Irrigation and Drainage Systems 9: 189-204, 1995.
© 1995 Kluwer Academic Publishers. Printed in the Netherlands.
Modeling solute transport in sub-surface drained soil-aquifer
system of irrigated lands
S . K . K A M R A 1, K . V . G . K . R A O 1 & S I T A R A M S I N G H 2
lCentral Soil Salinity Researeh Institute, Karnal-132001, India; 2Water Technology Centre for
Eastern Region, VII-H/176-180, Sailashrivihar, Bhubaneswar- 751016, India
Accepted 6 April 1994
Key words: modeling, sub-surface drainage, salinity control, finite element method, watet table,
impervious layer, water quality, drain design
Abstract. A two-dimensional finite element model of solute transport in a file - drained soil aquifer system has been applied to study the effects of the depth of impervious layer and quality
of irrigation water on salt distribution during drainage of an initially highly saline soll. The model
assumes steady state water movement through partially saturated soil and to drains in the saturated
zone. The exact in time numerical solution yields explicit expressions for concentration field at any
future tirne without having to compute concentrations at intermediate times. The model facilitates
predictions of long-term effects of different irrigation and drainage practices on concentration of
drainage effluent and salt distribution in the soil and groundwater. The model results indicated
that the depth of impervious layer from drain level, d I, does not significantly influence the salt
distribution in the surface 1 m root zone of different drain spacings (drain spacing (2S) = 25, 50,
75 m; drain depth (dd) = 1.8 m), its effect in the aquifer becomes dominant as drain spacing
increases. It was also observed that d I significantly governs the quality of drainage effluent. The
salinity of drainage watet increases with increasing d r in all drain spacings and this effect
magnifies with time. The model was also applied to study the effects of salinity of irrigation water
in four drain spacing-drain depth combinations: (2S = 48 m, dd = 1.0 m; 2S = 67 m, d d = 1.5
m; 2S = 77 m, da = 2.0 m; 2S = 85 m, dä = 2.5 m). The results indicated that a favorable salt
balance can be maintained in the root zone even while irrigating with water up to 5 dS/m salinity
in drains installed at 48 to 67 m spacing and 1.0 to 1.5 m depth. Further, irrespective of the quality
of irrigation watet, the deep, widely spaced drains (dä = 2.5 m, 2S = 85 m) produced rauch saline
drainage effluent during the initial few years of operation of the drainage system than the more
shallow, closely spaced dralns, thus posing a more serious effluent disposal problem.
Résumé. Considérant les conséquences potentiellement sérieuses de ia pollution du sol et de l'eau
souterralne dans l'agriculture irriguée, il est devenu absolument nécessaire de développer des
modèles de simulation en vue d'évaluer les effects ä long terme des méthodes agricoles modernes.
Un modèle d'éléments finis ä deux dimensions du transport en solution dans un système de sol
aquifère drainé au moyen de tuyaux a été développé et validé sur le terrain (Kamra et al. 1991 a,
b). Le modèle assume le mouvement de l'eau ä régime constant ä travers un sol partiellement saturé
et jusqu'aux drains dans la zone satnrée. La solution numérique exacte dans le temps produit des
expressions explicites pour le champ de concentration ä un temps future quelconque sans avoir ä
calculer les eoncentrations aux temps intermédiares. Le modèle facilite les prédictions des effets
ä long terme des diverses méthodes d'irrigation et de drainage sur la concentration des effluents
de drainage et sur la distribution de la salinité dans le sol et dans l'eau souterraine. Les résultats
190
du modèle relatifs aux effets de la profondeur de la couche imperméable et de la qualité de l'eau
d'irrigation sur la distribution de la salinité lors du drainage d ' u n sol fortement salé ä l'origine sont
mentionnés dans la présente communication.
