Nitrogen Dynamics and Biomass Production in a Vertical Flow Constructed Wetland Cultivated with Forage Rice and their Mathematical Modeling

Journal of Water and Environment Technology, Vol. 7, No. 4, 2009

Nitrogen Dynamics and Biomass Production in a
Vertical Flow Constructed Wetland Cultivated with
Forage Rice and their Mathematical Modeling
Masaki SAGEHASHI*, Sheng ZHOU*, Tatsuro NARUSE**, Mari OSADA**,
Masaaki HOSOMI*
* Graduate School of Engineering, Tokyo University of Agriculture and Technology, Tokyo
184-8588, Japan
** (Former Affiliation) Graduate School of Engineering, Tokyo University of Agriculture and
Technology, Tokyo 184-8588, Japan
ABSTRACT
Forage rice has high potential to produce biomass, and the vertical flow (VF) constructed
wetland in which forage rice is cultivated is one of the effective ways to achieve the purification
of eutrophicated water and biomass production simultaneously. To design and manage the VF
constructed wetlands cultivated with forage rice adequately, nutrient dynamics and the growth
of the rice should be understood quantitatively. In this study, we performed a series of
experiments replicating VF constructed wetlands involving the cultivation of a variety of forage
rice ("Kusahonami") using river water (supply rate : 0.1, 0.2, and 0.6 m3/(m2·day)) for 169 days.
The results showed that the rice biomass increased with the river water supply rate. A
mathematical model was developed based on these experimental observations in order to
quantitatively explain the nitrogen dynamics in VF constructed wetlands cultivated with forage
rice. The changes in both the rate of nitrogen assimilation by rice and the denitrification rate
with the change in the rate of water supply were simulated with the proposed model.
Keywords: forage rice, nitrogen dynamics, vertical flow wetland.

INTRODUCTION
Excess nutrient loading in a water environment causes various problems such as
deterioration of water quality, landscape damage, etc. Nutrient loading sources can be
divided into two types, namely, point and non-point sources, with the latter sources
usually more difficult to mitigate. Constructed wetlands represent one of the promising

techniques for removing nutrients from relatively large-scale water bodies. This
technique can be utilized in response to non-point nutrient sources. Many studies have
been performed to clarify the performance of various constructed wetlands (e.g.,
Nungesser and Chimney, 2006; Gu et al., 2006; Behrends et al., 2007; Zhou and
Hosomi, 2008a).
The constructed wetlands can be roughly divided into two types, i.e., the free water
surface flow (FWSF) and the sub-surface flow (SSF) types, with the latter including the
horizontal sub-surface flow (HSSF) and vertical flow (VF) types (Zhou and Hosomi,
2008b). In the VF constructed wetland, the wastewater is poured into the soil layer with
a distribution pipe, and the treated water flows out from the bottom through a drainage
pipe (Brix and Arias, 2005).
Forage rice has received attention as biomass cultivated in constructed wetlands. The
Address correspondence to Masaki SAGEHASHI, Graduate School of Engineering, Tokyo University of
Agriculture and Technology, Email:
Received February 16, 2009, Accepted August 6, 2009.
- 251 -


Journal of Water and Environment Technology, Vol. 7, No. 4, 2009

water penetration rate is an important factor for the growth of rice in paddy field.
Meanwhile, the water penetration rate in VF constructed wetland can be controlled
artificially. Moreover, the VF constructed wetland has high potential for nitrogen
removal (Zhou and Hosomi, 2008b). Therefore, the VF constructed wetland cultivated
with forage rice is a fascinating subject. To design an adequate VF constructed wetland
using forage rice, it is required to understand in detail the dynamics of nutrients in the
soil layer and resultant rice growth.
The purpose of this study is thus to clarify the effects of the water supply rate on the
dynamics of nitrogen in VF constructed wetlands cultivated with forage rice. This was
done both by experimental and model analysis approaches.

