ARTICLE

Prediction of root zone water and nitrogen balance in an irrigatedrice field using a simulation model

Ch. Jyotiprava Dash • A. Sarangi • D. K. Singh •

A. K. Singh • Partha Pratim Adhikary

Received: 1 February 2012 / Revised: 10 October 2013 / Accepted: 15 April 2014

� The International Society of Paddy and Water Environment Engineering and Springer Japan 2014

Abstract In irrigated semi-arid regions, knowledge of

groundwater recharge and nitrate leaching is essential for

sustainable management of water resources. In this study,

the potential recharge and nitrate leaching below the root

zone of rice crop were estimated using a root zone mod-

elling approach through simulation with HYDRUS-1D

model. The field data collected for a single season were

used for calibrating the model and validated for the next

season. The simulated results, when compared with the

measured soil water and nitrate contents at different soil

depths showed good agreement between the HYDRUS-1D

simulation and field data. The validated results indicated

that 55.5 % of the applied water percolated below the root

zone. Nitrogen balance within 120 cm deep soil profile

indicated that volatilization, denitrification, plant uptake

and leaching losses were 24.0, 18.9, 109.0 and

28.5 kg N ha-1, respectively. The magnitude of nitrate

leaching was 23.7 % of added fertilizer. Different nitrate

leaching scenarios were generated using different fertilizer

application rates exceeding the optimal dose. These sce-

narios showed an increasing trend with increase of

fertilizer rates, fitted with a second-order polynomial

equation (R2 = 0.99). The equation can be used for esti-

mation of nitrate leaching below root zone under different

fertilizer input scenarios of rice grown in similar hydro-

agro-climatic regions.

Keywords Rice � Nitrate leaching � Groundwater �HYDRUS-1D � Fertilizer use scenario

Introduction

Like many other Asian countries, rice is the main staple

food in India. Over the last decade, the compound growth

rate of rice production is 1.78 % (Economic Survey of

India 2011–2012), as against the decadal population

growth rate of 17.64 % (Census 2011). These two growth

anomalies necessitate the requirement of huge additional

quantity of rice. Further, the gradual decrease of cultivable

area hinders the horizontal growth of rice production

leaving the only option of vertical growth. Among the

various production maximization factors, the most impor-

tant is use of high dose of nutrients, especially nitrogen

(N). However, intensive use of N is not always in harmony

with ecological balance. Farmers often increase the dose

without considering the plant’s N uptake capacity. The

unused N in the soil profile mainly leaches to the

groundwater.

Nitrate leaching is a result of conspicuous interaction

among land use practices, on-ground nitrogen load,

groundwater recharge, soil nitrogen dynamics, soil char-

acteristics and depth to water table (Vinten and Dunn 2001;

Almasri and Kaluarachchi 2007). There have been a

plethora of field studies to quantitatively evaluate nitrate

leaching, which considered the effect of several factors

Ch. Jyotiprava Dash � A. Sarangi � D. K. Singh �A. K. Singh � P. P. Adhikary

Water Technology Centre, Indian Agricultural Research

Institute, PUSA, New Delhi 110012, India

Present Address:

Ch. Jyotiprava Dash � P. P. Adhikary (&)

Central Soil and Water Conservation Research and Training

Institute, Research Centre, Koraput, Odisha 763002, India

e-mail: ppadhikary@gmail.com

Present Address:

A. K. Singh

Rajmata Vijayaraje Scindia Krishi Vishwa Vidyalaya,

Gwalior, Madhya Pradesh 474002, India

123

Paddy Water Environ

DOI 10.1007/s10333-014-0439-x

including irrigation schedule, fertilizer amount, soil prop-

erties, cropping system and different tillage practices. The

results of those studies indicate that the high nitrate

leaching was often associated with excessive irrigation and

fertilizer use (Gehl et al. 2002; Sogbedji et al. 2006; Li

et al. 2007). However, these field studies are time con-

suming and cumbersome. To solve this problem, modelling

may be considered as one of the most suitable alternative

methods.

The modelling of water balance and nitrogen in rice

field becomes a challenge as rice is a highly water and

nitrogen demanding crop, therefore, poses more risk of

nitrate pollution in groundwater (Chowdary et al. 2004).

Moreover, rice is a shallow-rooted crop and the domain of

root zone is about 30 cm below the soil surface, which can

lead to considerable nitrate loss by leaching under irrigated

or high rainfall conditions. Therefore, many simulation

models have tried to describe the behaviour of nitrate in the

flooded rice fields (Antonopoulos 2008). As the water

movement in a rice field is vertical due to constant ponding

condition, one directional simulation model (HYDRUS-

1D) can be used effectively (Phogat et al. 2010). The

purpose of this study was to use a one-dimensional process-

based model HYDRUS-1D to simulate water percolation

and nitrate leaching under rice and to estimate the nitrate

leaching below the root zone under various fertilizer doses

by using the simulation of the validated model.

