Available water of a ferrallitic soil along the Kossou

Hydrological Sciences -Journal- des Sciences Hydrologiques,39,2, April 1994
Available water of a ferrallitic soil along
the Kossou Lakeside slope (Ivory Coast) as
affected by soil texture
N. R. YAO
Laboratory of Bioclimatology, ENSA, 08 BP 35 Abidjan 08, Côte d'Ivoire
O. AMADOU
Laboratory of Agricultural Engineering, ENSA, 08 BP 35 Abidjan 08, Côte
d'Ivoire
Abstract The neutron probe method for in situ measurement of field
capacity was applied to ferrallitic soils on the Kossou Lakeside slope
(Ivory coast). For better utilisation of the natural resources, a study was
conducted to determine the influence of the lake on the availability of soil
water to plants along the lakeside slope. Two methods, one gravimetric
and the other neutronic, were used to evaluate soil water content and its
variation with time. Five access tubes were placed along the slope and
were monitored during the study. The infiltration rate determined at three
of the measuring sites was higher in the top soil layers than in the deeper
soil layers. Drainage and redistribution data show a juxtaposition of two
infiltration kinetics, the first being exponential and the second linear. The
extrapolation of the second kinetic (slow) to the start of drainage gave,
for the first 600 mm of soil, field capacities of 154, 184 and 163 mm for
sites T l , T3 and T5 respectively. These differences were specifically
related to the porosity of the sites under study. An analysis of the soil
water profiles showed that the soil moisture near the lake was always
higher than the permanent wilting point. The influence of the lake was
not apparent during the long rainy season because of the steady decline
of the water table. The water availability then depended upon the rainfall
distribution and the specific characteristics of each of the sites, especially
soil texture and structure. A linear relationship was found between soil
moisture estimated by the neutronic and the gravimetric methods with
correlation coefficients ranging from 0.80 to 0.96. The data also showed
that the neutron probe method did not systematically underestimate the
soil moisture of the top layers. This response seemed to be associated
with the fraction of coarse sand. As soil moisture increased, data from
the neutron probe method converged toward the gravimetric data
regardless of the site.
L'eau disponible d'un sol ferrallitique le long de la
toposéquence du lac de Kossou (Côte d'Ivoire) en relation
avec la texture du sol
Résumé La méthode neutronique pour la détermination in situ de la
capacité au champ a été appliquée aux sols le long d'une toposéquence
du lac de Kossou. Dans le but de mieux exploiter nos ressources
naturelles, une étude a été menée pour déterminer l'influence de la
présence du lac sur la disponibilité de l'eau pour les cultures le long de
la toposéquence. Deux méthodes, l'une gravimétrique, l'autre
neutronique, ont été utilisées pour la détermination de la teneur en eau
Open for discussion until 1 October 1994
95
N. R. Yao & O. Amadou
du sol et de sa variation au cours du temps. Cinq tubes d'accès installés
le long de la toposéquence ont été suivis durant l'étude. La vitesse
d'infiltration déterminée sur trois des sites de mesure a été plus élevée
en surface qu'en profondeur. Les résultats de drainage et de ressuyage
donnent une juxtaposition de deux cinétiques d'infiltration, l'une
exponentielle, l'autre linéaire. L'extrapolation de la deuxième cinétique
(lente) au début du drainage donne pour les 600 premiers mm du sol des
capacités au champ de 154, 184 et 163 mm respectivement pour les sites
T l , T3 et T5. Ces différences sont notamment liées à la porosité des
différents sites étudiés. L'analyse des profils hydriques montre que la
teneur en eau du sol proche du lac a toujours été supérieure au point de
flétrissement permanent. L'importance du lac n'est pas évidente pendant
la grande saison des pluies à cause de la baisse constante du niveau
d'eau. Pendant cette période, la disponibilité en eau dépend de la
distribution des pluies et des caractéristiques intrinsèques de chacun des
sites, notamment de la texture et de la structure du sol. Une relation
linéaire a été mise en évidence entre les stocks d'eau déterminés par les
méthodes neutronique et gravimétrique avec des coefficients de
corrélation variant entre 0.80 et 0.96. Les résultats montrent aussi que
la sonde à neutrons ne sous-estime pas systématiquement les teneurs en
eau de surface. Ce comportement semble être lié au pourcentage de sable
grossier. Si le sol devient de plus en plus humide, alors les mesures
neutroniques convergent vers les mesures gravimétriques quel que soit le
site.