MATERIALS AND METHODS
Pot Experiment
A series of experiments was performed at an open-air experimental station located in
Ibaraki Prefecture, Japan. Figure 1 shows the outline of the experimental apparatus used
in this study. The apparent density and porosity of the ando soil were 0.44×106 g/m3 and
0.676 m3/m3, respectively. Gravels (diameter = ca. 1 to 2 cm) were put into the gravel
zone, and unwoven cloths were installed at the upper part of the gravel zone and the
water effluent port. River water obtained from the Sanno-gawa River was supplied at
the top surface of the pot, and it penetrated into the soil zone and flowed out from the
bottom of the pot. Soil water samplers (DIK-8391, Daiki Rika Kogyo, Japan) were
installed at depths of 10 cm and 20 cm in the soil zone to sample the soil interstitial
water. The water supply rate was controlled to achieve preset supply rates, and the water
level was adjusted by back pressure at the water effluent port.
In this study, a kind of forage rice known as “Kusahonami” was employed.
"Kusahonami" is a new variety developed for whole-crop silage. It can produce a larger
quantity of biomass, and has a higher tolerance for nitrogen loading than the commonly
used rice variety (Zhou and Hosomi, 2008a; Sakai et al., 2003). Two rice seedlings were
transplanted in a pot on May 8, 2005 and harvested on October 24, 2005.
In the experiments, the river water supply rates were set at 0.1, 0.2, and 0.6 m3/(m2·day),
and the time courses of the concentrations of inorganic nitrogen compounds (i.e.,
ammonium, nitrate, and nitrite nitrogen) in the supplied water, soil interstitial water, and
effluent water were monitored as well as the leaf number and plant height of the rice.
The concentrations of inorganic nitrogen ions were measured with an ion
chromatograph, and the total nitrogen (T-N) concentration in water was analyzed by
absorption spectrophotometry after decomposition with potassium peroxodisulfate
(K2S2O8) (Hosomi and Sudo, 1986). Furthermore, the weight of rice in each pot was
measured at the end of the experiment after being oven-dried at 80 °C for 48 h.

- 252 -

Journal of Water and Environment Technology, Vol. 7, No. 4, 2009

forage rice

water supply
(river water)

surface area
= 0.08 m2
surface water
depth = 0.05 m

penetration

ando soil zone
depth = 0.35 m

water effluent

gravel zone
depth = 0.05 m

Fig. 1- Experimental Pot Used in This Study

Nrice

Water
supply

Middle soil Upper soil

NHsurf

Lower soil

Surface water

Mathematical Modeling
Structure
A mathematical model which describes the fate of inorganic nitrogen compounds in soil,
water and rice was developed. The model structure is shown in Fig. 2. The gravel zone
was not considered in the model.

Advection
& diffusion

NHus,s

NHus,w
Adsorption/
desorption
Advection
& diffusion

NHms, s

NHms,w
Adsorption/
desorption

Water
supply

Nitrification
Plant
uptake

NHls, s

NHls,w
diffusion

Advection
& diffusion

NOus

Plant
uptake

Denitrifi
-cation

0.05 m

Advection
& diffusion

NOms

Nitrification

Advection
& diffusion
Adsorption/
desorption

0.05 m

NOsurf
Plant
uptake

Nitrification

Plant
uptake

gas

Denitrifi
-cation

0.20 m

Advection
& diffusion
Plant
uptake

Plant
uptake

NOls

Nitrification

Denitrifi
-cation

0.10 m

diffusion

Effluent

Effluent

Fig. 2- The Structure of the Model Developed in This Study

- 253 -

0.35 m


Journal of Water and Environment Technology, Vol. 7, No. 4, 2009

The model is composed of four compartments, namely the surface water compartment
(surf), upper soil compartment (us), middle soil compartment (ms), and lower soil

compartment (ls). In each compartment, complete mixing was assumed. Basically, the
model describes the dynamics of ammonium nitrogen (NH4-N) and nitrate + nitrite
nitrogen (NO2+3-N). The adsorption of NO2+3-N on soil was ignored.
The organic nitrogen dynamics was not considered because the differences between
organic nitrogen (T-N minus inorganic nitrogen) in supplied water and in effluent was
not so significant (supplied water = 0.80±0.50 mg-N/L; effluent = 0.39±0.22 mg-N/L
(Q=0.6); 0.36±0.26 mg-N/L (Q=0.2); 0.53±0.57 mg-N/L (Q=0.1); average±SD of
observations. Q is the water supply rate [m3/(m2･day)].) compared to that of inorganic
nitrogen in the experimental results.
Furthermore, total nitrogen in rice was employed as a state variable. Above-ground rice
biomass, and underground rice biomass were calculated from the total nitrogen in rice
with a certain proportional constant obtained from our experiments.
Basic Equations
The nitrogen flow in the system can be described by the following mass conservation
equations.
dNH i
= radv , NH , i + rdif , NH ,i − rnit , i − rup , NH ,i
dt