Les résultats du modéle ont indiqué que la profondeur de la couche imperméable depuis le niveau du drain, dl, n'influence pas d ' u n e fa9on significative la distribution de la salinité dans la
zone superficielle radiculaire de 1 m des divers écartements de drains (écartement de drains, 2S =
25, 50, 75 m; profondeur des drains, d d = 1.8 m); son effet dans l'aquifère devient dominant ä
mesure que l'écartement de drains augmente. O n a aussi eonstaté que le niveau du drain d I influence d ' u n e manière significative les effluents du drainage. La salinité de l'eau de drainage augmente ä mesure que d I augmente dans tous les écartements de drains et cet effet s'amplifie avec
le temps. Le modèle a été aussi appliqué pour étudier les effets de la salinité de l'eau d'irrigation
dans le eas de quatre conbinaisons d'écartement de drain et de profondeur de drain: (2S = 48 m,
dä = 1 , 0 m ; 2 S = 6 7 m , d d = 1 , 5 m ; 2 S = 7 7 m , d d = 2 , 0 m ; 2 S = 8 5 m , d d = 2 , 5 m ) . L e s
résultats ont indiqué q u ' u n bilan de salinité favorable peut ~tre maintenu dans la zone radiculaire
m ê m e en irrigant avec de l'eau d ' u n e salinité de 5 d S / m dans des drains installés ä u n écartement
de 48 ä 67 m et une profondeur de 1,0 ä 1,5 m. De plus, i n d é p e n d a m m e n t de la qualité de l'eau
d'irrigation les drains profonds ä grand écartement (d d = 2,5 m, 2S = 85 m) produisaient une
grande quantité d'effluents salés de drainage durant les quelques premières années de l'exploitation du système de drainage par rapport aux drains peu profonds ä écartement serré, posant ainsi
u n problème plus serieux d'évacuation des effluents.
Les résultats du développement et de l'évaluation du modèle ont montré qu'il peut être utilement
employé en vue d ' u n e évaluation judicieuse de la variation de temps eseomptée dans la salinité des
effiuents de drainage lors de la reise en valeur des sols salins et peut ainsi aider ä formuler son règlement plus sür du point de vue environnement et les projeets d'évacuation.
Introduction
Drainage is generally required to combat the twin problems of waterlogging
and soil salinity and to ensure sustainted irrigated agriculture in the arid and
semi-arid regions. While the benefits of drainage can be counted in terms of
improved crop yields and increased economic gains, environmental considerations related with the disposal of saline drainage effluents, sometimes also containing high concentrations of plant nutrients, trace elements and pesticides,
impose severe constraints on the design and operation of drainage and related
water management projects (Tanji 1990). Reliable long-term estimates of the
volume and composition of drainage effiuent in time are required to plan sustainable strategies for its disposal or treatment. Simulation models are generally required to predict the long-term consequences of the management decisions
on the performance of drainage systems and evolving a working balance between the maintenance of agricultural productivity and protection of natural
resources.
Several simulation models of saturated-unsaturated water flow have been
developed to relate drainage system design to soil properties and climatic conditions (Skaggs 1978; Feddes et al. 1978; Belmans et al. 1983; Lesaffre &
Zimmer, 1988). The scope of these models has now been extended for arid and
191
semi-arid regions to study-the effect of a given designon the salt distribution
in the soil profile and or quality of drainage water (Pickens et al. 1979; Nour
el-Din et al. 1987 a, b; Tracy & Marino 1989; Evans et al. 1989; Kamra et al.
1991 a, b). Numerical models, based on standard finite difference or finite element techniques, transform the space derivatives of the governing partial
differential equations of water and or solute transport into a finite set of approximate algebraic equations. The time derivative is mostly discretized by iterative finite differences which involve marching through the intermediate time
steps to develop solution at the desired time. Kamra et al. (1991a, b) used a
semi-discrete approach in which only space was discretized and an exact in-time
analytical solution of the system of ordinary differential equations yielded explicit expressions for the concentration field at any future time without needing
to compute it at the intermediate times. The two dimensional finite element
model of Kamra et al. (1991 a, b) simulates solute transport in sub-surface
drained lands under assumptions of steady state watet flow in the unsaturated
and saturated zones, and includes the effect of convective transport, dispersion
and linear adsorption. The model provides long-term predictions of the
desalinization of a drained soil, and of the associated changes in the quality of
the groundwater and the drain effluent.
The basic features of the model, and its calibration and field validation are
briefly discussed and the model applications on the effect of depth of impervious layer and the quality of irrigation watet on transient movement and distribution of dissolved chemicals in sub-surface drained soil-aquifer system are
presented in this paper.
Modei formulation
Fig. 1 schematically shows the movement of water and dissolved solutes to
parallel drains in a tile-drained soil-aquifer system. The space co-ordinate X
is positive towards right, whereas Y is positive downward. Infiltrating rain and
irrigation water is assumed to flow vertically downward through the partially
saturated soil before reaching the arch shaped steady stare ground water table,
JE. After reaching the water table, water and dissolved salts move twodimensionally towards the parallel drains.
