi were
calculated only for the years 1988 and 1989. Three methods were used to
calculate the annual fluxes: two stochastic methods (no. 1 and no. 2) and one
deterministic method (no. 3).
Method no. 1
In this method, the annual flux (E, in tons) is obtained by multiplying the
annual discharge-weighted mean of the concentration by the mean annual
water discharge, as follows:
Fa =
(C~'Q,)
]#, }
(Qi) " Q . ' k .
(11)
where C~and Q~ are respectively the instantaneous concentration (rag 1-~ ) and
water discharge (m3s -~) measured each month, Q, is the mean annual
discharge (m 3s -j ) and ka is a time correction factor (ka = 31.536).
Method no. 2
The annual dissolved flux results from an ex)rapoiation of the mean instantaneous flux to the entire year:
LLi= I
where Fa, C~, Qi and ka are as in method no. 1.
252
J.-L. PROBST ET AL.
M e t h o d no. 3
In this method, one first calculates theoretical mean monthly concentrations
of dissolved solids using the equations Ci = f ( O i ) calculated in Table 2. In
these equations, one substitutes Qi (the instantaneous water discharge) by the
mean monthly discharge (Ore) to obtain ltheoretical mean monthly concentrations (Cm) of dissolved substances which permit calculation of the annual
flux (Fd) as follows:
12
Fa
=
~
(Cm" am"
kin)
(13)
i=1
where Fd is the annual flux (tons), Qm is the mean monthly water discharge
(m 3s- ~), Cm is the mean monthly theoretical concentration (mgl- ~,
Cm = f(Qm)) and k m is a t:.me correction factor (kin = 2.628).
TABLE 3
Annual dissolved fluxes (106 tons calculated for the Congo and the Ubangi rivers using t.hree different
methods
Rivers
Year
Method
SiO2
Congo
1987
1987
1987
I
2
3
13.66 0.15
13,67 0.1~
-
2,53
2,54
2.38
1,88 2,93
!.88 2.93
1,73 2,8~
i,68
i,68
1,68
1988
1988
1988
I
2
3
11.63 0,01
11,92 0,01
-
2.29
2.29
2.39
1,56 2.85
1,56 2,85
1.73 2,89
1989
1989
1989
!
2
3
12,16 0.00
12,29 0.00
-
2.41
2,44
2,38
1
2
3
12.51 0.05
12.63 0.05
-
Ubangia 1988
1988
1988
I
2
3
1989
1989
1989
Mean
Mean
Mean
Mean
Mean
Mean
NH2
Na + K +
Ca z+ Mg 2+ H C O f
SO~-
NOr
CI-
15.43
15.44
16.51
1.76
!.76
1.3'7
1.95
1.95
-
1,99 44.01
2.00 44.03
1.69 4i.38
1,64
1.64
1,68
16,00
15.99
16,51
1.32
1,32
!.37
0,14
0,14
-
1.65 39.44
1,65 39,41
1,68 41,67
1.66 2,86
1.68 2,89
1.70 2.85
I.'72
1,74
1.67
17,49
17,68
16,52
!.20
1.21
1.38
0,08
0.08
-
1.46 41.11
1.48 41.55
1.67 41.16
2,42
2.42
2.38
1.70 2.88
i.71 2.89
1.72 2.87
1.68
!.69
1.68
16.32
16.37
16.51
1.42
1.43
1.38
0.72
0.72
-
1.70 41.54
1 . 7 1 41.67
1.68 41.69
1,31 0.00
1,06 0.00
-
0.25
0.20
0,15
0.17 0.32
0.14 0.26
0,13 0.30
0.14
0,11
0,13
!.93
1.57
1.63
0.07
0,06
0.07
0.06
0,05
-
0.08
0.06
0.07
4.22
3.43
3.87
1
2
3
1.05 0.00
1.04 0.00
-
0.12
0.12
0.13
0.11
0.11
0,11
0.26
0.26
0.26
0.11
0.11
0.11
1.48
1.47
1,42
0.06
0.06
0.06
0.10
0.10
-
0.06
0.06
0.06
3.38
3.36
3.27
!