(1)

dNOi
= radv , NO ,i + rdif , NO , i + rnit , i − rden, i − rup , NO ,i
dt

(2)

dN rice
= ∑ rup , NH , i + ∑ rup , NO , i
dt

i
i

(3)

where NHi and NOi represent NH4-N and NO2+3-N in the ith layer [g-N/m2], respectively,
and Nrice is the rice nitrogen [g-N/m2]. The terms radv,NH,i and radv,NO,i are the NH4-N and
NO2+3-N inflow into the ith layer due to advection, respectively. The terms rdif,NH,i and
rdif,NO,i are the NH4-N and NO2+3-N inflow into the ith layer due to diffusion,
respectively. The advection was calculated using conventional equations. The diffusion
in soil was calculated from the soil water content, soil porosity, and diffusion coefficient
in water based on the literature (Millington and Quirk, 1961; Shearer et al., 1973).
Adsorption
Ammonium nitrogen is adsorbed on soil. After Jury and Horton (2004), linear and
instantaneous adsorption were assumed, and the concentration of nitrogen in soil
interstitial water, CNH,i, is calculated as

- 254 -


Journal of Water and Environment Technology, Vol. 7, No. 4, 2009

C NH ,i =

(K

ads , NH

NH i
⋅ ρ s + θ s ) ⋅ Δzi

(4)

where Kads,NH is the linear adsorption coefficient (3.48×10-6) [m3/g-soil], ρs is the soil
density (440,000) [g/m3], θs is the water content (0.676) [m3/m3], and Δzi is the depth of
the ith compartment. The adsorption coefficient was obtained from an adsorption
experiment (data not shown), and the soil density and water content were determined
based on the physical properties of the soil mentioned above.
Nitrification and Denitrification
The nitrification at the surface water is described as
(T − 20 )
rnit,surf = knit,surf ⋅ χ nit
⋅ NH surf

(5)

where knit,surf is the nitrification rate constant at the surface water [/day], χnit is the
temperature constant for nitrification (1.05) [-] (based on Mayo and Bigambo, 2005),
and T is the temperature [°C]. The parameter knit,surf was calibrated in this study. The
temperature variation is described below. On the other hand, the denitrification at the
surface was ignored.
Mayo and Bigambo (2005) employed a model which considered the nitrification and
denitrification both by biofilm on soil aggregates and plant roots. Based on this report,
we assumed two mechanisms for rnit,i and rden,i as follows.

⎛
w ⎞ (T − 20 )
(T − 20 )
rnit,i = knit , i ⋅ χ nit
⋅ NH i = ⎜⎜ k nit,s,i + knit,r,i ⋅ r,i ⎟⎟ ⋅ χ nit

⋅ NH i
Δzi ⎠
⎝

(6)

w ⎞ (T − 20)
⎛
(T − 20 )
rden, i = kden ,i ⋅ χ den
⋅ NOi = ⎜⎜ k den, s ,i + kden, r ,i ⋅ r ,i ⎟⎟ ⋅ χ den
⋅ NOi
Δzi ⎠
⎝

(7)

where knit,i is the overall nitrification rate constant at the ith compartment at 20 °C [/day],
knit,s,i is the nitrification rate constant by the bacteria not attached to the root at 20 °C
[/day], knit,r,i is the nitrification rate constant by the bacteria attached to the root at 20 °C
[/day], wr,i is the root weight at the ith compartment [g/m2], kden,i is the overall
denitrification rate constant at the ith compartment at 20 °C [/day], kden,s,i is the
denitrification rate constant by the bacteria not attached to the root at 20 °C [day], kden,r,i
is the denitrification rate constant by the bacteria attached to the root at 20 °C [/day],
and χden is the temperature constant for denitrification (1.05) [-] (based on Mayo and
Bigambo, 2005). The parameters knit,s,i, knit,r,i, kden,s,i and kden,r,i were calibrated in this
study.
The root weights at the ith layer are calculated as

- 255 -

Journal of Water and Environment Technology, Vol. 7, No. 4, 2009

wr,i = ζ i ⋅ wr , tot

(8)

wr , tot = (1 − f above ) ⋅ wtot

(9)

wtot = wr ,tot + wa =

N rice
f Nw,rice

(10)

