You are on page 1of 17

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/273755489

A Coupled Mathematical Model to Predict the Influence of Nitrogen


Fertilization on Crop, Soil and Groundwater Quality

Article in Water Resources Management · December 2014


DOI: 10.1007/s11269-014-0664-5

CITATIONS READS

31 537

2 authors, including:

Irina Marinov
University of Pennsylvania
54 PUBLICATIONS 4,164 CITATIONS

SEE PROFILE

All content following this page was uploaded by Irina Marinov on 19 March 2015.

The user has requested enhancement of the downloaded file.


Water Resour Manage (2014) 28:5231–5246
DOI 10.1007/s11269-014-0664-5

A Coupled Mathematical Model to Predict the Influence


of Nitrogen Fertilization on Crop, Soil and Groundwater
Quality

Irina Marinov & Anca Marina Marinov

Received: 17 May 2013 / Accepted: 5 May 2014 /


Published online: 10 October 2014
# Springer Science+Business Media Dordrecht 2014

Abstract Intensified use of nitrogen based fertilizers in agriculture has resulted in a significant
increases in soils and groundwater nitrate concentrations all over the world. Here we propose a
new coupled model which describes the physical transport and biogeochemical dynamics of
the water and nitrogen compounds in a soil-water-plant-groundwater system. Our model takes
into account water infiltration into an unsaturated porous medium, the complex biogeochem-
ical cycle of nitrogen in soils, nitrate leaching from the agricultural system toward the aquifer’s
water table, and the dispersion of nitrates in the groundwater. We calibrate our model and
analyse the influence of soil type, precipitation or irrigation regime and fertilization schedules
on leaching to groundwater as well as the temporal and spatial evolution of the nitrate pollutant
plume in the aquifer. Simulations indicate that in order to achieve high crop yields while
minimizing nitrogen loading to soils and groundwater we need to create an optimal balance
between the amount of chemical fertilizers and water applied to crops on one hand, and the
amount of nitrate and water used by plants on the other. We find that medium soils are more
suitable for a sustainable corn production than coarse soils, ensuring both higher yield and less
nitrate pollution of the aquifer. Regardless of soil type and irrigation schedule, a gradual
fertilization throughout the plant life cycle reduces the potential for leaching and aquifer
pollution. For medium soils, a more modest irrigation schedule results in more nitrate available
for crops and less net leaching to the groundwater.

Highlights Design new coupled model for agricultural management:


nitrogen transfer in unsaturated soil-water-plant system,
nitrate dispersion in groundwater.
Electronic supplementary material The online version of this article (doi:10.1007/s11269-014-0664-5)
contains supplementary material, which is available to authorized users.
I. Marinov (*)
Department of Earth and Environmental Science, University of Pennsylvania, 240 S, 33rd Str, Hayden Hall
254B, Philadelphia, PA 19104, USA
e-mail: imarinov@sas.upenn.edu

A. M. Marinov (*)
Department of Hydraulics, Hydraulic Machinery and Environmental Engineering, University Politechnica of
Bucharest, Splaiul Independenţei, Nr. 313, Bucharest 060042, Romania
e-mail: ancamarinov@gmail.com
5232 I. Marinov, A.M. Marinov

Keywords Groundwater pollution . Chemical fertilizers . Irrigation . Nitrogen leaching .


Hydrological modeling . Sustainable agriculture

1 Introduction

Nitrate is the most widespread of all groundwater contaminants. Irrigated agriculture