Materials and methods

Study area

The study was conducted to simulate the water and nitrate

movement through the vadose zone under irrigated rice in an

experimental block situated at the research farm of Indian

Agricultural Research Institute (IARI), located in Delhi,

India (28�3702200–28�390N and 77�804500–77�1002400E) with

an average elevation of 230 m above mean sea level. The

climate is tropical semi-arid with an average annual rainfall

of 700 mm and a mean annual temperature of 24 �C (average

of 30 years). About 84 % of the annual rainfall was received

during monsoon and the rest in winter. The daily meteoro-

logical data of the study area for the observation period were

obtained from the meteorological observatory, IARI.

Soil sampling and analysis

The soil samples were collected from four different depths

(0–30, 30–60, 60–90 and 90–120 cm) before transplanting

and after harvesting of the crop for two seasons namely

kharif (rainy) 2008 and kharif 2009. Soil physicochemical

parameters like bulk density, saturated hydraulic

conductivity and texture were analyzed with standard

procedure (APHA 2005). Depth-wise gravimetric soil

moisture content, nitrate and ammonium content were

determined. The soil physicochemical properties are given

in Table 1. The rice crop was transplanted and the rec-

ommended package of practice was followed.

120 kg N ha-1 was applied as urea in three splits doses

(60 kg ha-1 during transplanting, 30 kg ha-1 at 30 days

after transplanting (DAT) and 30 kg ha-1 at 60 DAT).

Standing water condition of approximately 6 cm depth was

maintained in the rice field throughout the growth period.

The crop was harvested after 120 DAT.

Model description

The HYDRUS-1D software version 5.0 (Simunek et al.

2009) developed by US Salinity Laboratory was used for

simulation of water and nitrate leaching. The simulation

processes of soil water and nitrate movement considered in

the model are described in Fig. 1. One-dimensional soil

water flow in a variably saturated, homogeneous and iso-

tropic porous medium is described by Richards’ equation

as follows:

ohot¼ o

ozKðhÞ oh

ozþ KðhÞ

� �� S ð1Þ

where h is the volumetric water content (cm3 cm-3), h is

the pressure head (cm), K is the unsaturated hydraulic

conductivity (cm day-1), z is the distance below the soil

surface (cm), t is the time (day) and S is the root water

uptake term (cm3 cm-3 day-1), which is the volume of

water removed from a unit volume of soil per unit time due

to plant water uptake. The root water uptake term is

specified in terms of a potential uptake rate and a water

stress factor (Feddes et al. 1978):

SðhÞ ¼ aðhÞ � Sp ð2Þ

where Sp is the potential water uptake rate (cm3 cm-3 -

day-1) and a(h) is the dimensionless function of the soil

water pressure head in response to water stress, taking

values between 0 and 1. Feddes et al. (1978) proposed a

Table 1 Physicochemical properties of the soil

Soil properties Value

Bulk density (Mg m-3) 1.35

Saturated hydraulic conductivity (cm day-1) 5.5

Sand (%) 37.75

Silt (%) 31.75

Clay (%) 30.50

Textural class Clay loam

Available nitrogen (kg ha-1) 250.2

Paddy Water Environ

123

piecewise linear reduction function parameterized by four

critical values of the water pressure head,

h4 \ h3 \ h2 \ h1 to describe water stress

aðhÞ ¼

h� h4

h3 � h4

; h3 [ h [ h4

1; h2� h� h3h� h1

h2 � h1

; h1 [ h [ h2

0; h� h4 or h� h1

8>>>>><>>>>>:

ð3Þ

Water uptake is at the potential rate when the pressure

head is between h2 and h3, drops off linearly when h [ h2

or h \ h3 and becomes zero when h \ h4 or h [ h1. In

general, the value of h3 is expected to be a function of

evaporative demand. The values of Feddes parameters used

in this experiment are presented in Table 2 (Phogat et al.

2010).

The soil hydraulic properties were modelled using the

van Genuchten constitutive relationships (van Genuchten

1980), represented by

hðhÞ ¼ hr þðhs � hrÞ

1þ jahjnð Þm h\0

hs h� 0

8<: ð4Þ

KðhÞ ¼ KsSle 1� ð1� S1=m

e Þm

h i2

ð5Þ

Se ¼hðhÞ � hr

hs � hr

ð6Þ

where, h(h) is the soil water retention (cm3 cm-3), hs is the

saturated volumetric water content (cm3 cm-3), hr is the

residual volumetric water content (cm3 cm-3), Ks is the

saturated hydraulic conductivity (cm day-1), Se is the

degree of saturation, a is the air-entry parameter, n is the

pore size distribution parameter with m = 1 - 1/n and l is

the pore connectivity parameter.

Prediction of nitrate leaching is made using convection–

dispersion equation (CDE). Soil nitrate fluxes are con-

trolled by physical transport, chemical interactions and

biological processes. For one-dimensional soil profile, the

CDE for solute movement is given by (Simunek et al.