where ζi is the root abundance at the ith layer [g/g], wr,tot is the total weight of the rice
roots [g/m2], fabove is the above-ground fraction of rice (0.7) [g/g] (based on the
measurement in this study), wtot is the total weight of the rice [g/m2], wa is the weight of
the above-ground part of the rice [g/m2], and fNw,rice is the rice nitrogen / weight ratio
[g-N/g]. The variations of ζi and fNw,rice are discussed below.
Nitrogen Uptake by Rice
The rice growth rate is affected by radiation, temperature, and growth stage (Horie
1987; Horie et al., 1991), and the growth can be described by the logistic model (Sheehy,
J.E. et al., 2006). The nitrogen uptake by rice is closely related with growth. Assuming
Monod-type nitrogen uptake kinetics (Selim and Iskandar, 1981) and ignoring the
temperature effects, the NH4-N and NO2+3-N uptake by rice is described as

rup , NH ,i = k N ,up ⋅

rup , NO ,i = k N , up ⋅

f rad =

wa , p − wa
wa , p

⋅ f rad ⋅ f gs ⋅ wr ,i ⋅

C NH ,i
K N + C NH ,i + C NO ,i

wa , max − wa
C NO ,i
⋅ f rad ⋅ f gs ⋅ wr ,i ⋅
wa , max
K N + C NH ,i + C NO , i

IR
IRmax

(11)

(12)

(13)

where kN,up is the maximum nitrogen uptake by rice per unit root weight [g-N/g-root],

wa,p is the potential value of the rice above-ground biomass (3,000) [g/m2] (assumed
based on the observations in this study), frad is the radiation factor, fgs is a factor
dependent on growth stage, wr,i is the root weight per unit area [g-root/m2], KN is the
half-saturation constant for nitrogen uptake (1.0) [g/m3] (based on Selim and Iskandar,
1981), IR is the daily radiation [MJ/m2], and IRmax is the maximum daily radiation
during the experimental period [MJ/m2]. The variations of fgs are discussed below.
Forcing Functions
Some forcing functions were included in the model. The daily average temperature and
daily irradiation were obtained from a weather station near the experimental site (Japan
Meteorological Agency). The photosynthetic activity is controlled by the LAI (leaf area
index), and it varies according to the growth stage. Nitrogen uptake depends on the
photosynthetic activity, and fgs was assumed to be as shown in Fig. 3 based on the
observed variation of leaf number (see "Observation of Rice Growth" in "RESULTS

- 256 -


Journal of Water and Environment Technology, Vol. 7, No. 4, 2009

AND DISCUSSION"). The factor fNW,rice was determined as shown in Fig. 4 based on
another experimental observations (data not shown). The vertical distribution of the root
abundance, ζi, was assumed to be as shown in Fig. 5.

Factor depend on growth stage, fgs [-]

Calculation
The mass balance equations were solved numerically by the 4th Runge-Kutta methods
using STELLA ver 9.0.3J (isee systems, inc., USA). The calculation step (Δt) was set at
1/180 day. The initial condition of the nitrogen allocation was calculated by a
sufficiently long (100 days) simulation (Δt =1/100 day) without rice under the

corresponding water supply rate and the temperature on the transplantation day.

1.0
0.8
0.6
0.4
0.2
0.0
0

50

100

150

Time after transplantation [day]

Factor depend on growth stage, fNw,rice
[g-N/g]

Fig. 3- Assumed Time Course of the Factor Dependent on Growth Stage, fgs

0.05
0.04
0.03
0.02
0.01
0
0

50

100

150

Time after transplantation [day]

Fig. 4- Time Course of the Ratio Between Rice Nitrogen and Weight of Rice, fNw,rice

- 257 -


Root abundance at ith layer, ζi [g/g]

Journal of Water and Environment Technology, Vol. 7, No. 4, 2009

1.0
0.8
upper soil
middle soil
lower soil

0.6
0.4
0.2
0.0
0

50

100

150

Time after transplantation [day]

Fig. 5- Assumed Time Course of Vertical Distribution of Root Abundance, ζi
RESULTS AND DISCUSSION
Observation of Rice Growth
The time courses of leaf number and plant height are shown in Fig. 6. In the 0.6
m3/(m2∙day) case, the water level was increased from early October, and clogging was
presumed. However, the discharge rate was maintained, and we concluded that the
effect of this clogging on the rice growth was not critical. Obviously, the leaf number
and plant height were increased with the water supply rate. As described earlier, the
photosynthesis activity is controlled by the leaf area. Roughly, the plant height increased
until 120 days after transplantation, and thereafter a constant height was maintained in
every case. The leaf area index (LAI), however, increased until heading and then
decreased (Hasegawa et al., 1991). To determine whether this observed stoppage in the
growth in plant height is related to the heading, the factor dependent on growth stage, fgs,
is assumed as before (Fig. 3).
140

