Download as pdf or txt
Download as pdf or txt
You are on page 1of 4

440 Plant Soil (2013) 366:433–451

Table 3 Parameters of soil hydraulic functions (van Genuchten water content for C vs. all V scenarios (KzzA – anisotropy factor/
1980), values of the minimized objective function and R2 de- tensor in the vertical direction). Parameters of dual-porosity
scribing the relationship between simulated and measured cu- model are shown for mobile/immobile soil media
mulative outflow at the soil column bottom and average soil

Scenario θr θs α n Ks Objective function R2

[cm3 cm−3] [cm3 cm−3] [cm−1] [−] [cmday-1] 0c 2c

Control 0.05a 0.43a 0.032±0.002b 1.53±0.02b 29.1±0.3b 0.005 1.216 0.999


Vegetation (anisotropy KzzA 01) 0.05 a
0.38 a
0.06 a
3.50 a
3208±0.01 b
0. 249 9.150 0.992
(anisotropy KzzA 010) 0.05a 0.38a 0.049±0.001b 2.78±0.08b 33.6±0.67b 0.204 12.10 0.987
(dual-porosity–mobile) 0.00a 0.23a 0.040a 2.54±0.20b 17.9±1.28b 0.032 – 0.996
(dual-porosity–immobile) 0.05a 0.15a 0.015a 1.50a – – –
a
Set values
b
95% confidence limits of optimized soil hydraulic parameters
c
The value of the objective function (0—measured volumetric water content - average through soil profile; 2—Cumulative boundary
fluxes across a bottom boundary 0 cumulative seepage face)

the experiment did not provide enough data to optimize boundary). Initial metal concentrations were defined
all parameters, many of them were set to values estimated according to the measured (by ICP-OES) concentra-
based on previous studies and examples listed in HYD- tions in soil water extracted from the soil, using suction
RUS programs (Radcliffe and Šimůnek 2010). The cups (Table 2). The bulk density ρd was measured in the
parameters of the soil water retention curve in the immo- 100-cm3 soil samples (Table 4).
bile (θs,im, θr,im, αim and nim) and mobile (θs,mo, θr,mo and The parameters of the Freundlich adsorption iso-
nmo) domain (Table 3) were set as follows: saturated soil- therms were previously measured and determined by
water content in the mobile and immobile domain Trakal et al. (2012). In order to evaluate the sorption
equaled to 0.23 and 0.15 cm3 cm−3, respectively, (which behaviour of chosen metals, a batch sorption experiment
was based on the soil structure observation) θr,mo in the was performed. Metal sorption, represented by KF val-
mobile domain was zero, θr,im equaled to the θr for ues decreased in the order KF ðPbÞ > KF ðCuÞ >>
equilibrium water flow, and the greater retention ability KF ðZnÞ > KF ðCdÞ. Furthermore, sorption parameters
in the immobile domain in comparison to the mobile (Table 4) showed that root presence and production of
domain was assumed and characterized by lower values the root exudates had a negligible effect on the metal
of αim and nim in the immobile domain. Remaining sorption behaviour (Trakal et al. 2012). Additionally, a
parameters (αmo, Ks,mo and ωw) were numerically opti- new batch experiment was performed according to the
mized using the cumulative outflows at the soil columns methodology described in Trakal et al. (2012). Citrate
bottoms over a period of time using the dual-porosity (representing a root exudate involved in metal mobili-
model in HYDRUS-2D. zation processes) was considered during the determina-
tion of Freundlich parameters in scenario V. More
precisely, 0.5 and 5mM citric acid was mixed with
Simulation of metal transport metals solution in order to evaluate its effect on the final
Freundlich parameters (Poulsen and Hansen 2000).
Soil hydraulic parameters resulting from the analysis of Measured metal concentrations did not provide
water flow were then used as input data for direct enough information for optimizing values of longitu-
simulation of the metals transport. The same finite dinal and transversal dispersivities; therefore the val-
element mesh, boundary and initial conditions for water ues of 5.0 and 0.5 cm were set as the longitudinal and
flow were used as before. The following boundary transversal dispersivity, respectively, which was esti-
conditions were applied for the metals transport simu- mated based on the studies performed by Kodešová et
lation (no flux at both lateral boundaries and a third al. (2009) on undisturbed soil columns and by
type boundary condition at the top and bottom Kodešová et al. (2012a) on repacked soil columns.
Plant Soil (2013) 366:433–451 441

θ [cm3 cm-3 ]
The solute transfer between the mobile and immobile
0.0 0.1 0.2 0.3 0.4 0.5
domains (when using the dual-porosity model) due to
the water transfer was only assumed. 1000