Consider the governing equation of two dimensional solute transport in
unsaturated-saturated porous media (Kamra et al. 1991 a)
0ReSt -SX
-
0Dxx~+0D~y~ + ~
~ ( q x C) + ~ff(qyC)
+ ¢(X,Y,t)
0Dex~-~+oDyy
(1)
192
L K
X
0
Land surface
Y
dd
J"~"-Strlp stak
"-Water divide
TJle draln"~
Stream
,~sh Aqui~
line
d[
Streamline
H',=
S
q
'-ImpermeabJe Barrier
Fig. 1. Flow domain for solute transport in a tile -
drained soil - aquifer system.
in which
C
Rf
=
=
=
Kd =
0
=
dissolved solute concentration, M/L3;
retardation factor, equal to { 1 + ~Kd/0 }, dimensionless;
bulk density of porous medium, M/L3;
distribution coefficient of solute species, L3/M;
volumetric watet content (equal to porosity in the saturated zone),
L3/L3;
Dxx , D ~ y ~ y y x, Dyy -- components of the dispersion coefficient tensor,
Jl-a
/
~ ,
I
qx, qy = Darcian specific discharge components, L / T ;
(X,Y,t) = source or sink term, being positive for sources and negative for
sinks, M/L3T;
X,Y = space coordinates, L;
t = time, T.
The retardation factor Rf in (1) accounts for linear equilibrium interactions
between the solute and the porous medium. The dispersion coefficients for a
193
two-dimensional isotropic porous medium were adopted from Scheidegger
(1961) which relate Dxx, Dxy, Dyx, Dyy to qx, qy, 0, qs, aL and a T where ai~ and
a T are the longitudinal and transverse dispersivities (L), respectively and qs is
the magnitude of specific discharge vector (L/T).
Steady state water m o v e m e n t
Wierenga (1977), Beese and Wierenga (1980) and Destouni (1991) have shown
that transport models based on steady state water flow can produce concentration distributions that are comparable to those obtained with transient water
flow models, but with considerably less input data requirements than the transient models. The steady state formulations can be particularly useful for making long-term predictions by ignoring the often highly dynamic but short-term
oscillations in water content and solute concentration near the soff surface. Accordingly steady state water flow models were used in this study for both the
unsaturated and saturated zones.
Unsatnrated z o n e
The velocity field for the unsaturated part of the flow domain was obtained
by considering the water flow to be vertical (qx = zero) and taking qy equal
to the net upward or downward steady state flux. The net water flux during
a period was obtained from the water balance of the area and a reliable estimate
of the groundwater contribution to evaporation. If K and h respectively
represent the hydraulic conductivity and pressure head in soil and provided the
water retention 0(h) and hydraulic conductivity K(h) curves of the soil are available, the Darcy law equation may be integrated numerically to determine water
content 0 at required heights above the water table in a soff profile during
steady upward or downward water flow. The functional forms of unsaturated
hydraulic properties used in this study were those of Van Genuchten (1978).
Saturated z o n e
Kirkham (1958) analytically solved the Laplace equation for the water table
height, Z, above the drain axis, and the hydraulic head distribution in the flow
domain FGHIJF below the drain axis (Fig. 1) for a homogeneous aquifer. The
specific discharge components, qx and qy, were computed for the region (0 _
X _< S, 0 _< Y _< d~) with the help of Darcy's law (Kamra 1989). Solutions are
applicable to other half of the domain between the two drains because of
symmetry.
194
Initial conditions
The measured solute concentration in the flow domain before the beginning
of the simulation period is taken as the initial condition:
C(X,Y,0) = Co(X,Y) at t = 0
(2)
Boundary conditions
The solute flux is prescribed on a Cauchy boundary, the normal gradient of
concentration on a Neumann boundary, while concentrations are prescribed
on Dirichlet boundary nodes. A Cauchy boundary condition is generally applied to a boundary through which solute enters the region. The Neumann
boundary conditions are imposed on flow-through boundaries with outflow
from the region, and on impervious boundaries. The land surface KO (Fig. 1)
acts as a Cauchy boundary during the infiltration phase and the segments HG
(bottom basis layer), OG (water divide), and HI and JK (streamlines) are treated as impervious boundaries. The file snrface IJ and land surface KO during
evaporation are outflow boundaries. There are no Dirichlet boundaries in the
present study.
Finite element solution
In the finite element method, the flow region is divided into a network of
subregions called elements, whose corners are taken as the nodal points.