2009):

oðqsÞotþ oðhCÞ

ot¼ o

ozhD

oc

oz

� �� oqC

oz� ra � lwhc� lsqs

ð7Þ

where, D is the solute dispersion coefficient (cm2 day-1),

C is the solute concentration in liquid phase (g cm-3), s is

the solute concentration in solid phase (g g-1), q is the soil

bulk density (g cm-3) and q is the volumetric flux density

(cm day-1). lw and ls are the first-order rate constants for

solutes in the liquid and solid phases (day-1), respectively,

providing connections between individual chain species, ra

is the root nutrient uptake term (equal to the product of the

sink term S in the water flow equation and the concentra-

tion of solute, g cm-3 day-1).

The dispersion coefficient in the liquid phase, Dw

(cm2 day-1) is given by (Bear 1972)

hDw ¼ DL qj j þ hD0wsw ð8Þ

where, D0w is the molecular diffusion coefficient in free

water (cm2 day-1), sw is the tortuosity factor in the liquid

phase and DL is the longitudinal dispersivity (cm).

Fig. 1 Schematic

representation of the process

considered for simulation of

nitrate leaching to the

groundwater

Table 2 Values of root water uptake reduction parameters used in

the study

Crop h1 (cm) h2 (cm) h3high (cm) h3low (cm) h4 (cm)

Rice -10 -55 -160 -250 -15,000

Paddy Water Environ

123

Initial and boundary conditions

The schematic representation of the initial and boundary

condition is shown in Fig. 2. Initial soil water content and

measured ammonium and nitrate content of various soil

layers within the flow domain were given as initial con-

dition for nitrate leaching simulation. For simulation, it was

assumed that lateral water flow along the boundaries was

zero (zero-flux boundary condition). An atmospheric

boundary condition with surface layer (as most of the time

the rice field was under submerged condition, hmax = 6 cm)

was implemented along the top of the soil surface to allow

interactions between soil and atmosphere. These interactions

included rainfall, evaporation, fertilizer, and transpiration

(root uptake) were given in the time-variable boundary

conditions. The bottom boundary was defined by a unit

vertical hydraulic gradient, and simulating free drainage as

the water table was far below the root zone. The Neumann

boundary condition for nitrate leaching was used at the

bottom of the soil profile.

Input parameters of the model

Input parameters required for water transport in HYDRUS-

1D like hs, hr and empirical factors (a, n, l) were estimated

using a pedotransfer function Rosetta Lite version 1.1

(Schaap et al. 2001). The l value was set to 0.5. Saturated

hydraulic conductivity was obtained from field experiment.

The evapotranspiration (ETc) was computed from the

product of potential evapotranspiration (ET0), crop coeffi-

cient (Kc) (Allen et al. 1998; Tyagi et al. 2003). With ETc,

potential evaporation Ep was calculated using the equation

(Pachepsky et al. 2004):

Ep ¼ ETc � exp�bLAI ð9Þ

where, b is the radiation extinction coefficient and LAI is

the leaf area index. The b value for rice was taken as 0.5

(Kiniry et al. 2001).

With the HYDRUS-1D atmospheric boundary condi-

tion, water evaporates from the soil surface at the potential

rate Ep as long as the pressure head at the surface remains

above a threshold value, hcrit. In this study, for simulation,

hcrit was assumed to be -15000 cm. The root depth was

assumed to increase with time, achieving a maximum

depth at the end of the crop development stage.

Initial nitrate and ammonium concentrations in the soil

before sowing of crop were measured. Urea was mainly

used as the N fertilizer. After its application on soil, it

undergoes a series of first-order decay reaction. In this

reaction pathway, urea is hydrolyzed to form ammonium,

which is sequentially nitrified to nitrite and nitrate and

subsequently denitrified to form N2 and nitrous oxide.

Since the nitrification from nitrite to nitrate is a much faster

reaction than nitrification of ammonium, both nitrification

reactions are often lumped, thereby neglecting nitrite

content. Therefore, urea, ammonium and nitrate were

considered for simulations.

The model input parameter for solute transport such as

longitudinal dispersivity was considered varies between 10

and 20 cm, and molecular diffusion was neglected, as it

was negligible relative to dispersion. Dispersion coefficient

(Dw) was taken from the literature (Hanson et al. 2006). In

addition to solutes dispersion coefficients, the isotherm

partition coefficient Kd (cm g-1) which linearly relates the

solute in the soil solution and in the sorption sites is

required to study the solute. While urea and nitrate- were

assumed to be present only in the dissolved phase

(Kd = 0 cm3 g-1), ammonium was assumed to adsorb to

the solid phase using a distribution coefficient Kd of

3.5 cm3 g-1. Similar values were reported in the literature

(Lotse et al. 1992; Ling and El-Kadi 1998). The rate con-

stants of different transformation processes such as urea

hydrolysis, volatilization, nitrification and denitrification

were collected from the literature and corrected during the

calibration procedure.