The metal root uptake was simulated in the case of


the scenario V. The maximum concentration (the lim-

Pressure head [cm]


ited value of cRoot), which can be removed from the 100

soil due to the root uptake was adjusted manually


when necessary. Adjustment was performed by com-
paring the total quantity of metals accumulated in the 10

top part of the plant measured at the end of the second


vegetation period and the simulated cumulative solute
uptake. 1

Control
Vegetation (anisotropy KzzA = 1)
Results Vegetation (anisotropy KzzA = 10)
Vegetation (dual-porosity)
Vegetation (mobile soil medium)
Vegetation (immobile soil medium)
Water flux
Fig. 3 Soil water retention curves for C and all V scenarios
Transient water flow data (cumulative outflow at the (KzzA – anisotropy factor/tensor in the vertical direction). Dual-
soil column bottom and average soil water contents) porosity model was also considered for V scenario (retention
measured during the second vegetation period were curves for mobile/immobile mediums are shown in the figure)
used for calibrating the radially symmetric HYDRUS
model. The van Genuchten hydraulic parameters, were fixed at 0.06 cm−1 and 3.50, respectively, and Ks
which were optimized, are shown in Table 3. Results value was only estimated. In this case, the 95 % con-
demonstrate reliably estimated parameters in the case fidence limits for Ks were again very narrow, but
of the scenario C (when isotropic media was as- values of the objective functions were significantly
sumed), which is indicated by their narrow 95 % con- higher and the coefficient of determination R2 was
fidence limits, low values of the objective functions lower in comparison to those from the C scenario. In
and a high value of the coefficient of determination addition, the shape of the retention curve was very
R2. On the other hand, the parameter optimization for different in comparison to the shape of the retention
the scenario V (when assuming the isotropic media) curve for the C scenario (Fig. 3). The optimized Ks
led to very high values of α and n, which caused value from the V scenario was 100 times higher than
numerical instability. Therefore, the α and n values that from the value from the C scenario. The soil

Table 4 Parameters for metal simulations: bulk density (ρd; data shown are means±SD (n03)), longitudinal (DL) and (DT) transversal
dispersivity, parameters of the Freundlich adsorption isotherm (KF, β)

Scenario ρd D La DTa KFb [cm3β nmol1−β g−1] βb [−]

[gcm−3] [cm] [cm] Cd Cu Pb Zn Cd Cu Pb Zn

Control 1.43±0.04 5.00 0.50 51.0 234 337 74.5 0.245 0.276 0.264 0.230
Vegetationc 0 mM CA 1.43±0.04 5.00 0.50 51.6 246 358 87.6 0.249 0.282 0.275 0.256
0.5 mM CA 1.43±0.04 5.00 0.50 64.9 255 372 139 0.265 0.229 0.313 0.230
5.0 mM CA 1.43±0.04 5.00 0.50 70.3 172 187 86.7 0.317 0.156 0.412 0.405
a
Value of DL and DT were obtained from Kodešová et al. (2009, 2012a)
b
Freundlich adsorption isotherm parameters KF, β were obtained from Trakal et al. (2012)
c
Represented all 3 V scenarios (single-porosity with anisotropy KzzA 01, 10 and dual-porosity soil mediums)
CA citric acid
442 Plant Soil (2013) 366:433–451

parameters for scenario V were similar to sandy ma- curve for the entire flow domain is similar to that of
terial, which did not seem to be realistic. Since the the C scenario. The Ks,mo value is even slightly lower
very fast reaction of the soil porous system was mon- than Ks value for the C scenario (Table 3). The rela-
itored (rapid infiltration at the soil column top and tively low value of first-order rate coefficients ωw
almost immediate discharge at the bottom), the anisot- (Table 3; e.g. slow interaction between the mobile
ropy of soil caused by the vertical root orientation was and immobile domains) was obtained.
assumed (soil structure compression in the horizontal Table 5 and Fig. 4 show the measured and simulated
direction and thus vertical flow predisposition). The boundary fluxes (negative values indicate an inflow into
anisotropy factor KrrA always equalled 1, the anisotro- the modelled system). An excellent fitting of the modelled
py factor KzzA equalled 10, 15 and 20, respectively. cumulative inflow and outflow in the scenario C was
The resulting parameters for one simulation (KzzA 010) obtained. Whereas in the case of scenario V the curve
are shown in Table 3. It should be noted that the modelled using the single-porosity model (not shown in
hydraulic conductivity in the horizontal direction Fig. 4) was slightly different in comparison with observed
equals the value in the table and the hydraulic conduc- data. The reason was that the simulated soil water content
tivity in the vertical direction equals the value in the was considerably higher than the monitored one (see
table multiplied by the anisotropy factor. The shape of below), as more water had remained in the soil. Both
the retention curve for KzzA 010 is shown in Fig. 3. In mismatches were caused by optimized values for the
all cases, the 95 % confidence limits were very narrow, hydraulic parameters in the V column, which were esti-
but values of the objective functions were again higher mated with lower reliability (as mentioned above). The
and the coefficient of the determination R2 was lower model was unable to closely reproduce observed data, as
in comparison to those from the C scenario. However, indicated by the objective function values. The better
the shape of the soil hydraulic characteristics changed correspondence between the measured and simulated out-
significantly. The slope of the retention curve in- flows was obtained when using the dual-porosity model
creased with the increasing anisotropy factor KzzA. (Table 5; Fig. 4). The restricted water flux into the immo-
The Ks values for all three scenarios were close to bile domain was simulated due to the low value of ωw. As
the K s value for the C scenario. Next, the non- a result, the mobile domain was quickly saturated during
equilibrium water flow was assumed and the dual- irrigation allowing fast water discharge at the soil column
porosity model was applied to simulate observed fast bottom. Figure 4 and Table 5 also show a little water
response of the soil porous system. The 95 % confi- discharge due to the evaporation from the column surface
dence limits were slightly wider then for the previous (scenario C) and 10-times higher water discharge due to
simulations, but the value of the objective function the willows transpiration (scenario V), which were both
(calculated for outflow data) decreased and the coef- observed in the laboratory and set as input data.
ficient of determination R2 (calculated for outflow The modelled and measured data of the volumetric
data only) increased (Table 3). The retention curves water content showed an excellent agreement in the
for both domains and their sum (characterizing reten- scenario C (Fig. 5), but a weak correspondence of the
tion curve of the entire flow region) are shown in measured and simulated (using the single-porosity
Fig. 3. It is evident that the shape of the retention model) data were obtained in both V scenarios.