Different element shapes can be defined. The value of a variable within the
element is interpolated in terms of its values at the corner nodes. Simple poly-'
nomials (linear, quadratic or cubic) are frequently used as linearly independent
basis functions for the interpolation. The finite element numerical equations
are usually formulated with either the weighted residual Galerkin or the variational approach. In the more commonly used Galerkin scheme, a trial solution
made up of an expansion of basis functions is substituted in the differential
equations and the resulting residual (error) is forced to be zero by requiring its
orthogonality to each of the basis functions. Quadrilateral elements and linear
basis functions were employed in this model to approximate equation (1) with
a vector-matrix differential equation of the following form:
[AMI [dC/dt} = [DM] {C] + [F}
where
(3)
195
[AM]
[DM]
{F}
{c}
NN
NN x NN symmetric coefficient matrix;
NN x NN nonsymmetric matrix accounting for convection, dispersion and outflows boundaries;
NN × 1 vector representing the sources/sinks, and boundary conditions of the transport equation;
NN x 1 vector of nodal concentrations;
number of nodes in the discretized domain at which the concentration is unknown.
The typical elements of these matrices, the finite element evatuation of spatial
derivatives of equation (1) and the application of initial and boundary conditions has been presented in Kamra et al. (1991a). In the semi-discrete approach
of Kamra et al (1991a) only space was discretized and an exact-in-time analytical solution of the inhomogeneous matrix-vector equation (3) yielded explicit
expressions for the concentration field at any fnture time. In this approach, the
eigensystem (eigenvalues and eigenvectors) of the coefficient matrix resulting
from spatial discretization is first computed and then exponentiated. For solute
transport, the eigensystem may be complex (i.e. involve imaginary components) due to asymmetry created by the convection term in the governing convection dispersion equation. Compared to the standard numerical algorithms,
the semi-discrete method requires less computer time and offen yields smaller
truncation errors for long term predictions.
Model calibration and field validation
The model was validated against field results of a sub-surface file drainage experiment of Central Soil Salinity Research Institute, Karnal (India), conducted
on its Saline Soll Research Farm at Sampla (District Rohtak) in the State of
Haryana. Sub-surface drainage system, consisting of thrice replicated three
drain spacings of 25, 50 and 75 m and average drain depth of 1.80 m, was installed at Sampla in the summer of 1984 in a 10 ha saline area. The soll salinity
(ECe, electrical conductivity of the saturation extract) of the surface 15 cm soll
in the area ranged from 20 to 100 dS/m. The salinity was about 30 dS/m in
the 15-30 cm layer and 20 dS/m below 30 cm. Dissolved salts were mainly calcium, magnesium and sodium chlorides. Before installation of the drains, the
water table in the area typically fluctuated between a depth of 1.5 m (during
early summer) and the soll surface (during the rainy season). Salinity of the
groundwater near the water table varied from 10-40 dS/m. The soll in the
region is a sandy loam alluvium having hydraulic conductivity of 1.0 m/day
up to 1.75 m depth, followed by a loamy sand zone of 3 m/day hydraulic conductivity. This porous zone extends to a fine textured layer of low permeability
196
at 3-4 m which was treated as the impermeable boundary (Rao et al. 1986). The
values of selected hydraulic and drainage system parameters, including steady
annual water fluxes, are listed in Table I.
Table 1. Values o f selected soil hydranlic and drainage system parameters.
Parameter
Value(s)
Drain Spacing, 2S
Drain Depth, d d
Depth o f impervious layer
below drain axis, d I
Saturated hydraulic
conductivity of aquifer, K s
25, 50, 75 m
1.8 m
1.2, 2.0, 5.0 m
Soil hydraulic parameters
(Van Genuchten, 1978)
Ks
Os
0r
ct
n
m
3.0 m / d a y
1.0 m / d a y
0.4486
0.1004
0.0088 1/cm
1.6715
0.4017
Soil bulk density, E
Distribution coefficient, K d
(cm3/g)
1.5 g / c m 3
0.0
Longitudinal dispersivity, ŒL
Transverse dispersivity, IXT
0.8 m
0.08 m
Annual steady water flux for
25, 50 and 75 m drain spacing
1.0, 0.7, 0.4 m m / d a y
The observed seasonal drain discharge rates, after correcting for estimated
lateral seepage and upward water fluxes from water table during summer in individual plots, were combined to compute seasonal and annual watet fluxes
which were highly variable for different drain spacing plots. Numerical results
corresponding to different values of longitudinal dispersivity a L, were used
for calibrating the model to observed soil solution and drain effluent concentrations during 1984 and the selected value of GtL w a s used to validate the
model against field observations of 1985. The transverse dispersivity, a T, was
assumed to be always one tenth of mE. The model was then applied to make
10 year predictions on salt distribution in soil, groundwater and drainage effluent.
Kamra et al. (1991, b) report that the predicted results on salinity in soil pro-
197
file matched favourably with observed ones except in the surface 20 cm layer.