The first-order decay coefficient lw for urea, repre-

senting hydrolysis, was set to be 0.30 day-1 (Ling and El-

Kadi 1998; Chowdary et al. 2004). The rate constant for

volatilization was ranged between 0.04 and 0.2 day-1

(Hutson and Wagenet 1992; Chowdary et al. 2004; Sog-

bedji et al. 2006). Nitrification from ammonium to nitrate

was modelled using the rate coefficient varied between

0.02 and 0.72 day-1, and denitrification rate coefficient

was ranged between 0.01 and 0.24 day-1 (Lotse et al.

No flux boundaries

Atmospheric boundary with surface layer

hmax= 6 cm

Free drainage

25 cm

Fig. 2 Schematic representation of the initial and boundary

conditions

Paddy Water Environ

123

1992; Ling and El-Kadi 1998; Jansson and Karlberg 2001;

Chowdary et al. 2004; Antonopoulos 2008).

Model calibration and validation

The model was calibrated for the van Genuchten hydraulic

parameters such as hr, hs, a, n, l and dispersivity coefficient

for the soils of the rice field with the observed values of soil

moisture, ammonium and nitrate content pertaining to soil

depths of 30, 60, 90 and 120 cm using data of kharif 2008.

The best overall parameters were determined based on

minimum difference between observed and simulated val-

ues. With the calibrated parameters, the model was then

validated using the measured data of kharif 2009 up to

effective root zone depth of 120 cm. Further, the validated

model was simulated up to 120 cm depth below the ground

surface.

Model performance criteria

To assess the level of agreement between the model-pre-

dicted and field-observed data, four statistical procedures

were used viz. mean absolute error (MAE), root mean

square error (RMSE), Nash–Sutcliffe model efficiency (E)

(Nash and Sutcliffe 1970) and index of agreement (AI).

They were calculated using Eqs. 10, 11, 12 and 13,

respectively.

MAE ¼ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiXn

i¼1

ðPi � OiÞ=n

sð10Þ

RMSE ¼ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiXn

i¼1

ðPi � OiÞ2=n

sð11Þ

E ¼ 1�

Pni¼1

ðOi � PiÞ2

Pni¼1

ðOi � OÞ2ð12Þ

AI ¼ 1�

Pni¼1

ðOi � PiÞ2

Pni¼1

ðjO0ij � OÞ2

ð13Þ

where, Pi is the predicted values, Oi is the observed values,

O is the mean of observed values and n is the total no of

observations. The estimated values of MAE and RMSE

close to zero indicate that the model predictions are very

close to the observed values. However, model efficiency

(E) of 1 represents a perfect prediction. The index of

agreement (AI) is a measure of the degree to which the

predicted variation precisely estimates the observed vari-

ation, AI = 1 when there is a perfect agreement (Almasri

and Kaluarachchi 2007; Hu et al. 2008).

Results and discussion

Model calibration and validation

Observed data of kharif 2008 pertaining to soil depths of

30, 60, 90 and 120 cm were used as input parameters to

calibrate the model. It was assumed that up to a depth of

30 cm the soil profile was saturated and below that the soil

was under field capacity (FC). This assumption pertains to

the situation prevailing in wetland rice with deep water

table condition. The low percolation in the top soil layer in

the rice fields and occurrence of water table below 14 m

depth used to permit the moisture content of soil profile at

least up to effective root zone depth to either remain at FC

or less than FC. The calibrated hydraulic parameters for the

rice field at different depths are presented in Table 3. With

the calibrated parameters, the model was then validated

using the measured data of kharif 2009.

The model performance statistics of validation results

for water and nitrate simulation at different soil depths are

presented in Table 4. The validation dataset indicted that

there was a close agreement between measured and model

predicted soil water and nitrate content at different depths.

So, the model can be used effectively to predict the soil

water and nitrate content at various depths for rice crop.

The MAE and RMSE values for soil water content at all the

four depths were close to 0 (MAE 0.011–0.016 and RMSE

0.018–0.025). Both the MAE and RMSE values showed a

trend of gradual decrease with increase of soil depth. The

E and AI values were close to 1 and increased gradually

with increase of soil depth. All the performance parameters

clearly indicated that the simulation was better at lower

depth.

Moreover, the model performance statistics to predict

nitrate concentration at different depths also showed sim-

ilar result. The MAE and RMSE for all four depths ranged

from 0.013 to 0.021 and 0.018 to 0.024, respectively. The

MAE and RMSE values decreased and E and AI values

increased with increase in soil depth. The model efficiency

E was found maximum (0.796) at 120 cm and minimum

(0.634) at 30 cm soil depth, implying a good agreement

between observed and simulated values. Similarly, AI

values for majority of soil depths were close to 1

(0.835–0.895).