2
3
!.18 0,00
1,05 0.00
-
0,18
0.16
OI4
0.14 0.29
0.12 0.26
0.12 0.28
0.12
0.11
0.12
1.70
1.52
!.53
0.07
0.06
0.07
0.08
0.07
-
0.07
0.06
0,07
3.':0
3.,~0
3.:~
For the Ubangi, the samples of September 1988 and March 1989 are lacking.
TDS = total inorganic dms,: e ~ solids.
TDS
253
DISSOLVED MAJOR ELEMENT EXPORT: CONGO AND UBANGi RIVERS
a - Congo r i v e r
b - U b a n g i river
DISSOLVED SUBSTANCES
Fig. 8. Annual fluxes of dissolved substances exported by (a) the Congo and (b) the Ubangi during the years
1987-1989.
As seen in Table 3, the results obtained by 'the three methods are not very
different. Among dissolved species, HCO3 and SiO2 show the greatest annaal
fluxes (Fig. 8). The TDS flux (without organic matter) amounts to 4045 X 106 tons year-~, of which about 10% comes from the Ubangi drainage
basin (Fig. 9). The contribution of the Ubangi region to the dissolved fluxes
of the Congo River is low as compared with its drainage area, but slightly high
.
.
. contribution
.
.
.
.
.
.
. (6-8°/.)
.
.
.
.tn. the
.
. Congo
.
.
. discharge.
.
.
.
~ j
relative to its water
For HC37
and Ca 2+, the contribution of the Ubangi is stil- higher. These ions, con~,lg
/
/
/
~
_
8
~[~
Fig. 9. Contributions of the Ubangi River (in ,r~ercent} to the annual discharge (Q,) and the dissolved fluxes
of the Congo River during the years 1988-~989.
254
J.-L, PROBST ET AL.
probably from the drainage of the paleocryptokarsts, raise the contribution to
a level higher than it would be without the carbonate dissolution. Indeed, the
thick ferricrete mantle which covers the Ubangi basin is depleted in mobile
elements. Furthermore, the Ubangi region is less watered than the remainder
of the Congo basin, with a specific discharge of 5.71s -t km -2 versus
12.81 s -~ km -2. The 'disturbing effect' of the presence of carbonate rocks in
the Ubangi drainage area has also been pointed out by Probst (1990b) who
estimated that 25% of the HCO~- originates from carbonate dissolution.
CONCLUSIONS
The main objective of the scientific programme which supports this work
is to follow the hydroclimatic fluctuations in the Congo basin and to analyse
their consequences for river transport, that is, to understand the response of
this second largest tropical-equatorial forest ecosystem to climatic variav:ons.
Maybe the coupling of the time hydrological records of the Congo discharge
and the study of the deposited materials in the adjacent Atlantic Ocean could
give some indications of the past. The data collected during the first three
years of this programme are not yet sufficient to discuss such a problem and,
moreover, the three years studied do not exhibit much hydrologic contrast.
However, some important results have been already obtained.
(1) The Congo and the Ubangi rivers have diluted waters that mainly
contain dissolved silica and bicarbonates and which always present anionic
deficit, probably owing to dissolved organic materials.
(2) The chemical compositions of the~e river waters are stable over time, i.e.
the contributions of the different anions and cations to the total anionic and
cationic charges do not vary over the hydrological cycle, whereas monthly
concentration variations of most dissolved elements are cyclical and vary
inversely with discharge.
(3) Good relationships between concentration and discharge could be
determined for most elements: for the Congo the concentrations are simply
diluted by river discharge, whereas, for the Ubangi, onc can detect a significant supply of elements from the draining surface waters.