The differences in the surface layers exist due to the assumption of steady water
movement during a time interval in which short term and highly dynamic oscillations of solute concentration near soil surface get ignored. These differences
can be minimized by taking a smaller time interval for computing the steady
stare watet flux. Predicted results on long term improvement in salinity of
groundwater and drainage effluent did not match with observed field values
because salt load of seepage water from surrounding areas was not considered
in the model. Component of seepage is being incorporated in the model. Further details on the calibration, field validation, long-term predictions and sensitivity analysis of a number of model parameters like drain spacing-drain
depth combinations, size of drain, initial salinity of ground water, longitudinal
and transverse dispersivities and aquifer layering can be found in Kamra et al.
(1991 b), while detailed analysis on the effect of adsorption on salt dynamics
in sub-surface drained soil-aquifer system can be found in Kamra et al. (1994).
Results and discussion
The results related with long-term effects of the depth of impervious layer and
salinity of irrigation water on salt distribution in tile-drained lands are discussed in these sections. Since field experiments involving different drainage
system parameters and salinities of irrigation water have not been conducted,
the model predictions may be treated as qualitative only. Nevertheless, the
model results do provide improved understanding of the mechanisms and sensitivity of salt distribution in subsurface-drained soils to these two parameters
for inland saline sandy loam soils of Haryana (India).
Effect of depth of impervious layer
From field observations of hydraulic head and drain discharge, the average
depth of the impervious layer, d I at Sampla was estimated at 1.2 m below
drain axis (Rao et al. 1986). The model results corresponding to three values
of d I (= 1.2, 2.0 and 5.0 m) on salt distribution in the soil profile and the
aquifer at midplane of different drain spacing plots, two and five years after
operation of drainage system, are presented in Fig. 2. The results for aquifer
in all drain spacings were restricted to a depth of 2.7 m from drain level below
which salinity at any time did not vary with depth. These observations are similar to water movement results of Childs (1943) which indicated that about 75%
of the total flow to a field drain in a deep homogeneous soil takes place within
a depth equal to 1/20 of the drain spacing from the drain axis.
198
4
0
S ~
8
ECe (d S / m )
12
16
20
I
l
~,,~ ~..~
-~~,~
24
.i . . . .
I
28
:32 048 50
, _ _ . _ . . ~ » . _ +_.+l.~
e,°,ù°,SChù,
-%,--...~
L .......
~,~~,-,.. ~..'~'*-,.-
-i
--.«-..... \ \
4"5
O 0
(o) 2S=25m
(p'~~
I
I
""-.~i
" I
I
I
-
E
"\~~),.
~
I
I
T = 2 Years
T= 5 Yeors
I I
....
2
e-
o.
3
a
4.5
0
'~'(~L~ ~
(c) 2 S = 7 5 m
4"5
I
I
I
I
I
"x~
--
Fig. 2. Effect of depth of impervious layer, di, on salt distribution in soil profile and aquifer for
three drain spacings: (a) 25 m, (b) 50 m, (c) 75 m; drain depth = 1.8 m.
T h e d e p t h o f i m p e r v i o u s layer, d p h a d little influence o n the salt d i s t r i b u t i o n in 1.0 m effective r o o t z o n e o f all d r a i n spacings (Fig. 2) m a i n l y b e c a u s e
t h e w a t e t m o v e m e n t was a s s u m e d to b e s t e a d y a n d its e q u a t i o n for the u n s a t u r a t e d zone ( K a m r a et al. 1991a) d i d n o t a c c o u n t for d r T h e small differences
199
2824-
16-
8-
028-
24~
E
16"10
UJ 8 -
0
I
r
f
I
m
J
I
I
2
4
J
I
I
J
___L
I
28-24
16--
8-(c) 2 S = 7 5 m
0
6
P
8
l
i
I0
Time (yeors)
Fig. 3. Effect o f depth of impervious layer, d~, on salinity (ECd) of drainage effluent for three
drain spacings: (a) 25 m, (b) 50 m, (c) 75 m; drain depth = 1.8 m.