Table 5 Simulated vs. measured water cumulative outflow, evaporation and transpiration for C and all V scenarios

Scenario Cumulative outflow [l] Cumulative Cumulative


evaporation [l] transpiration [l]

Measured Modelleda Measured Modelled Measured Modelled

Anisotropy (KzzA 01) Anisotropy (KzzA 010) Dual-porosity

Control 12.2 11.8 – – 0.94 0.94 – –


Vegetation 12.5 10.3 10.9 12.7 – – 11.6 11.6
a
V scenario modelling with applied anisotropy factor/tensor in the vertical direction – KzzA and with dual-porosity model consideration
Plant Soil (2013) 366:433–451 443

Control Vegetation
1500 0.15 1500 1.5

1000 0.10 1000 1.0

Flux (cm3 day-1)


Flux (cm3 day-1)

Flux (cm day-1)


Flux (cm day-1)
500 0.05 500 0.5

0 0.00 0 0.0

-500 -0.05 -500 -0.5

-1000 -0.10 -1000 -1.0

-1500 -0.15 -1500 -1.5


15000 15000
Measured data (C) Measured data (V)
10000
Cumulative flux (cm3)

10000

Cumulative flux (cm3)


5000
5000
0
0 -5000
Atmosperic boundary (E/T) -10000
-5000 Variable flux (irrigation)
Seepage face (outflow)
-15000
-10000
-20000
-15000 -25000

1 8 15 22 29 36 43 50 57 1 8 15 22 29 36 43 50 57
Time (days) Time (days)

Fig. 4 Measured and simulated boundary fluxes and cumulative water fluxes of the C and V scenario (only dual-porosity consideration
is shown in the figure). Fluxes, which are positive, are out of the flow domain, negative are into the flow domain

Specifically, for the first week of monitoring, the mea- were similar. The higher simulated water storage in the
sured data were in agreement with the modelled ones, soil column in comparison to the measured one (which
but the subsequently observed data were remarkably for Δθ00.13 cm3 cm−3 would be 1020 cm3) corre-
different in comparison to the simulated ones (a dif- sponds to the lower simulated cumulative outflow
ference of about 0.13 cm3 cm−3; Fig. 5). However, the compared to the measured one.
increasing/decreasing trends of the modelled and mea- Additionally, and for illustration, Fig. 6 (document-
sured water content behaviour after/before irrigation ing simulated water content distributions one day be-
fore irrigation, during irrigation, and one day after
Time (days) irrigation) shows how water may have infiltrated into
1 8 15 22 29 36 43 50 57
the soil and how rapidly water may have redistributed
within the soil column. Figure 6 also explains why
0.4
water may have drained from the soil at the time of
irrigation in both cases; when θs was reached on the
0.3 bottom of the column (boundary seepage face). Fig-
θ (cm3 cm-3)

ure 6 clearly show considerably different water distri-


0.2 butions within the soil columns. While the soil water
content near the column bottom is close to saturation
0.1
for KzzA 01 for all 3 stages, the soil water content for
KzzA 010 is lower and larger in the column before and
after the irrigation, respectively. Figure 6 also shows
0.0
greater saturation in the centre of the column during
Control the infiltration for KzzA 010. In the case of the result of
Vegetation (anisotropy KzzA = 1)
Vegetation (anisotropy KzzA = 10)
the dual-porosity model, the soil water contents in the
Measured data (Control) mobile domain simulated during the irrigation are
Measured data (Vegetation)
closer to saturation then water contents in the immo-
Fig. 5 Measured and simulated soil water contents of the C and
bile domain as well as water contents simulated using
all V scenarios (KzzA – anisotropy factor/tensor in the vertical the single-porosity model. Thus greater outflow was
direction; was considered for the V scenario) obtained as mentioned above.

You might also like