with excessive nitrogen (N) fertilizers can result in excessive nitrate concentrations in
groundwater above the 10 mg/L of N-NO3- limit established by the World Health
Organization and the US EPA for drinking water safety. Above this limit, nitrate in
groundwater can cause health problems such as blue baby syndrome, methemoglobi-
nemia, non-Hodgkin lymphoma (Ward et al. 1996; Su et al. 2013) and potentially
miscarriages (Schubert et al. 1997).
Fertilizers typically used in agriculture contain a mixture of nitrate (NO3), ammonia (NH4)
and organic nitrogen. When applied to the soil, these fertilizers are converted into the highly
soluble nitrate form, which is readily taken by plants in the top root zone (L1 in Fig. 1a). If the
quantity added exceeds the plant uptake, the remaining nitrate is leached out of this layer and
gradually penetrates the next layer (L2 in Fig. 1a), eventually reaching the groundwater table
(GWT). Indeed, it has been observed that crops use approximately 50% of applied nitrogen
(Tilman et al. 2002). Three main grain crops (rice, corn, and wheat) account for 56% of world
N fertilizer consumption (Cassman et al. 2003), hence an urgent concern with improving the
efficiency of agriculture to reduce nitrate loads.
Recent observational studies have highlighted a positive correlation between groundwater
nitrate and N rate applications (Costa et al. 2002), for example West Bengal villages using a
higher rate of N-fertilizer recorded a higher concentration of nitrate in their groundwater
(Kundu and Mandal 2009). Some improvement in the quality of surface waters was obtained
upon lowering nitrate surpluses in agriculture in developed countries (Kumazawa 2002;
Oenema et al. 2005). Excessive rain or irrigation can increases the groundwater free surface
level and its vulnerability to nitrate pollution (Nolan et al. 2006) while environmental
conservation practices can reduce runoff and transport of nutrients at the watershed level
(Rocha et al. 2012).
Romania covers an area of roughly 23.8 mil ha, with 39% (9.4 mil ha) agricultural land and
about 3.1 mil. ha irrigated land. Intensified use of nitrogen based fertilizers in agriculture has
resulted in a significant increase in nitrates in soils and groundwater over the past few decades.
The first estimate of nitrate vulnerability of Romanian soils performed in 2003 established 255
areas (a total agricultural land area of 6500000 ha) as “nitrate vulnerable zones”. Among the
potentially vulnerable regions are parts of the Balta Brailei region, in the S-E Plains of
Romania, characterized by intensive irrigation and fertilization applied since 1973 and an
aquifer close to the surface.
The development of models that simulate N processes in soils and evaluate
environmental impacts associated with nitrogen management is now recognized as
being critical for improving cropping technique and cropping systems, optimizing
irrigation scheduling practices (Rinaldi 2001) and more generally the irrigation effi-
ciency in agriculture or horticulture (Greenwood et al. 2010; Letey and Vaughan
2013). An in depth understanding of the interactions between irrigation method,
nutrient application, crop root distribution and nutrient uptake will help improve water
management practices (Hopmans and Bristow 2002).
Several one dimensional models have been proposed in the literature to simulate water flow
and solute transport, crop uptake of water and nutrients, as well as the biological and chemical
A Coupled Mathematical Model to Predict the Influence of Nitrogen 5233

Legend

L1-unsaturated soil layer with roots,


P (or IRIG) ETP L2-unsaturated soil layer without
roots,
GW-groundwater,
SW SWP-soil-water-plant system,
Z1- height of L1,
Z2- depth of water-table under the
QR soil surface,
P-rain intensity,
z1 QI IRIG-irrigation intensity,
z2 L1 ETP-evapotranspiration,
QIR QR-surface runoff,
L2 QI- water discharge entering the soil
QE through a unit surface,
GW QIR- water discharge between L1
and L2,

(a)

Input data for Input data for BNTM Input data for
WTUM WTGM
P, ETP, z1,z2, kd, kam, kra, kni, krn, kab, B.
WP, total fertilizers amount: FNIT, Dispersivity
crop type, FAM, FORG, fertilizer α L αT
soil type, spreading schedules, Boundary
soil properties: initial conditions: N-NO3 (0), conditions
ρb, C, Cw, θr, N-NH4+ (0), N-NO3n (0), Initial conditions
θsat Norg(0). Velocities vx. vz

Water Hydraulic Biochemical BNTM Water and


transport in results: nitrogen results: nitrate
unsaturated transfer Model transport in
soil-Model IRIG, N-NO3 , groundwater
WL1, (BNTM) N-NH4+, Model
(WTUM) WL2, QI, N-NO3n,
QIR, QE Norg (WTGM)

(b)

Fig. 1 a Schematic of the coupled model for aquifer-plant-soil system (CMAPS) b Components of the coupled
model CMAPS

transformation of nitrogen in soil (e.g. Hutson 2000; Antonopoulos 2001; Rinaldi 2001). A
growing number of modelling studies are considering the effect of soil type and crop
management on nitrate leaching. Among the most sophisticated we mention the HYDRUS-
2D studies of Gardenas et al. (2005) and Hanson et al. (2006) or the STICS soil-crop model
study of Jego et al. (2008). Gardenas et al. (2005) investigated the effects of soil type and of the
timing between the fertigation and irrigation on nitrate leaching from citrus, grape, tomato, and
strawberry fields using HYDRUS-2D. Hanson et al. (2006) analyse the effects of a combina-
tion of micro-irrigation systems, different fertigation strategies and soil profiles on soil nitrate
and nitrate leaching. Rocha et al. (2012) use the SWAT model to predict the impact of soil
management practices on watersheds, under variable soils, land uses.
Here we study the behaviour of a soil-water-plant system connected with the groundwater
(Fig. 1a) considering a 90,000 m2 surface cultivated with corn and irrigated at the surface.
5234 I. Marinov, A.M. Marinov

Building on our previous work (Marinov and Diminescu 2008), we design a new coupled
model (Fig. 1b) for the aquifer-plant-soil system (CMAPS) which combines sub-models for
water transport in unsaturated soils, nitrogen cycling in soil and pollutant dispersion in
groundwater. Our specific objective is to investigate via numerical simulations the effects of
soil type, irrigation rate, and nitrogen fertilizer application rate on nitrogen leaching toward
groundwater and on the behaviour of the pollutant plume in the aquifer.
Since meteorological and hydrogeological conditions, soil type, fertilization and irrigation
rates used in the study are characteristic for the Balta Brailei region of Romania, this study can
be used to understand the potential vulnerability of this region to nitrogen pollution. We
propose that our model can be used to search for good agricultural management strategies in
order to (i) develop ways to use scarce water resources most efficiently while maximizing plant
nitrate uptake and crop production and (ii) minimize the N accumulation below the root zone
and in groundwater.