200
in individual drain spacing plots can be attributed to differences in moisture
distributions due to different water table profiles obtained with different
values of d I. However, because of increase in saturated flow domain with increasing di, its effect on salt distribution in aquifer increased; the effect becoming more pronounced with increasing drain spacing. The model results for
50 and 75 m drain spacings also indicated the aquifer at 1.2 m from drain level
to be relatively more saline when d I is equal to 1.2 m than the case when d I _>
2.0 m. As discussed by Childs (1969), the sections EFG, G H and HI (Fig. 1)
of the flow domain constitute a bounding streamline. The referred zone for di
= 1.2 m (point G in Fig. 1) represents a stagnation point where, due to n/ 2
change in direction of streamline, the flow velocity becomes zero resulting in
little improvement in the salinity of the region. The model results indicated
similar behaviour in corresponding areas for cases when d I is equal to 2.0 and
5.0 m. Point H (Fig. 1) is another stagnation point which, though not discussed
in Fig. 2, showed little improvement in salinity.
Fig. 3 presents the effect of d~ on the time variation in the salinity of
drainage effiuent, ECa, in different drain spacings. It is observed that the
depth of impervious layer significantly governs the quality of drainage effluent. EC d in all drain spacings increased with increasing dl, apparently due
to extension of saline groundwater domain, and the effect magnified with time.
In the first two years, the difference in EC a due to increasing d I are seen to be
more dominant in the closer drains. This is because during initial stages of
reclamation, the salts leached from soil profile contribute more than the salts
drained from aquifer towards the salt load of drainage effluent. Since leaching
is more efficient (faster) in closer drains, this trend continues for a longer period in widely spaced drains. After initial 2-3 years, the fraction of salts leached
from soil profile decreases and the effect of d I on EC d manifests itself more
significantly.
Effect of quality of irrigation water
The model was also applied to study the effect of salinity of irrigation water
on solute transport under alternate drainage designs. Four drain spacing (2S)d r a i n d e p t h ( d d ) combinations:2S = 4 8 m , d d = 1 . 0 m ; 2 S = 6 7 m , d a =
1.5 m; 2S = 77 m, d a = 2.0 m; 2S = 85 m, d d = 2.5 m, having drain discharge vs. hydraulic head relationships identical to those of the recommended
combination of 2S -- 75 m, d d = 1.8 m for Sampla, were used in the comparison. The iso-salinity contours within the flow domain, corresponding to two
salinities (Cin = 0.5 and 5.0 dS/m) of irrigation watet, for 2S = 67 m, d d -1.5 m and 2S = 85 m, d d = 2.5 m are presented in Fig. 4, and the time variation
in the salinity of drainage effluent for the four designs are presented in Fig. 5.
201
2S:67m~dd=l'Sm
0.4
0'6
0'8
"K,.~/rlß ~~
~ ~
drain[~ß.k_~..j,s ~ , 8 ~ _
~
0"0
0"2
»o~~O-~/-~,
x~
,
,
1.0
, ~.~/~
0.0
0-2
~f-~/-xL
«roßt3-~l~/f~~
3.~/~-~lly,
2 S =85m»dd= 2 ' 5 m
0-4
0-6
0.8
1.0
I ~~_...jV-x~~_/..~ ~ i ~, ,«
I ~ ',~,~~15 ~,\
~ß
~//~~k\
X/S
,
l/,_LL.7, \¥3~/I
--~X/S
(a) C i n = O ' 5 d S / m » T = 2
yrs.
ù ~ S o i l surface
o~~-._=.~~ -.«-~ ,~._~ ~~~
I 5 ~~''L'~----'- t° ~
=° - ' ' ' ' - - ~
3;oß\~,ù'~,~~,2,~.,";'-~ ~
0'0
0'2
(0"4
0'6
0'8
W,Impervious layer
~-~\~___.,.-.-1"~ 1
2
1.5~[~\~.__i_,4 ~
3.0 ~. "~ls-..J'x I r"A'~l ~
0'0
0'2
0'4
~X/S
~
I'0
,o~«~_._~,»~«~.~~~c.~__.~~_
~I' .l
3.0, ~ l l ,//~-'%~-~~ ~\~.]
0'0
2~
"---~
i "tS~/TA
0'8
0'4
0'6
0'8
I'0
(b) Cin= 0 ' 5 dS/m., T = 5 yrs.
~ "
0'6
0'2
I'0
1"5
Il
3.O
0"0
(c)Cin = 5 dS/m» T=5
9)
i
'
0"2
t
,
~11
0"4
0"6
-,- X/S
~
0'8
I
i
I'0
yrs.
(Length units in m»
~H.~in
dS/m)
Fig. 4. E f f e c t o f s a l i n i t y o f i r r i g a t i o n w a t e r , Cin , o n salt d i s t f i b u t i o n in s u b s u r f a c e - d r a i n e d soils
f o r t w o d r a i n s p a c i n g - d e p t h c o m b i n a t i o n s o f 2S = 67 m , d d = 1.5 m (left side) a n d 2S = 85
m , d d = 2.5 m ( r i g h t side).