INTRODUCTION
For agronomie purposes, field capacity is defined as the water content found
when a thoroughly wetted soil has drained for about two days (Marshall &
Holmes, 1979). In other words, the field capacity is the water content when the
drainage rate is considered to be negligible. Field capacity then depends only
on the arbitrary choice of a negligible drainage rate and the hydraulic
properties of the profile (Campbell, 1985). This term lacks critical physical
definition, as stated by Marshall & Holmes (1979), but it has an established
utility especially for the estimation of irrigation requirements. The
determination of field capacity in the laboratory is subject to certain difficulties.
In fact, the soil structure is generally destroyed during the experiment; the
suction pressure which should be applied to the soil sample is known but
remains empirical. As a result several research scientists have proposed simple
techniques allowing a quick and precise measurement of the soil field capacity
(Féodoroff, 1965; Dancette, 1970; Dancette & Maertens, 1974; Marcesse &
Couchât, 1974; Puard et al., 1980).
The determination of soil water content is essential for the understanding
of the chemical, mechanical and hydrological responses of soils and for the
study of plant growth (HUM, 1974). Destructive and non-destructive methods
exist for measuring soil moisture (Hénin, 1977). The objectives of the present
study were: (a) to compare surface soil moisture values obtained from
gravimetric and neutronic methods; and (b) to determine the impact of the
Kossou Lake on the soil moisture distribution along the lakeside slope.
97
Available water of a ferrallitic soil
MATERIAL AND METHODS
Site
The experimental fields studied were sown in 1985 and 1986 on the Kossou
Lakeside slope near Bouaflé (Central Ivory Coast) at 6°59'N, 5°45'W and at
an altitude of 187 m. One half of the fields were sown with maize and the
other half with rice. The average annual rainfall at the site is 1350 mm with a
potential evaporation of 1700 mm giving an annual water deficit of 350 mm.
The rainfall distribution is bimodal resulting in two rainy seasons from March
to July and from September to November. The soils are ferrallitic, slightly
desaturated, impoverished and with hardpan. The upper horizons have a sandy
texture while the lower ones have sand-clayey or loam-sand-clayey texture with
massive structure.
Experimental design
Five access tubes were placed along the Kossou Lakeside slope with a 35 m
spacing between tubes and 30 m between the lake and the first access tube
(Fig. 1). The length of the access tube ranged from 1000 mm to 1200 mm
depending upon the position of the hardpan within the lower soil layers.
O
HII
Access tubes
Horizon A11
|T[Q Horizon A12
[fjJJI Horizon A12g
^ ^
Horizon A3
USB Horizon AB
E H ] Horizon B
B U
Horizon B
E 3
Horizon B
V
,-—--"*""
>4^)5?TT—|-
Ë S 2 Horizon Bfe
S 3
Horizon 8
J
F
M
A
M
J
Months
J
A
S
O
N
of year
Fig. 1 Kossou lake: implantation and representation of the soil
profiles, variation of the water table in 1985 (closed line) and
1986 (dashed line).
N. R. Yao & O. Amadou
98
Infiltration measurements were conducted on three of the five access
tubes (Tl, T3, T5) from 28 to 31 January 1987, during the dry season on a dry
and good drainage soil. The method consisted of monitoring the infiltration and
redistribution kinetics using a neutron probe (Solo 25, Nardeux France). At
each site, the access tube was located in the centre of a double ring
infiltrometer with infiltrating surfaces of 1 and 4 m2 respectively (Vachaud
et al., 1978). The two rings werefilledwith water to 250 mm. The measuring
site was not protected against evaporation, leading, therefore, to a slight underestimation of infiltration.