Water balance within root zone

The simulation results for soil water content obtained in

rice for different depths are depicted in Fig. 3. The

cumulative bottom flux of percolating water below 30, 60,

90 and 120 cm depths were 110, 100, 92 and 83 cm,

respectively. Further, considering 120 cm deep soil profile

as a single system, water balance was carried out. The

Paddy Water Environ

123

water balance components were irrigation, rainfall, runoff,

evapotranspiration, percolation below 120 cm depth and

change in water storage within the profile. The simulated

water balance components are presented in Table 5. The

amount of water added as irrigation (108 cm) was more

than double of the rainfall (53.2 cm). The amount of

evapotranspiration (59.8 cm) was 39 % of the applied

water indicating only less than half of the added water can

be utilized by the crops. The moisture flux percolating

below 120 cm depth was 55.6 % of supplied water. The

amount seems to be high but realistic because rice crop

need standing water condition throughout its growing

period. So there was more chance of movement of water

below the root zone. The result was found to be in close

agreement with the results obtained by Tyagi (2006) for

kharif rice at Indo-Gangetic basin (percolation

loss = 59.2 % of water supplied).

Nitrate balance within root zone

The simulated nitrogen balance below 120 cm depth for

the validation data set is presented in Table 6. The vali-

dated HYDRUS-1D was operated for the rice field and the

depth of nitrate load converged at 120 cm depth below

ground surface and the predicted amount was

28.5 kg N ha-1. In this study, the moisture content below

the root zone depth was assumed to be at FC; however,

model simulation with variable soil moisture content

acquired from field observations will provide a different

value of the moisture and nitrate leaching below root zone.

Surface application of urea in saturated soil triggered

ammonia volatilization from rice field. In this experiment,

N loss due to volatilization was calculated as 24.0 kg ha-1

(20.0 % of applied fertilizer) with application of

120 kg N ha-1 in three splits under submerged condition.

The simulated volatilization loss was slightly higher than

the results obtained from a study done by Ebrayi et al.

(2007) at IARI farm. Some earlier studies are also in prop

up with the present finding (Katyal et al. 1987; Aulakh and

Singh 1997; Parashar et al. 1998). They reported that the

volatilization loss from rice field was 20–30 kg N ha-1 for

Indo-Gangetic region. The denitrification process was

found as one of the main pathways for loss of fertilizer N in

rice, and accounted for 15.7 % of applied N during the crop

growth. As the soil is submerged during major crop

growing period, anaerobic condition developed and the

denitrification process gets initiated. High denitrification

loss of N immediately after flooding of dry soil has been

reported by Buresh et al. (1991) and George et al. (1992).

The high denitrification loss in rice is attributed to wet soil

conditions and higher temperatures during kharif season. In

the present study, the simulated loss through denitrification

(18.9 kg ha-1) was slightly lower than the results obtained

by Ebrayi et al. (2007) and Pathak et al. (2003) at IARI

farm. The low value may be due to change in microbial

composition and moisture regime within the soil profile. In

another study, Aulakh et al. (2001) estimated that 23–33 %

of the applied fertilizer N could be lost via denitrification

during the growth period of rice crop.

The depth-wise simulation of nitrate leaching is presented

in Fig. 4. Leaching was higher below 30 cm depth and

reduced substantially for the bottom layers. This may be due

to direct contact of top layer with fertilizer and standing

water, which promotes faster moving. Nitrate leaching

below root zone was found to be 28.5 kg ha-1. For the first

few days after transplanting of rice, the leaching was

Table 3 Values of soil texture, bulk density qb, residual moisture content hr, saturated moisture content hs and optimized van Genuchten soil

hydraulic parameters (a and n) used for the validation of the model

Soil

depth

Textural

class

Sand (%) Silt (%) Clay (%) Bulk density

(Mg m-3)

hr (cm3 cm-3) hs (cm3 cm-3) a n

0–30 Clay loam 41 29 30 1.33 0.06 0.427 0.016 1.372

30–60 Clay loam 37 32 31 1.35 0.06 0.426 0.014 1.390

60–90 Clay loam 36 35 29 1.36 0.06 0.426 0.012 1.423

90–120 Clay loam 37 31 32 1.35 0.06 0.427 0.015 1.377

Table 4 Model performance statistics for simulated soil water and

nitrate content at different depths for the validation data set

Depth (cm) MAE RMSE E AI

Water content

0–30 0.016 0.025 0.672 0.837

30–60 0.014 0.023 0.783 0.894

60–90 0.013 0.021 0.795 0.903

90–120 0.011 0.018 0.825 0.915

Nitrate content

0–30 0.021 0.024 0.634 0.835

30–60 0.019 0.023 0.725 0.882

60–90 0.018 0.021 0.758 0.891

90–120 0.013 0.018 0.796 0.895

MAE mean absolute error, RMSE root mean square error, E Nash–

Sutcliffe model efficiency, AI index of model efficiency

Paddy Water Environ

123

maximum (Fig. 5d), due to less nutrient and evapotranspi-

ration demand during that period. Moreover, the standing

water condition in the field created favourable atmosphere

for nitrate leaching. Leaching accounted for 23.7 % of

applied N in rice. Some earlier studies have shown that the

leaching loss was 10–25 kg ha-1 with application of

120 kg N ha-1 in rice (Katyal et al. 1987; Aulakh and Singh

1997; Parashar et al. 1998). It was also observed from the

simulated result that nitrate leaching from rice field seizes

after 100 days of transplantation (Fig. 4). The drying of the

soil that normally occurs during harvest of rice crop favours

Fig. 3 Simulated cumulative bottom flux in rice grown in a clay loam soil obtained through HYDRUS-1D model for a 0–30 cm b 30–60 cm