The results indicate that irrespective of the salinity of irrigation water up to
5 dS/m, the desalinization of the soll profile is more effective with the deeper
drains. However, the shallower drains are also reasonably effective in rapidly
reducing the salinity of top 1 m soil profile that is involved with crop production. It is seen from Figs. 4b and 4c that an increase in Cin from 0.5 d S / m to
5.0 d S / m in 1.5 m deep drains increases the salinity of effective root zone and
of aquifer after five years from 9 to 12 d S / m and 12 to 14 dS/m, respectively.
The corresponding increases for root zone and aquifer in 2.5 m deep drains are
from 6 to 10 and 10 to 12 d S / m respectively, indicating marginal long-term advantage of deeper drains in using saline irrigation water up to 5 dS/m.
The flow patterns of salt movement in 1.5 and 2.5 deep drains in a 3.0 m soilimpervious layer domain (Fig. 4) are quite different. The salinity contours of
1.5 m deep drains are almost symmetric around the drain till a lateral distance
of 0.05S and uniform (parallel contours) in the remaining area indicating the
leaching of salts both from above and below the drain level. In 2.5 m deep
drains, the convergence (concentration) of salts extends to a lateral distance o f
0.15S from the drain, with most of salts from above the drain appearing to be
leached through a sink in the bottom impervious layer. The presence of similar
202
30-
( o ) Ctn =0.5 dS/m
24-
0 2S=48m~dd=l.Om
O 2S=67m» dd =1.5 m
X 2S=77m» d d = 2. 0m
=
18-
=
•
12A
E
o~
'I0
6-
"10
IJJ
t.
_=
0
I
I
I
I
I
2
4
6
8
I0
0
]=
C
30-
o
"10
ö
(J
bJ
24-
18-
12"d
6-
0
I
I
I
I
I
2
4
6
8
I0
Time~ f (years)
Fig. 5. Predicted changes in EC d in different drainage designs using irrigation water of salinity
(Cin) (a) 0.5 dS/m and (b) 5 dS/m.
203
sinks farther away from drain are probably due to the appearance of oscillations in numerical results of wider drains with a relatively coarser discretization.
Fig. 5 indicates that, irrespective of the quality of irrigation water, EC d for
the deepest and the most widely spaced drains is higher than for the shallower
and closer drains during the first three years, after which it reduces sharply to
less than those for the other cases. This is understandable since a much larger
soff volume is involved in the leaching process for deeper drains, resulting in
an initial much heavier salt load to drains. However, once most of the salts are
removed, the soff profile for the deepest drain becomes relatively salt-free and
EC d reduces relatively more rapidly than the shallower drains. Further, it appears that at any time the increase in the salinity of drainage effluent (and also
of root zone), corresponding to an increase in ein from 0.5 tO 5.0 dS/m, is
slightly more in deeper drains (dd _> 2.0 m) than in shallower drains.
Summary and conclusions
A two dimensional finite element model of salt transport in tile drained soilaquifer system has been presented. The input data requirements of the model
include drainage system parameters (such as drain depth, drain spacing, and
radius of the drain), aquifer parameters (porosity and hydraulic conductivity
of aquifer material, depth to impervious layer, and groundwater salinity), soil
parameters (notably the soil water retention and unsaturated hydraulic conductivity functions and initial soil salinity), solute adsorption parameters (the
equilibrium distribution coefficients of the saturated and unsaturated zones),
and inflow parameters (rainfall, evapotranspiration, quantity and quality of
irrigation water).
The model results indicate that though the depth of impervious layer, di,
has little effect on root zone salinity; it significantly governs the quality of
drainage effluent. The salinity of drainage water increases with increasing d I
in all drain spacings and this effect magnifies with time. These results emphasize the need of careful and intensive investigations on d I in a reclamation
project to judiciously assess the expected time variation in the salinity of
drainage effluent under alternative drainage designs. This can help to plan and
execute environmentally safer disposal and utilizational schemes of the saline
drainage water. The model results also indicate that for inland saline sandy
loam soils of Haryana, India, favorable salt balance can be maintained in the
root zone even while irrigating with 5 dS/m saline water and using sub-surface
drains of 48 to 67 m spacing and 1.0-1.5 m depth. The quality of effluent of
drains wider and deeper than these limits, especially if installed at depths _>2.5
m, is more saline than that of shallower drains during the initial years which
may pose a relatively more difficult surface disposal problem.