120
Q = 0.6
Q = 0.2
Q = 0.1
Q = 0.6
Q = 0.2

Q = 0.1

80
60

120
100
80
60

40

40

20

20

0

0
0

20

40

60

80

Plant height [cm]

Leaf number [/hill]

100

100 120 140 160

Time after transplantation [day]

Fig. 6- Time Courses of Leaf Number and Plant Height
(“Q” : river water supply rate into each pot [m3/(m2∙day)])

- 258 -


Journal of Water and Environment Technology, Vol. 7, No. 4, 2009

Rice above-ground biomass [kg/m2]

Model Predictability
The final above-ground rice biomass of each experimental case as calculated with the
finally calibrated parameters is shown in Fig. 7 along with the measured values.
Underestimations and overestimation were occurred in the low-water-supply case and
high-water-supply case, respectively. However, considering the simple structure of the
model, the predictability of the rice production seemed to be acceptable.
3.0
2.5
2.0

measurement
calculation

1.5
1.0
0.5
0.0
0.1

0.2
0.3
0.4
0.5
0.6
3
2
Water supply rate [m /(m day)]

Fig. 7- Comparison Between Calculated and Measured Above-Ground Rice Weight
Figure 8 shows the time courses of the measured and calculated NO2+3-N concentrations
in soil interstitial water and effluent water in each case. Note that the calculated
NO2+3-N concentration in effluent water from about 2 months after transplantation was
low in every case. In the model calculation, overestimation was especially observed in
the soil interstitial water in the 0.6 m3/(m2∙day) case. However, as in the rice production
prediction, the model predictability for the NO2+3-N concentrations in soil interstitial
water and effluent water was permissible considering the simple structure of the model.
The NH4-N concentration in effluent water in each case from about 2 months after
transplantation was also low (Fig. 9). In the 0.6 m3/(m2∙day) case, some measurements
in effluent water were relatively higher than others, and the model calculation was
significantly different from these values (Fig. 9). There was a possibility that these high

values were caused by the temporal change in the concentration of NH4-N in the
supplied water. In this model, however, the concentration of NH4-N in the supplied
water was assumed to be constant as was that of NO2+3-N. Therefore, the temporal
change in water quality cannot be predicted by the model. However, the general trends
in the NH4-N concentration variations were predicted by the model calculation.

- 259 -


2.5

NO2+3-N concentration [mg-N/L]

NO2+3-N concentration [mg-N/L]

Journal of Water and Environment Technology, Vol. 7, No. 4, 2009

Q = 0.1

2.0

1.5

1.0

0.5
0.0
0

20

40

60

2.5

Q = 0.2

2.0

1.5

Supplied water

1.0

Supplied water
(assumed in calc.)

0.5
0.0

80 100 120 140 160

0

20

NO2+3-N concentration [mg-N/L]

Time after transplantation [day]
2.5

40

60

80 100 120 140 160

Time after transplantation [day]

Q = 0.6

Soil interstitial water1)
(meas.)
Soil interstitial water2)
(calc.)
Effluent water
(meas.)

2.0
1.5

Effluent water
(calc.)

1.0

1) Average of 10 cm and 20 cm depth

2) The middle compartment in the model

0.5

0.0
0

20

40

60

80

100 120 140 160

Time after transplantation [day]

2.5

NH4-N concentration [mg-N/L]

NH4-N concentration [mg-N/L]

Fig. 8- Comparison Between Calculated and Measured NO2+3-N Concentration
(“Q”: river water supply rate into each pot [m3/(m2·day)])
Q = 0.1

2.0

1.5

1.0

0.5

0.0
0

20

40

60

80 100 120 140 160

NH4-N concentration [mg-N/L]

Q = 0.2

2.0

1.5

Supplied water

1.0

Supplied water
(assumed in calc.)

0.5

0.0
0

Time after transplantation [day]

2.5

2.5

20

40

60

80

100 120 140 160

Time after transplantation [day]

Q = 0.6

Soil interstitial water1)
(meas.)

Soil interstitial water2)
(calc.)
Effluent water
(meas.)

2.0

Effluent water
(calc.)