c 60–90 cm and d 90–120 cm depth

Table 5 Water balance within

the 120 cm soil profile

simulated by HYDRUS-1D for

rice crop grown in a clay loam

soil

Water balance

components

Amount

Rainfall (cm) 53.2

Runoff (cm) 8.2

Effective rainfall

(cm)

45.0

Irrigation (cm) 108.0

Evapotranspiration

(cm)

59.8

Bottom flux/

percolation (cm)

85.0

Percolation (% water

input)

55.5

Input (cm) 153.0

Output (cm) 144.8

Input–output (cm) 8.2

Table 6 Nitrogen balance within the 120 cm soil profile simulated

by HYDRUS-1D for rice crop grown in a clay loam soil

Nitrogen balance components Amount

Fertilizer (kg ha-1) 120.0

Atmospheric deposition(kg ha-1) 0

Mineralization (kg ha-1) 75.6

NH3 volatilization (kg ha-1) 24.0

Crop uptake (kg ha-1) 109.0

Denitrification (kg ha-1) 18.9

Nitrate leaching (kg ha-1) 28.5

Nitrate leaching (% fertilizer applied) 23.7

Input (kg ha-1) 195.6

Output (kg ha-1) 180.4

Input–output (kg ha-1) 15.2

Paddy Water Environ

123

aerobic N transformations resulting in nitrification, and

subsequent fallow period is prone to losses by denitrification

and leaching during flooding of in rice (Buresh et al. 1991;

George et al. 1992; Pathak et al. 2003).

Nonetheless, it was observed that the loss of nitrate was

substantial during the early days of crop growth due to

saturated condition of soil, less water requirement of crops,

less atmospheric demand, etc. Water application through

irrigation and rainfall and its subsequent movement

through vadose zone triggered the leaching of nitrate. The

results when compared with the previous studies corrobo-

rated that the HYDRUS-simulated results were able to

predict the leaching behaviour satisfactorily for the crop

under consideration. However, detailed analysis on vari-

able moisture content within the root zone, ammonia vol-

atilization from the surface zone, variation of soil pH and

calcium carbonate concentration which trigger the volatil-

ization, and denitrification loss for rice will help to get

more realistic results.

Nitrate leaching under different fertilizer use scenario

The validated model was simulated under varying quantity

of fertilizer i.e. with 10, 20, 30, 40 and 50 % more than the

recommended fertilizer dose in rice, keeping all other

model input parameters constant. This operation of the

model generated different nitrate leaching rates under

changing fertilizer use scenarios. The result showed that

with 10 % increase in fertilizer dose, the leaching of nitrate

below 120 cm increased to 30.5 kg N ha-1 (Table 7), i.e.

the increase was 6.9 % as compared to the recommended

practice. Whereas 50 % increase in fertilizer rate has

increased the nitrate leaching by 37.6 %. So, under these

simulated scenarios, it can be concluded that the rate of

nitrate leaching is directly proportional to increment of

fertilizer use over recommended dose. Although the nitrate

leaching increases with increase of fertilizer, the relation-

ship is not linear. Beyond 40 % higher than recommended

dose, there was a sudden increase of leaching. This may be

due to the fact that barring leaching, all other nitrogen

balance components are biologically controlled, and very

high dose of nitrate makes the system toxic. Therefore, at

nitrogen concentration 40 % higher than the recommended

dose, all other nitrogen balance components reach to the

maximum threshold limit; beyond that physic-chemical

phenomenon leaching become predominant. The amount of

nitrate leaching decreases with increase in the depth of soil

profile within the vadose zone up to the depth of model

Fig. 4 Simulated nitrate leaching in rice grown in a clay loam soil obtained through HYDRUS-1D model for a 0–30 cm b 30–60 cm

c 60–90 cm and d 90–120 cm depth

Paddy Water Environ

123

convergence. The percent increase in the recommended

fertilizer dose and the nitrate leaching depth pertaining to

model convergence was plotted and the trend line was

generated to develop a relationship between these two

parameters. It was observed that the trends were best fitted

with second-degree polynomial equation with coefficient

of determination (R2) of 0.99. The equations for the nitrate

leaching predicted by HYDRUS-1D for rice (NLR120)

below 120 cm depth under different percent increase of

recommended fertilizer dose (P) in % is

NLR120 ¼ 0:007 P2 � 0:21 Pþ 32:1 R2 ¼ 0:99� �

ð14Þ

The generated scenarios of nitrate leaching below root

zone under different fertilizer doses will be a ready surmise

of nitrate load generating from a particular cropping

system.