(4) Three different methods used to calculate the total annual inorganic
dissolved fluxes exported by the Congo give similar values, ranging from
39 × 106 to 44 x 106 tons according to the year, of which about 10% is
supplied by the Ubangi (3.4 x 106 to 4.2 x 106 tons).
(5) The annual fluxes of dissolved silica and bicarbonates have been
estimated respectively at 11.6-13.7 x 106 and 15.4-17.5 x 106 tons for the
Congo, and at 1-1.3 x 106 and 1.5-1.9 x 106 tons for the Ubangi.
These early results also permit some preliminary observations. For the
DISSOLVED MAJOR
a -
1987
ELEMENT
EXPORT: CONGO
AND UBANG!
Congoriver
1988
IN9
YEARS
255
RIVERS
b -
• . .
Ulmagi river
1988
I~9
T.D.$.
YEARS
Fig. 10. Mean annual water discharge and fluxes of total inorganic dissolved solids exported by (a) the
Congo and (b) the Ubangi during the years 1987-1989.
Congo River, the flux of the TDS in 1987 was slightly higher than that of the
following years (Fig. 10), although the mean annual water discharge for this
year is medium. However, in 1987 the flux of NOr seemed to be abnormally
high relative to the two other years, which explains this high flux of TDS. On
the other hand, in 1988, the most humid year of the study period, the flox of
TDS was the lowest. For the Ubangi River, the two years of measurement
contrast more climatically than for the Congo, and the dissolved fluxes follow
the water discharge in the same proportion.
For the continuation of this programme, which is planned to run for at least
l0 years, the main question is "what will happen to the river transport, to the
eros;ion balance and the consumption of atmospheric CO2, to the global
biogeochemical cycle of each element, and finally to the respiration of this
forest ecosystem, when this large drainage basin is affected by a drought or a
very humid period?"
ACK N O W L E D G E M ENTS
This work was carried out with the support of INSU (Institut National des
Scie:aces de l'UnLivers),ORSTOM (Institut Fran~ais de Recherche Scientifique
pour le D6veloppement en Coop6ration) and CNRS (Centre National de la
Recherche Scientifique). "/he authors are grateful to the ORSTOM hydrologists of Brazzaville and Bangui, the Service des Voies Navigables du Congo
and the Laboratory of Water Chemical Analyses (Centre de G6ochimie de la
Surfa,ce, CNRS, Strasbourg). Contribution CNRS/INSU No. 4.
REFERENCES
Boulvert, Y. and Salomon, J.N., 1988. Sur l'existence de pal6o-crypto-karsts dans le bassin de
l'Oubangui (R6publique Centrafricaine). Karstologia, i 1/ i 2: 37-48.
256
J.-L. PROBST ET AL.
Bricquet, J.P., 1990. Le bassin du Congo: essai de bilan hydrog6ochimique. Compte-Rendu des
Journ6es Laboratoires ORSTOM, Centre ORSTOM de Bondy, 18-20 September 1990, pp.
1-23.
Clerfayt, A., 1955. Les eaux d'alimentation au Congo. Bull. Trimest. CEBEDEAU, 29/Iii:
186-197.
Deronde, L. and Symoens, J.J., 1980. L'exportation des 616ments dominants du bassin du fleuve
Zah'¢: une r66valuation. Ann. limnol., 16(2): 183-188.
Etchanchu, D., 1988. G6ochimie des eaux du bassin de la Garonne. Transferts de mati6res
disso~tes et particulaires vers l'oc6an Atlantique. Th6se Doe., Univ. Paul Sabatier de
Toulouse, 181 pp.
Figu6res, G., Martin, J.M. and Meybeck, M., 1978. Iron behaviour in the Zaire estuary. Neth.
J. Sea Res., 12(3/4): 329-337.
Giresse, P., Ouetiningue, R. and Bai'usseau, J.P, 1990. I~tude des fractions min6rales des
suspensions et des alluvions du Congo et de i'Oubangui. Sci. G~ol. Bull., Strasbourg,
43(2-4): 151-173.