Soil moisture was measured at all five tubes (Tl, T2, T3, T4, T5),
starting from the bottom of the access tube, with measurements every 100 mm
upward to 100 mm from the surface using a reflector. Measurements were
made once a week during the four months of the experimental period. The data
were used to calculate volumetric soil water content based on calibration
curves. The purpose of making measurements in the top soil layer is to
determine whether soil sampling after calibration can be avoided by simply
applying the calibration coefficients to neutron data from top soil. A field
method was used for the calibration of the neutron probe using neutronic
measurements, soil bulk density and gravimetric moisture of the soil samples.
Calibration After installation of the access tubes, gravimetric soil
samples and neutron probe readings were taken simultaneously every 100 mm
over the whole length of each tube as a basis for generating an appropriate
calibration curve. Three replications of soil samples were taken within 450 mm
around each access tube. Then during the monitoring period soil samples were
also taken at three depths (0-100, 100-200, 200-300 mm) with three
replications within 2 m around each access tube, once a week, giving a total
of 45 samples (3 x 3 x 5). The volumetric soil moisture (Hv, %) was then
calculated using the following formula:
Hv = HpxPs
(1)
where Hp = gravimetric soil moisture; and ps = bulk density.
The following formula was, however, used to determine volumetric soil
moisture at wilting point based on laboratory data:
Hv = HpcXps
(2)
where:
Hpc = (HpexFe) + (HprXFr)
(3)
with Hpr = gravimetric moisture of coarse material fraction; Hpe = gravimetric
moisture of fine material fraction; Fr = fraction of coarse soil material; and Fe
= (1 - Fr) = fraction of fine soil material.
The soil bulk density was measured using a gamma-densimeter (DR 18),
removing a 200 mm soil layer after each measurement.
99
Available water of a ferrallitic soil
RESULTS AND DISCUSSION
Soil physical characteristics
The soil texture was mainly sandy with fine or coarse sands (Table 1). The fine
sands were dominant in the humus and humus penetration horizons. Coarse
sands were abundant in the B horizons. In most cases, the fraction of clay and
loam was about 30% of the total. The bulk density varied from 1.2 at the
surface to 1.7 in lower layers (Table 2). There was, however, a sharp
difference between the 0-200 mm horizon (1.2-1.3) and those underneath (1.61.7).
Table 1 Soil physical analysis at sites Tl, T3, and T5
Depth
(mm)
400-700
T3
0-200
200-400
400-700
600-900
ftuticle size (%):
Water content {%):
Density
(f)
Clay
Loam
12.8
15.3
16.2
13.3
14.5
13.3
63.8
52.4
49
5.2
16.1
19.4
6.3
7.3
7.6
16.5
15.5
14.6
1.3
.3
1.6
6
1.7
50
3.9
39
35.8
14.3
16.5
17.7
13.5
10.4
8.2
53.3
35.1
26.1
15.7
34.2
45.3
7.4
9.1
9.:
D.3
10/.
16.6
15.8
16.3
1.2
1.7
1.6
54.7
35.8
39.6
22.3
37
35.1
38.8
8.3
11.5
13.6
17.3
18 9
17.1
18.6
23.2
1.3
1.7
1.7
1.7
50.9
35.8
35.8
35.8
27.7
11
10.1
9.6
10.8
Fine sand Coarse sandl.53 MPs 0.03 MPs
17.9
The total porosity calculated from soil density was poorly developed
(n < 50%) except in the top layers (Table 1). The high porosity recorded at
the surface can be explained by tillage on one hand and by the presence of
small holes caused by the soil fauna (Hénin, 1976). Soil moisture content at the
wilting point was higher in the lower than in the upper layers. This was mainly
due to the higher clay content in the lower horizons. However, it was lower
than what would have been obtained using the Van Wambeke (1974) equation
which considers the water content at -1.5 MPa and the clay content. The
available water declined with depth.