1.5

1.0

1) Average of 10 cm and 20 cm depth
2) The middle compartment in the model

0.5

0.0
0

20

40

60

80

100 120 140 160

Time after transplantation [day]

Fig. 9- Comparison Between Calculated and Measured NH4-N Concentration
(“Q”: river water supply rate into each pot [m3/(m2·day)])
Calibrated Parameters
Some parameters were calibrated based on experimental observations, and the finally
calibrated values are shown in Table 1.
The parameter knit,r,us was calibrated as 0.01 (m3/g-root)/day. Considering the root
weight during the experiment, the overall nitrification rate constant in the upper soil,

- 260 -


Journal of Water and Environment Technology, Vol. 7, No. 4, 2009

knit,us, was estimated to be <37 /day. In the same way, the overall denitrification rate
constant, kden,i, was estimated to be 40 /day, although many studies have reported
smaller values for the nitrification and denitrification rate constants (e.g., Selim and
Iskandar (1981); Chowdary et al.; Rao et al., (1984)). However, Shelbley et al. (2003)
reported values of 4.8 ~ 378 /day for nitrification and 0.48 ~ 190.8 /day for
denitrification, and suggested that such high values for nitrification and denitrification
were caused by the high water exchange rate in the soil column. They employed an
experimental soil column of 4.92 cm in diameter, and the water was supplied at a rate of
1.0±0.1 mL/min. Thus, the water supply rate was comparable to that in our study. The
other experimental conditions were slightly different (e.g., they employed an upflow
type reactor); however, the same reason for high nitrification and denitrifaction rate
constant calibrated in this study can be considered.
The maximum rice uptake rate was 0.01 (g-N/g-root)/day. Duan et al,.(2006) reported a

maximum NH4-N uptake rate of 10 μmol/(g·hr) (i.e., 0.004 (g-N/g-root)/day). A direct
comparison is difficult due to differences in the model structure, half-saturation constant,
rice species, etc. However, these two values are similar, and the calibrated value is
thought as feasible.
Table 1- Calibrated Parameter Values

parameter
k nit,surf

calibrated value
0.48

unit
/day

k nit,s,us

4.8

/day

k nit,s,ms , k nit,s,ls

0

/day

k nit,r,us

0.02

(m3/g-root)/day

k nit,r,ms , k nit,r,ls

0

k den,s,us , k den,s,ms , k den,s,ls

0.48

(m3/g-root)/day
/day

k den,r,us , k den,r,ms , k den,r,ls

0.01

k N,up

0.01

(m3/g-root)/day
(g-N/g-root)/day

Simulation of Nitrogen Dynamics
The proposed model simulates the dependency of the overall nitrogen dynamics on the
water supply rate in the experiment performed in this study. The distribution amount
within a 169-day simulation is shown in Fig. 10. Each distribution amount increased
with the water supply rate. The time courses of the denitrification, rice uptake, and

effluent rates under 0.1 m3/(m2·day) and 0.6 m3/(m2·day) water supply rates are shown
in Fig. 11. A high effluent rate occurred particularly in the early stage. The
denitrification increased with elapsed time; this occurred because the root biomass
increased and the microbiological activity was increased by the root (Eqs. (6) and (7)).
The distribution ratio ( = each distribution amount / total influent) is shown in Fig. 12.
This figure shows that the effluent ratio increased with the water supply rate. This
increment in the effluent ratio was considered to be caused by the decrease in the rice
uptake ratio.

- 261 -


Journal of Water and Environment Technology, Vol. 7, No. 4, 2009

Distribution amount [g-N/m2]

250

200
rice uptake
denitrification (lower soil)

150

denitrification (middle soil)

100

denitrification (upper soil)
effluent

50

0
0.1

0.2

0.3

0.4

0.5

0.6

Water supply rate
Fig. 10- Simulation of Nitrogen Flow for Rice Uptake, Denitrification, and Effluent

2

Distirbution rate [g-N/(m day)]

[m3/(m2∙day)]

1.2

Q = 0.1 [m3/(m2∙day)]

1.0

0.8

denitrification

0.6

rice uptake
effluent

0.4
0.2
0.0
0

50

100

150

2

Distribution rate [g-N/(m day)]

Time after transplantation [day]

1.2

Q = 0.6 [m3/(m2∙day)]

1.0
0.8

denitrification

0.6

rice uptake
effluent

0.4
0.2
0.0
0

50

100

150

Time after transplantation [day]

Fig. 11- Simulation of Time Courses of Denitrification, Rice Uptake,
and Effluent Rates (“Q”: river water supply rate)

- 262 -


Journal of Water and Environment Technology, Vol. 7, No. 4, 2009

Distribution ratio [g-N/g-N]

1.0
0.8
0.6

denitrification
rice uptake
effluent

0.4
0.2
0.0
0.1

0.2

0.3

0.4

0.5

0.6

[m3/(m2∙day)]