Gunnerson, C.G., 1967. Strcamfiow and quality in the Columbia river basin. J. Sanit. Eng. Div.
Am. Soc. Cir. Eng., SA6: 1-16.
Hall, F.R., 1970. Dissolved solids-discharge relationships. !. Mixing models. Water Resour.
Res., 6(3): 845-850.
Hall, F.R., 1971. Dissolved solids-discharge relationships. 2. Applications to field data. Water
Resour. Res., 7(3): 591-601.
Hart, F.C., King, P.H. and Tchobanoglous, G., 1964. Predictive techniques for 0rater quality
inorganics. J. Sanit. Eng. Div., 90, SAS: 63-64.
lneson, J. and Downing, R.A., 1964. The ground-water component of river discharge and its
relationship to hydrogeology. J. Inst. Water Eng., 18: 519-541.
Jouanncau, J.M., Lapaquellerie, Y., Latouche, C. and Tastet, J.P., 1990. R6sultats pr61iminaires
de la campagne Oubangui-Congo de novembrc 1988. Microgranulom6trie, min6ralogie,
analy.~e,~chimiques des mati6res en suspension. Sci. G6ol. Bull., Strasbourg, 43(I): 3-14.
Kattan, Z., 1989. G6ochimi¢ ct hydrologic des eaux fluviales des bassins de la Moselle et de la
Mossig. Transports dissous et particulaires. Cycles biog6ochimiques des 616ments. Th6se
Doe., Univ. Louis Pasteur, Strasbourg, 220 pp.
Kattan, Z. and Probst J.L., 1986. Transport en suspension et en solution par la Moselle en
p6riode de crue. Actes desjourn6es d'hydrologie, "Crues et inondations", Strasbourg, i 6- i 8
October, 1986, pp. 14~-167.
Krempp, G., 1988. Techniques de pr616vement des eaux naturelles et des gaz associ6s. M6thodes
d'analyse des eaux et des roches. Notes techniques de l'Institut de G6ologie (nou,~elle
~dition), Univ. Louis Pasteur, Strasbourg, 19, 79 pp.
Ledbetter, J.O. and Gloyna, E.F., 1964, Predictive techniques for water quality inorganics. J.
Sanit. Eng, Div., SA i: 127-151,
Livingstone, D.A., 1963, Chemical composition of rivers and lakes. U.S. Geol. Surv., Prof.
Pap., 440G, 64 pp.
Mariotti, A., Gadel, F., Gircsse, P. and Kinga-Mouzeo, 1991. Carbon isetope composition and
geochemistry of particulate organic matter in the Congo river (Central Africa): application
to the study of Quaternary sediments off the mouth of the river. Chem. Geol. (lsot. Geosci.
Sect.), 86: 345-357.
Martins, O. and Probst, J.L,, 1991. Biogeochcmistry of major African rivers. Carbon and
mineral transport. In: E.T. Degcns, S. Kemp¢ and J.E. Richey (Editors), Biogeochemistry
of Major World Rivers. Wiley, Chichester, pp. 129-i57.
DISSOLVED MAJOR ELEMENT EXPORT: CONGO AND UBANG! RIVERS
257
Meybeck, M., 1978. Note on elemental contents of the Zaire River. Neth. J. Sea Res., 12(3/4):
293-295.
Molinier, M., 1979. Note sur les d6bits et la qualit6 des eaux du Congo d Brazzaville. Cabiers
ORSTOM, S~r. Hydrol., XVI(i): 55-66.
Molinier, M. and MBemba, E., 1979. The Congo-Zaire river basin. The great rivers of the
world, part 2. Water Quai. Bull., 4:83-85 and 97.
Moukolo, N., Bricquet, J.P. and Biyedi, J., 1990. Bilans et variations des exportations de
mati6res sur ie Congo d Brazzaville. Hydrol. Continent., 5(I): 41-52.