Table 2 Soil bulk density
Sites
Tl
T2
T3
T4
T5
Depth (mm)
0-2T»
1.3
1.2
1.2
1.2
1.3
200-400
1.7
1.6
1.7
1.7
1.7
400-600
600-800
800-100)
1.6
1.7
1.6
1.7
1.7
1.9
1.9
1.6
1.8
1.7
1.6
1.8
1.7
Figure 1 shows the evolution of the water level of the lake in 1985 and
part of 1986. As can be seen, the water table declined from December to July
100
N. R. Yao & O. Amadou
and rose from August to November as a consequence of the rainy season
upstream of the Bandama River in northern Ivory Coast and the Sahel.
Figure 1 also demonstrates that sites Tl to T3 were at one time under water
from November 1985 to early March 1986. However, the decline of the water
table during the major growth season (April to July) led to a lesser influence
of the lake on the crop water consumption, especially for the sites further from
the lake shore. This result is evidence that crop growth from March to July
along the lakeside slope mainly depended on the local rainfall and its
distribution. A stronger influence is, however, expected during the short rainy
season (September to November) with a high risk of flooding and yield losses
as was the case in 1985.
Infiltration
Figure 2 shows the water content profiles obtained during the infiltration test.
The results show that water rapidly infiltrated to the soil layer at 200 mm depth
before accumulating at the surface. The difference between the dry water
profile and the first profile of the infiltration test (Fig. 2) is clear evidence of
that situation. Figure 2 also shows that the soil moisture at 200 mm depth
quickly increased at the start of the test reaching that of the 100 mm depth
before any substantial increase in the latter. This result can be explained, on
the one hand, by the low soil moisture content at the 200 mm depth before the
infiltration test and, on the other hand, by the difference in soil structure above
and below the 200 mm depth (Féodoroff, 1965). As a matter of fact, the sharp
difference in the soil bulk density and the texture between the 0-200 mm layer
and those underneath, and the relatively high porosity of the top layer (Table 1)
can explain the apparent difference in infiltration rates between the top and the
bottom layers. As a consequence, water preferentially accumulated in the top
layers.
The decline of the infiltration rate with depth was confirmed in Fig. 3.
There was a reduction in the slope of the moisture curves (2.8% min"1 at
100 mm, 1.5% min"1 at 200 mm, 0.7% min"1 at 300 mm, 0.6% min"1 at
500 mm and 0.5% min"1 at 700 mm) with depth. Soil tillage to 200 mm depth
prior to sowing had favoured infiltration in the top soil layer (Féodoroff,
1965). Figure 2 demonstrates that moisture content at saturation was equivalent
to total porosity (33 to 37%) in the horizons below 200 mm depth; however,
this equivalence was not confirmed for the 0-200 mm layer probably because
of calibration error due to neutron loss from the soil despite the use of a
reflector.
Internal drainage
Soil water profiles during evaporation-free drainage are presented in Fig. 4.
The sharp difference between the dry profiles and those at saturation in the top
layer (0-200 mm) indicates that the soil porosity was greater in the top layer
101
Available water ofa ferrallitic soil
than in the bottom layers in relation to tillage. However, those top layers
quickly lost water because of their low clay content. Within 24 h from the start
of drainage, the top layer had lost 28, 55 and 34% of its water, respectively,
for sites Tl, T3 and T5. The differences in soil water content between the three
sites were mainly due to the porosity.
Volumetric water content (Hv % )
2
10
18
26
34
42
50
58
I
L
I
I
1
I
I
J
Fig. 2 Moisture profiles during an infiltration test on ferrallitic
soil at site T3; Ps = dry profile; numbers indicate time (min)from
start.