Water supply rate
Fig. 12- Simulation of the Each Distribution Ratio

Model Applicability
The approximate characteristics of the nitrogen dynamics in a VF constructed wetland
in which forage rice was cultivated could be predicted with the proposed mathematical
model. Such a constructed wetland can be used not only for the treatment of water but
also for the production of biomass. The outcomes obtained in the experiment and the
proposed mathematical model are quite significant. In particular, it was important that
both a high rice yield and favorable nitrogen removal were observed in the 0.6
m3/(m2∙day) case.
The model included several assumptions to simplify its structure. Future research
validating and adequately modifying these assumptions would strengthen the model’s
predictability and generality. Areas where the model can be improved are discussed
below, along with some discussion of the calibrated parameters.
The equations regarding the dynamics of roots and the relevant mechanisms employed
in the model were very simple. Obviously, it is ideal to precisely determine the growth
of roots. The assumptions regarding root distribution (Fig. 5) were made for the sake of
convenience, and were not directly based on actual observations. To determine the
validity of these assumptions and to improve the model, more detailed study of the rice
root morphology in VF constructed wetlands is necessary. This is also true of the
biofilm on the root surface. Microscopic (cellular level) nitrogen decomposition
dynamics on the root surface should be clarified to enable the model to make more
precise predictions.
As described earlier, the nitrification and denitrification rate constants were higher than
many of the values in reported in the literature except for the values reported by
Sheibley et al. (2003), and it is possible that this discrepancy seemed to be caused by
the high infiltration rate. To clarify the mechanism by which such high rate constants
can be produced, the effect of a high infiltration rate on environmental factors which
were not considered in this study (e.g., oxidation-reduction potential in soil, availability
of other compounds which affects on microbiological process, pH, etc.) should be

- 263 -

Journal of Water and Environment Technology, Vol. 7, No. 4, 2009

considered.
Consideration of organic nitrogen production and decomposition processes is desirable,
which was not included in this model. In addition, it is likely better that nitrite-nitrogen
and nitrate-nitrogen are considered separately in the nitrification and denitrification rate
processes.
As noted above, there is room for improvement in the model. However, the model
proposed in this study was adapted for high-water-supply-rate VF constructed wetlands
using forage rice. To evaluate and optimize such constructed wetland systems, the
model proposed in this study provides useful knowledge about the effects of the water
supply rate on biomass production and nitrogen removal like in Figs. 7, 10, 11, and 12.
Hereafter, many studies of the cultivation of rice not only for food or forage but also for
energy production such as bioethanol are expected. Improvement of the model based on
the outcomes of such future studies is desirable to obtain better prediction of biomass
production and water treatment and a detailed understanding of the nitrogen dynamics
in VF constructed wetlands.
CONCLUSIONS
In the experimental pots replicating VF constructed wetlands, a variety of forage rice
("Kusahonami") was cultivated using river water under various supply rates (0.1, 0.2,
and 0.6 m3/(m2·day)) for 169 days. As a result, the growth of the rice increased with the
river water supply rate. Effective removal of inorganic nitrogen was also observed in
every case. A mathematical model which predicts the fate of inorganic nitrogen
compounds and the rice growth was developed based on these experimental
observations. The increases in both the rate of nitrogen assimilation by rice and the
denitrification rate with increases in the river water supply rate were calculated using
the proposed model. Also, the model was used to calculate the decrease in the ratio of
the rice assimilation amount to the total nitrogen inflow that occurs with the increase in

the water supply rate in the pots. The mathematical model and its quantitative outcomes
may be useful in the design and management of VF constructed wetlands in which
forage rice is cultivated.
ACKNOWLEDGEMENTS
This research was supported in part by the River Fund in charge of the Foundation of
River and Watershed Environment Management (FOREM) and a Grant-in-Aid for
Scientific Research (A) (No. 19201018) from the Japan Society for the Promotion of
Science (JSPS).
REFERENCES
Behrends, L. L., Bailey, E., Jansen P., Houke, L., and Smith, S. (2007). Integrated
constructed wetland systems: Design, operation, and performance of low-cost
decentralized wastewater treatment systems., Water Science and Technology Vol.55,
No.7, 155-161.
Brix, H., and Arias, C.A. (2005). The use of vertical constructed wetlands for on-site