NKounkou, R.R., 1989. Hydrog6odynamique actuelle du Congo et de l'Amazone. Cycle global
de reau et bilan de I'~rosion au cours des temps phan6rozoiques (derniers 600 millions
d'ann~s). Th6se Doc., Univ. Louis Pasteur, Strasbourg, 193 pp.
NKounkou, R.R. and Probst, J.L., 1987. Hydrology and geochemistry of the Congo River
system. In: E.T. Degens, S. Kempe and G. Weibin (Editors), Transport of Carbon and
Minerals in Major World Rivers, Part 4. Mitt. Geol-Pal~.ont. Inst. Univ. Hamburg, SCOPE/
UNEP Sond., 64: 483-508.
NKounkou, R.R., Krempp, G. and Probst, J.L,, 1990. G~ochimie et nydrologie des eaux de
surface: exemple du bassin du fleuve Congo. Compte-Rendu des Journ6es Laboratoires
ORSTOM, Cez,',re ORSTOM de Bondy, 18-20 September 1990, pp. 25-36.
Olivry, J.P., Bricquet, J.P., Thi6baux, J.P. and NKamdjou, S., 1988. Transports de mati6res sur
les grands fleuves des r6gions intertropicales" les premiers rt~sultats des mesures de flux
particulaires sur le bassin du fleuve Congo. Internat. Assoc. Hydrol. Sci., Pub., 174: 509-521.
Orange, D., 1~90. Hydroclimatologie du Fouta Djalon et dynamique actuelle d'un vieux
paysage lat6ritique. Th6se Doc., Univ. Louis Pasteur, Strasbourg, 220 pp.
Probst, J.L., 1983. Hydrologie du bassin de la Garonne. Mod61es de mt~langes. Bilan de
r6rosion. Exportation des phosphates et des nitrates. Tht~se Doc. 3e cycle, Univ. Toulouse
Ill, 148 pp.
Probst, J.L., 1990a. Transports de mati6res min6rales dissoutes et particulaires et bilans de
1'6rosion sur les grands bassins fluviaux intertropicaux p6riatlantiques. Rapport de fin de
Programme PIRAT (INSU-ORSTOM), Op6ration Grands Bassins Fiuviaux, Th~me
Mati6res Min6rales (1986-1991), Strasbourg, March 1991, 20 pp.
Probst, J.L., 1990b. G6ochimie et hydrologie de r6rosion continentale. Mccanismes, bilan
global actuel et fluctuations au cours des 500 derniers millions d'ann6es. T!,6se Doc. d'Etat,
Univ. Louis Pasteur, Strasbourg, Vol. I (text), 185 pp., Vol. 2 (publications), 393 pp.
Probst, J.L. and Tardy, Y., 1987. Long range streamflow and world continental runoff fluctuations since the beginning of this century. J. Hydrol., 94:289-31 i.
Probst, J.L., Amiotte-Suchet, Ph. and Tardy, Y. 1992. Global continental erosion and fluctuations of atmospheric CO2 consumed during the last 100 years. Proc. 7th Int. ,qymp.
Water-Rock Interaction (WR!-7). A,A. Balkema, Park City, USA, in press~
Stallard, R.F. and Edmond, J.M., 1983. Geochemistry of the Amazon. 2. Influence of geology
and weathering environment on dissolved load. J. Geophys. ges., 88(C14): 967!-9688.
Steele, F.D., 1968. Digital-computer applications in chemical quality studies of surface water
in a small watershed, h Assoc. Hydrol. Sci., Pub., 80: 203-214.
Symoens, J.J,, 1968. La mineralisation des eaux natureiles. Hydrobiological survey of lake
Bangweulu and Luapula river basin. Cercle Hydrobiol. Bruxelles, 2(!): !-199.
Van Bennekom, A.J., Berger, G.W., Helder, W. and De Vries, R.T.P., 1978. Nutrient distribution in the Zaire estuary and river plume. Neth. J. Sea Res., 12(3/4): 296-323.