An analysis of the evaporation-free drainage revealed two kinetics (Figs 5
and 6). The first one is exponential (Figs 5(b) and 6(b)) and related to rapid
flow by gravity. A semilog paper used to plot the data shows a linear
relationship between soil moisture and time at the early stage of drainage.
Therefore the first kinetic is effectively exponential. The second kinetic, of
linear form (Figs 5(a) and 6(a)), is slower and related to capillary forces. These
results partly confirm those of Marcesse & Couchât (1974) who stated that the
two kinetics were exponential.
102
N. R. Yao & O. Amadou
Time (mn )
Fig. 3 Evolution of soil moisture at different depths (100, 200,
300, 500 and 700 mm) during an infiltration test at site T3.
Volumetric
water content
( Hv % !
Fig. 4 Moisture profiles during aferrallitic soil drainage at sites
Tl, T3 and T5; Ps = dry profile; numbers indicate time (min)
from start of drainage.
The similarity of the slopes of the lines during both the fast and slow
kinetics (Figs 6(a) and 6(b)) demonstrates that the drainage rate through the
0-600 mm soil layer was identical for the three sites. However, the drainage
rate through the 0-300 mm soil layer was different between the sites (Figs 5(a)
Available water of a ferrallitic soil
Field capacity
n =so
So - 3 0 c m )
T3=83
Time
(hours)
f r o m t h e s t a r t of
103
drainage
Fig. 5 Moisture change in the 0-300 mm soil layer during
drainage at sites Tl, T3 and T5; (a) regular paper; (b) similog
paper.
Fied capacity
(0-60cm)
T I : 154mm
T3=
184mm
TS= 163mm
16
20
Time (hours) from
300p.
24
the
28
s t a r t of
32
drainage
Fig. 6 Moisture change in the 0-600 mm soil layer during
drainage at sites Tl, T3 and T5; (a) regular paper; (b) semilog
paper.
and 5(b)). Extrapolation of the slow kinetic to the start yielded field capacities
of 80, 88 and 81 mm for the 0-300 mm soil layer; 74, 96 and 82 mm for the
300-600 mm layer and a total of 154, 184 and 163 mm for the 0-600 mm layer
respectively for sites Tl, T3 and T5. This water retention capacity is
considered as the soil moisture content at the start of the slow kinetic. From the
agronomic viewpoint, it is the upper limit of the water stored in the soil and
usable by the plants (Marcesse, 1967). The differences between sites were
mainly associated with soil porosity.
104
N. R. Yao & O. Amadou
Water profiles
Figure 7 shows the water profiles recorded at the experimental site T4. The
soil moisture between 0 and 300 mm depth was calculated using both the
neutronic and gravimetric methods. The volumetric soil water content at depths
beneath 300 mm was calculated using the neutronic method. Soil water content
at -1.5 MPa (Table 3) was calculated using equations (2) and (3). The
gravimetric humidity of the coarse fraction was estimated to be 6.5% at
-1.5 MPa based on work by Boa (1983). Figure 7 shows clearly the
importance of the correction made in soil moisture calculations based on data
from the fine soil fraction (less than 2 mm) and taking into account the
percentage of coarse material.
In general, the volumetric humidity at -1.5 MPa was very low (less
than 12%) in the top layers and increased with depth in relation to the higher
clay content. The 200-400 mm soil layer acted like a pool that was emptied
only when there was a long dry period (this was the case on day 91 after
Volumetric water content
10
—1
14
1
18
1
22
I
( Hv % )
26
I
30
l_
\
-1.5MPa
non corrected
Fig. 7 Soil moisture profiles at site T4 in 1986: the numbers
indicate the days after sowing.