- 264 -


Journal of Water and Environment Technology, Vol. 7, No. 4, 2009

treatment of domestic wastewater: New Danish guidelines., Ecological Engineering
Vol.25, 491-500.
Chowdary, V.M., Rao, N.H., and Sarma, P.B.S. (2004). A coupled soil water and
nitrogen balance model for flooded rice fields in India., Agriculture, Ecosystems and
Environment Vol.103, No.3, 425-441.
Duan, Y.H., Zhang, Y.L., Shen, Q.R., and Wang, S.W. (2006) Nitrate Effect on Rice
Growth and Nitrogen Absorption and Assimilation at Different Growth Stages.,
Pedosphere Vol.16, No.6, 707-717.
Gu, B., Chimney, M.J., Newman, J., and Nungesser, M.K. (2006). Limnological
characteristics of a subtropical constructed wetland in south Florida (USA).,

Ecological Engineering Vol.27, No.4, 345–360.
Hasegawa, T., Horie, T., and Yandell, B.S. (1991). Improvement of yielding ability in
japonica rice cultivars and its impact on regional yield increase in Kinki district,
Japan., Agricultural Systems Vol.35, No.2, 173-187.
Horie, T. (1987) A Model for Evaluating Climatic Productivity and Water Balance of
Irrigated Rice and Its Application to Southeast Asia., Southeast Asia Studies Vol.25,
No.1, 63-74.
Horie, T., Yajima, M., and Nakagawa, H. (1992). Yield forecasting., Agricultural
Systems Vol.40, 211-236.
Hosomi, M., and Sudo, R. (1986). Simultaneous determination of total nitrogen and
total phosphorus in freshwater samples using persulfate digestion., International
Journal of Environmental Studies 27(3-4), 267-275.
Japan Meteorological Agency Homepage ( />Jury, W.A., and Horton, R. (2004). Chemical Transport in Soil. In: Soil Physics, 6th edn,
John Wiley & Sons, Inc., New Jersey, pp. 225-275.
Mayo, A.W., and Bigambo, T. (2005). Nitrogen transformation in horizontal subsurface
flow constructed wetlands I: Model development., Physics and Chemistry of the
Earth Vol.30, 658-667.
Millington, R.J., and Quirk, J.M. (1961). Permeability of porous solids., Transactions of
the Faraday Society Vol.57, 1200-1207.
Nungesser, M.K., and Chimney, M.J. (2006). A hydrologic assessment of the Everglades
Nutrient Removal Project, a subtropical constructed wetland in South Florida (USA).,
Ecological Engineering Vol.27, No.4, 331–344.
Rao, P.S.C, Jessup, R.E, and Reddy, K.R. (1984). Simulation of nitrogen dynamics in
flooded soils., Soil Science Vol.138, No.1, 54-62.
Sakai, M., Imbe, T., Nemoto, H., Horisue, N., Nakagawa, N., Sato, H., Hirasawa, H.,
Takadate, M., Tamura, K., Ando, I., Ishii, T., Iida, S., Maeda, H., Aoki, N., Ideta, O.,
Hirabayashi, H., and Ohta, H. (2003). A new rice variety for whole-crop silage
"Kusahonami"., National Institute of Crop Science Report No. 4, 1-15. (in Japanese)
Selim, H.M. and Iskandar, I.K. (1981). Modeling nitrogen transport and transformations
in soils: 1. Theoretical considerations., Soil Science Vol.131, No.4, 233-241.

Shearer, R.C., Letey J., Farmer, W.J., and Klute, A. (1973). Lindane diffusion in soil.,
Soil Science Society of America Journal Vol.37, No.2, 189-193.
Sheehy, J.E., Mitchell, P.L., Allen, L.H., and Ferrer, A.B. (2006). Mathematical
consequences of using various empirical expressions of crop yield as a function of
temperature., Field Crops Research Vol.98, 216-221.
Sheibley, R.W., Jackman, A.P., Duff, J.H., and Triska, F.J. (2003). Numerical modeling
of coupled nitrification-denitrification in sediment perfusion cores from the hyporheic

- 265 -


Journal of Water and Environment Technology, Vol. 7, No. 4, 2009

zone of the Shingobee River, MN., Advances in Water Resources Vol.26, No.9,
977-987.
Zhou, S. and Hosomi, M. (2008a). Nitrogen transformations and balance in a
constructed wetland for nutrient-polluted river water treatment using forage rice in
Japan., Ecological Engineering Vol.32, No.2, 147-155.
Zhou, S. and Hosomi, M. (2008b). Trends of Research and Application of Constructed
Wetlands., Journal of water and waste Vol.50, No.2, 41-50. (in Japanese)

- 266 -