204
References
Beese, F. & Wierenga, P.J. 1980. Solute transport through soil with adsorption and root uptake
computed with a transient and a constant flux model. Soil Sci. 129: 245-252.
Belmans, C., Wesseling, J.G. & Feddes, R.A. 1983. Simulation model of the water balance of
cropped soll: SWATRE. J. Hydrology 63: 271-286.
Childs, E.C. 1943. The water table, equipotentials and streamlines in dralned lands. Soil Sci. 56:
317-330.
Childs, E.C. 1969. An Introduction to the Physical Basis of Soll Water Phenomena. Wiley, London. 493 pp.
Destouni, G. 1991. Applicability of the steady state flow assumption for solute advection in field
soils. Water Resour. Res. 27(8): 2129-2140.
Evans, R.O., Gilliam, J.W. & Skaggs, R.W. 1989. Effects of agricultural watet table management
on drainage water quality. Report 23 7, Water Resources Research Institute o f the University o f
North Carolina, Raleigh, U.S.A.
Feddes, R.A., Kowalik, P.J. & Zaradng, H. 1978. Simulation of water use and crop yield. Simulation monograph. PUDOC, Wageningen, The Netherlands. 189 pp.
Kamra, S.K. 1989. Finite element simulation of solute transport in groundwater aquifers. Unpublished Ph.D. Thesis, Punjab Agril. Univ., Ludhiana, India. 240 pp.
Kamra, S.K., Singh, Sita Ram, Rao, K.V.G.K. & Van Genuchten, M.Th. 1991 a. A semidiscrete
mode1 for watet and solute movement in tile-dralned solls: 1. Governing equations and solution.
Water Resour. Res. 27(9): 2439-2447.
Kamra, S.K., Singh, Sita Rare, Ran, K.V.G.K. & Van Genuchten, M.Th. 1991 b. A semidiscrete
model for water and solute movement in tile-dralned soils: 2. Field validation and applications.
Water Resour. Res. 27(9): 2449-2456.
Kamra, S.K., Singh, S.R. & Rao, K.V.G.K. 1994. Effect of depth of impervious layer and adsorption on solute transport in tile-drained irrigated lands. J. Hydrology 155: 251-264.
Kirkham, D. 1958. Seepage of steady ralnfall through soil into drains. Trans. Am. Geo. Union
39: 892-908.
Lesaffre, B. & Zimmer, D. 1988. Subsurface drainage peak flows in shallow soil. J. Irri. Drain.
Div., Am. Soc. Civ. Eng. 114(3): 387-406.
Nour el-Din, M.M., King, I.P. & Tanji, K.K. 1987 a. A salinity management mode1: I. Development. J. Irri. Drain. Dir., Am. Soc. Civ. Eng. 113(4): 440-453.
Nour el-Din, M.M., King, I.P. & Tanji, K.K. 1987 b. A salinity management model: II. 1 and 2-D
applications. J. Irri. Drain. Div., Am. Soc. Civ. Eng. 113(4): 454-468.
Pickens, J.F., Gillham, R.W. & Cameron, D. 1979. Finite element analysis of the transport of
water and solutes in tile drained soils. J. Hydrology 40: 243-264.
Rao, K.V.G.K., Singh, O.P., Gupta, R.K., Kamra, S.K., Pandey, R.S., Kumbhare, P.S. & Abrol,
I.P. 1986. Dralnage investigations for salinity control in Haryana. Bull. 10. Central Soil Salinity
Research Institute, Karnal, India. 95 pp.
Scheidegger, A.E. 1961. General theory of dispersion in porous media. J. Geophys. Res. 66(10):
3273-3278.
Skaggs, R.W. 1978. A water management model for shallow water table solls. Tech. Rep. 134,
Water Resour. Res. Inst., North Carolina State University, Raleigh, N.C., U.S.A. 178 pp.
Tanji, K.K. (Ed.). 1990. Agricultural Salinity Assessment and Management. Manual 71. Amer.
Soc. Civil Eng., New York. 619 pp.
Tracy, J.C. & Marino, M.A. 1989. Solute movement through root-soil environment. J. Irri. Drain.
Dir., Am. Soc. Civ. Eng. 115(4): 608-625.
Van Genuchten, M.Th. 1978. Calculating the unsaturated hydraulic conductivity with a new
closed form analytic model. Res. Rep. 78-WR-08, Water Resour. Program, Dep. o f Civil. Eng.,
Princeton Univ., Princeton, N.J.
Wierenga, P.J. 1977. Solute distribution profiles computed with steady state and transient water
movement models. Soil Sci. Soc. Am. J. 41(6): 1050-1055.