105
Available water of a ferrallitic soil
Table 3 (a) Fraction of coarse material; (b) water content at 1.5 MPa; and (c)
corrected water content at 1.5 MPa
Site
Tl
T2
T3
T4
T5
Depth (cm):
0-10
a
b
c
0.6
0.9
0.1
0.2
3.4
8.1
9.4
8.1
9.1
9.6
8.1
9.4
8.1
9.1
9.5
Depth (cm):
50-60
a
b
c
Tl
T2
T3
T4
T5
47.3 12.3 9.6
53.6 22.8 14.1
33.9 16.4 13.0
49.5 28.4 17.6
47.3 24.9 16.2
10-20
a
b
20-30
a
b
c
30-40
a
b
c
0.6 8.1 8.1
0.9 9.4 9.4
10.5
5.8 9.0 8.7
21.2 10.6 10.4
12.1 10.8
11.2
8.2
10.5
11.5
38.9
60-7C
b
a
70-80
a
b
c
62.7
74.7
14.5
19.1
9.5
9.7
49.5
56
30.1 18.4
30.1 16.9
47.3
53.6
33.9
49.5
56
c
14.5
24.7
16.2
30.1
30.1
10.7
14.9
12.9
18.4
16.9
40-50
a
b
c
11.8 11.2 11.2 11.8 11.2 47.3 12.3
15.4 14.7 53.6 20.1 12.8 53.6 22.8
15.7 14.7 33.9 16.6 13.2 33.9 16.4
16.2 15.1 22.1 18.6 15.9 32.6 20.3
17.3 13.1 47.3 22.4 14.9 47.3 23.7
80-90
a
b
c
49.5 30.4
56
28.2
18.6
16
c
9.6
Î4.1
13.0
15.8
15.6
sowing). It was generally observed that the soil moisture near the lake was
always higher than the wilting point (the water table being closer to the soil
surface) (Fig. 1). The total amount of soil water in the first 600 mm varied
from 150 to 97 mm (Fig. 8). There were sites with high water content (T2, T3
and T5) and sites with low water content (Tl and T4). However, on the basis
of the available water in the first 600 mm of soil, it was found that site T4
consistently had less available water than the other four sites. This result is
explained by both the high soil moisture (10 to 22%) at wilting point and the
low field capacity of site T4. The low filling of site T4 between days 40 and
50 after sowing is an indication that its field capacity was low.
10
20
30
40
50
60
70
80
90
100
Days after sowing
Fig. 8 Evolution of soil moisture (dashed line) and available
water (solid line) at the five sites for the first 600 mm of soil in
1986.
N. R. Yao & O. Amadou
106
A linear relationship was found between neutronic and gravimetric water
measurements in the top soil layer as shown in Fig. 9 for the top 300 mm. The
correlation coefficients in every case ranged from 0.80 to 0.96. This good
correlation implies that the surface water contents determined by both methods
had similar trends. However, each site should be calibrated, as is shown by the
spatial heterogeneity of the sites studied which were only 35 m apart. The
results demonstrate that the neutron probe method does not systematically
underestimate the top soil moisture content. This is partly associated with
variations in soil density and texture. In point of fact, Couchât (1983) has
shown that the neutron probe response depends upon soil bulk density; he also
reported a strong influence of the sand fraction on the calibration curve.
20
30
40
50
60
70
80
Neutronic water content (mm I
Fig. 9 Relationship between gravimetric and neutronic water
contents of the first 300 mm of soil at sites Tl and 75.
Particle size analysis showed that neutronic soil water measurements
were more or less associated with the fraction of coarse material and
particularly coarse sand. The fraction (%) of coarse sand was 10.6, 15.2,15.7,
23.5 and 26.2 respectively for sites Tl, T2, T3, T4 and T5. It should be
pointed out that the two sites with the highest fraction of coarse sand were the
ones where soil moisture was overestimated by the neutronic probe method. As
Available water of a ferrallltic soil
107
the soil moisture increased, however, the neutronic and the gravimetric
measurements converged. These results suggest that soil texture should be
integrated carefully in the calibration process and that the error margin on
neutronic water measurements gets higher as the soils are drier.