Conclusions

Good agreement was achieved to predict soil moisture and

nitrate leaching below root zone between the HYDRUS-1D

simulations and field measurements made for rice crop,

indicated by very low MAE and RMSE values. The model

efficiency (E) and index of agreement (AI) values pertain-

ing to prediction of soil moisture showed increasing trend

with increase in soil depth, whereas MAE and RMSE

showed decreasing trend with increasing soil depth. Model

simulations for soil water transport showed that the amount

of water percolating below root zone of rice crop was

85 cm accounting to 55.5 % of the total applied water.

Nitrogen balance within 120 cm deep soil profile indicated

that volatilization, denitrification, plant uptake and leach-

ing losses were 24.0, 18.9, 109.0 and 28.5 kg N ha-1,

respectively. The magnitude of nitrate leaching was 23.7 %

Fig. 5 Simulated nitrogen balance components in rice grown in a clay loam soil obtained through HYDRUS-1D model: a volatilization

b denitrification c uptake and d leaching

Table 7 Nitrate leaching

scenarios below root zone under

different fertilizer application

rates

RD recommended dose

Depth

(cm)

RD (120

kg N ha-1)

10 % over

RD (132

kg N ha-1)

20 % over

RD (144

kg N ha-1)

30 % over RD

(156 kg N ha-1)

40 % over RD

(168 kg N ha-1)

50 % over RD

(180 kg N ha-1)

Leaching, kg N ha-1 (percent increase in leaching)

30 35.7 38.9 (9.0) 40.4 (13.2) 42.4 (18.9) 46.0 (28.8) 51.4 (44.2)

60 29.4 31.7 (7.8) 33.2 (13.2) 34.2 (16.3) 37.2 (26.8) 41.3 (40.4)

90 29.2 31.4 (7.2) 32.4 (10.9) 33.4 (14.3) 36.2 (23.9) 40.4 (38.2)

120 28.5 30.5 (6.9) 31.2 (9.5) 31.8 (11.7) 34.8 (22.2) 39.2 (37.6)

Paddy Water Environ

123

of the added nitrogenous fertilizer. Different nitrate

leaching scenarios below root zone were generated using

different fertilizer application rates exceeding the optimal

dose up to 50 %. These scenarios showed an increasing

trend with increase of fertilizer rates, fitted with a second-

order polynomial equation (R2 = 0.99). The equation can

be used for estimation of nitrate leaching under different

fertilizer input scenarios of rice grown in similar hydro-

agro-climatic regions.

Acknowledgments The authors acknowledge Indian Agricultural

Research Institute, New Delhi for providing financial assistance in

terms of doctoral scholarship and other facilities to carry out the

research during the doctoral programme of the first author.

Conflict of interest The authors declare that they have no conflict

of interest.

References

Allen RG, Pereira LS, Raes D, Smith M (1998) Crop evapotranspi-

ration: guidelines for computing crop water requirements.

Irrigation and Drainage Paper 56, FAO, Rome, Italy

Almasri MN, Kaluarachchi JJ (2007) Modelling nitrate contamination

of groundwater in agricultural watersheds. J Hydrol 343:211–229

Antonopoulos VZ (2008) Modeling of water and nitrogen balance in

the ponded water of rice fields. Paddy Water Environ 6:387–395

APHA (2005) Standard methods for the examination of water and

wastewater, 21st edn. American Public Health Association,

Washington

Aulakh MS, Singh B (1997) Nitrogen losses and fertilizer N use

efficiency in irrigated porous soils. Nutr Cycl Agroecosys 47:1–16

Aulakh MS, Khera TS, Doran JW, Bronson KF (2001) Denitrification,

N2O and CO2 fluxes in rice-wheat cropping system as affected

by crop residues, fertilizer N and legume green manure. Biol Fert

Soils 34:375–389

Bear J (1972) Dynamics of fluid in porous media. Elsevier, New York

Buresh RJ, Datta SK, Samson MI, Phongpan S, Snitwongse P, Fagi

AM, Tejasarwana R (1991) Dinitrogen and nitrous oxide flux

from urea basely applied to puddled rice soils. Soil Sci Soc Am J

55:268–273

Census 2011 (2011) Ministry of Home Affairs, Government of India,

New Delhi

Chowdary VM, Rao NH, Sarma PBS (2004) A coupled soil water and

nitrogen balance model for flooded rice fields in India. Agric

Ecosyst Environ 103(3):425–441

Ebrayi KN, Pathak H, Kalra N, Bhatia R, Jain N (2007) Simulation of

nitrogen dynamics in soil using INFOCROP model. Environ

Monit Assess 131:451–456

Economic Survey of India (2011–2012) Ministry of Finance,

Government of India, New Delhi

Feddes RA, Kowalik PJ, Zaradny H (1978) Simulation of field water

use and crop yield. John Wiley and Sons, New York

Gehl RJ, Schmidt JP, Stone LR, Schlegel AJ, Clark GA (2002) In situ

measurements of nitrate leaching implicate poor nitrogen and

irrigation management on sandy soil. J Environ Qual 34:2243–2254

George T, Ladha JK, Buresh RJ, Garrity DP (1992) Managing native

and legume fixed N in lowland rice based cropping systems.