CONCLUSIONS
The infiltration rate was higher in the top soil layer due to the porosity. The
water accumulated in the first 200 mm of soil before redistributing.
There were two kinetics during drainage: one fast and of exponential
form and a slow second one of linear form. Differences in drainage rate were
mainly recorded in the top soil layers.
The field capacity determined using the neutron probe method was more
closely associated with soil porosity.
The effect of the lake was not evident during the major growth period
because of the water table decline associated with die delay in the rainy season
upstream. Under favourable rainfall conditions, soil water availability along the
lakeside slope depended upon the physical characteristics (texture and structure)
at each site. The study also revealed the importance of soil texture on neutronic
water measurements. An appropriate calibration is necessary for each site.
REFERENCES
Boa, D. (1983) Caractéristiques hydriques des gravillons ferrugineux dans les sols ferrallitiques.
Laboratoire de Pédologie ORSIÛM, Adiopodoumé, 16-33.
Campbell,G. S. (1985) Infiltration and redistribution. In: Soil Physics with Basic Transport Modelsfor
Soil-Plant Systems, 73-97. Elsevier, Amsterdam, The Netherlands.
Couchât, Ph. (1983) Les applications de la méthode neutronique dans la recherche agronomique.
Colloque International sur l'Emploi des Techniques des Isotopes et des Rayonnements dans les
Études sur la Physique des Sols et l'Irrigation, Aix-en-Provence, France, 18-22 April. IAEASM-267/42.
Dancette, C. (1970) Détermination au champ de la capacité de retention après irrigation dans un sol
sableux du Sénégal. Agron. Trop. 25,225-240.
Dancette, C&Maertens, C. (1974) Méthode d'estimation de la capacité au champ pour l'eau à partir
du pF 3. Bull. Assoc. Française pour l'Etude du Soli, 165-171.
Féodoroff, A. (1965) Etude expérimentale de l'infiltration de l'eau non saturante: cas d'un sol
initialement sec et d'un arrosage sans formation de plan d'eau en surface. Ann. Agron. 16, 231263.
Hillel, D. (1974) L'eau et le Sol, Principe et Processus Physiques, 75-85. Vander, Leuven, Belgium.
Hénin, S. (1976) Cours de Physique du Sol. I: Texture-Structure Aeratin. Edn. ORSTOM Paris,
EDITEST, Bruxelles.
Hénin, S. (1977) Cours de Physique du Sol. II: L'Eau et le Sol: les Propriétés Mécaniques. EDITEST,
Bruxelles.
Marcesse, J. (1967) Détermination in situ e la capacité de rétention d'un sol au moyen de l'humidimètre
à neutrons. In: Isotope and Radiation Techniques in Soil Physics and Irrigation Studies, 137146. IAEA, Vienna, Austria.
Marcesse, J. & Couchât, Ph. (1974) Etude hydrodynamique des sols à l'aide d'un humidimètre à
neutrons automatique. In: Isotope Techniques in Groundwater Hydrology, 277-280.IAEA,
Marshall, T. J.'& Holmes, J. W. (1979) Composition of soil. In: Soil Physics, 1-31. Cambridge
University Press, UK.
108
N. R. Yao & O. Amadou
Puard, M., Couchât, Ph. & Moutonnet, P. (1980) Application de la méthode gammaneutroniquc à une
étude d'infiltration d'eau sous rizière. Agron. Trop. 35, 25-27.
Vachaud, G., Dancette, G, Sonko, S. & Thony, J. L. (1978) Méthode de caractérisation hydrodynamique in situ d'un sol non saturé. Application à deux types de sol du Sénégal en vue de la
détermination des termes du bilan hydrique. Ann. Agron. 29, 1-76.
Van AMtmbeke, A. (1974) Management properties of ferrai soils. FAO Soils Bulletin 23.
Received 26 February 1993; accepted 24 September 1993