Plant Soil 141:69–91

Hanson BR, Simunek J, Hopmans JW (2006) Evaluation of urea–

ammonium nitrate fertigation with drip irrigation using numer-

ical modeling. Agric Water Manag 86:102–113

Hu K, Li B, Chen D, Zhang Y, Edis R (2008) Simulation of nitrate

leaching under irrigated maize on sandy soil in desert oasis in

inner magnolia, China. Agric Water Manag 95:1180–1188

Hutson JL, Wagenet RJ (1992) LEACHMN: Leaching Estimation and

Chemistry Model: A process-based model of water and solute

movement, transformations, plant uptake, and chemical reactions

in the unsaturated zone. Version 3. Department of Soil, Crop,

and Atmospheric Sciences Research, Series 92-3. Cornell

University, Ithaca, USA

Jansson PE, Karlberg L (2001) Coupled heat and mass transfer model for

soil-plant-atmosphere systems. Royal Institute of Technology,

Department of civil and environmental engineering, Stockholam, p 35

Katyal JC, Singh B, Vlek PLG, Buresh RJ (1987) Efficient N use as

affected by urea application and irrigation sequence. Soil Sci Soc

Am J 51:366–370

Kiniry JR, McCauley G, Xie Y, Arnold JG (2001) Rice parameters

describing crop performance of four U.S. cultivars. Agron J

93:1354–1361

Li Y, White R, Chen D, Zhang J, Li B, Zhang Y, Huang Y, Edis R

(2007) A spatially referenced water and nitrogen management

model (WNMM) for (irrigated) intensive cropping systems in the

North China Plain. Ecol Model 203:395–423

Ling G, El-kadi AI (1998) A lumped parameter model for nitrogen

transformation in the unsaturated zone. Water Resour Res

34(2):203–212

Lotse EG, Jabro JD, Simmons KE, Baker DE (1992) Simulation of

nitrogen dynamics and leaching from arable soils. J Contam

Hydrol 10:183–196

Nash JE, Sutcliffe JV (1970) River flow forecasting through conceptual

models, Part I, A discussion of principles. J Hydrol 10:282–290

Pachepsky YA, Smettem KRJ, Vanderborght J, Herbst M, Vereecken

H, Wosten JHM (2004) Reality and fiction of models and data in

soil hydrology. In: Feddes RA et al (eds) Unsaturated zone

modeling. Kluwer academic publishers, Dordrecht

Parashar DC, Kulshreshtha UC, Sharma C (1998) Anthropogenic

emissions of NOx, NH3 and N2O in India. Nutr Cycl Agroecosys

52:255–259

Pathak H, Timsina J, Humphreys E, Godwin DC, Singh B, Shukla AK

(2003) Simulation of rice crop performance and water, nitrogen and

carbon dynamics in northwest India using CERES Rice 4.0 model.

CSIRO Land and Water Technical Report. Griffith, Australia: CSIRO

Phogat V, Yadav AK, Malik RS, Kumar S, Cox J (2010) Simulation of salt

and water movement and estimation of water productivity of rice crop

irrigated with saline water. Paddy Water Environ 8:333–346

Schaap MG, Leij FL, van Genuchten MTh (2001) Rosetta: a

computer program for estimating soil hydraulic parameters with

hierarchical pedotransfer functions. J Hydrol 251:163–176

Simunek J, Sejna M, Saito H, Sakai M, van Genuchten M Th (2009)

The HYDRUS-1D Software package for simulating the one-

dimensional movement of water, heat, and multiple solutes in

variably-saturated media. Department of Environmental Sci-

ences, University of California, Riverside, CA

Sogbedji JM, Van Es HM, Agbeko KL (2006) Modeling nitrogen

dynamics under maize on ferralsols in Western Africa. Nutr Cycl

Agroecosys 74:99–113

Tyagi NK (2006) Irrigation and drainage management. In: Handbook

of agriculture. Indian Council of Agricultural Research, New

Delhi, India, pp 377–394

Tyagi NK, Sharma DK, Luthra SK (2003) Determination of evapo-

transpiration for maize and berseem clover. Irrig Sci 21:173–181

van Genuchten MTh (1980) A closed-form equation for predicting the

hydraulic conductivity of unsaturated souls. Soil Sci Soc Am J

44:892–898

Vinten AJA, Dunn SM (2001) Assessing the effects of land use on

temporal change in well water quality in a designated nitrate

vulnerable zone. Sci Total Environ 265:253–268

Paddy Water Environ

123