2 Methods: Model Description

This section describes the new coupled model for the aquifer-plant-soil system
(CMAPS) schematically described in Fig. 1-. The model contains three modules: a
1D water transfer model for the unsaturated soil (WTUM), a biochemical nitrogen
transfer model for the soil (BNTM), and a 2D model describing pollutant dispersion
in the groundwater (WTGM)
Plants develop their roots in the unsaturated layer (L1) and the root layer depth z1
depends on crop type. The excess nitrates from layer L1 can leach through the
unsaturated layer without roots (L2) to the groundwater table (z=z2). The hydraulic
model WTUM calculates the suction in the soil pores (h), the volumetric water
content (θ), the total water quantity (WL1) in the roots region (0-50 cm), and the
total water quantity (WL2) in the below-roots region (50–150 cm). WL1, WL2, the
discharges (QI) entering the soil surface from rain and irrigation (diminished by
evapo-transpiration), and the discharge (QIR) flowing through the unsaturated region
(50–150 cm), influence the nitrogen transformations and the quantity of nitrates
reaching the watertable. Nitrate leaching through the aquifer’s free surface is a
pollution source for groundwater. Nitrate concentration in the unconfined aquifer will
depend on leachate concentration, aquifer’s hydrogeological characteristics, and
boundary conditions.

2.1 Hydraulic Model: Water Transport in Unsaturated Soils

The isothermal one-dimensional fluid flow in an unsaturated porous soil is described by Richards’
equation, which combines the mass conservation equation with the generalized Darcy law:

  
∂h ∂ ∂h
CS ðhÞ ¼ K ðhÞ 1 ; ð1Þ
∂t ∂z ∂z

where CS(h)=dθ/dh is the specific soil water capacity, θ (z,t) [L3L-3] is the soil’s
volumetric water content, h (z,t) [L] is the water pressure head in soil pores, K (h) [L
T-1] is the hydraulic conductivity, z [L] is soil depth taken positive downwards, and t
[T] is time. The pressure of an unsaturated soil is always negative relative to
A Coupled Mathematical Model to Predict the Influence of Nitrogen 5235

atmospheric pressure: h<0. CS (h) and K (h) are time dependent functions of the local
soil water pressure head, h (z,t).
We use the mathematical relationships for soil water retention θ(h) and unsaturated soil
hydraulic conductivity (K) proposed by van Genuchten (1980):
 m
1
θðhÞ ¼ θr þ ðθs  θr Þ ð2Þ
1 þ ðαjhjÞn

h  m i2
K ¼ K s S e 1=2 1− 1−S e 1=m ð3Þ

where
θ−θr 1
Se ¼ ¼ ð4Þ
θs −θr ½1 þ ðα abs ðhÞÞn m
for h < 0, or Se = 1 for h > 0 or h = 0, θ (cm3cm-3) is the soil volumentric water content, h (cm)
the pressure head in soil pores, θr (cm3cm-3) the residual soil water content, θs (cm3cm-3) the
saturated soil water content, and α, n, m empirical soil water retention factors defining the
moisture retention characteristic shape. The θr, θs, α, n, m parameters are estimated from
observed soil water retention data and m=1−1/n. We use Van Genuchen’s optimized param-
eters corresponding to the different texture classes studied here (Table 1).
Equation 1 is numerically integrated using a fully implicit finite differences, stable and conver-
gent scheme (Marinov and Diminescu 2008) to calculate the pressure head h (z,t) and the hydraulic
conductivity K (z,t) in the two layered unsaturated soil profile shown in Fig. 1a. The water discharges
entering the soil through a unit surface, at depths z=0, z=z1 and z=z2 are computed as follows:
 
∂h
QIðtÞ ¼ −K þ K ¼ q0 ðtÞ ð5Þ
∂z z¼0;t

 
∂h
QIRðt Þ ¼ −K þ K ð6Þ
∂z z¼z1 ;t

 
∂h
QEðt Þ ¼ −K þK ð7Þ
∂z z¼z2 ;t

Table 1 Soil characteristics and optimized Mualem-van Genuchten parameters: saturated θs and residual θr soil
water content, α, n, m empirical soil water retention factors, relative bulk density ρb, wilting point Cw (%) m and
field capacity Cf (%) m in percent of the dry soil mass

Soil Type texture layer Ks cm/ θs θr α n ρb Cw (%) Cf (%)


hour (%) (%) m m

Chernozem medium Topsoil 12.061 43.9 1.0 0.0314 1.1084 1.34 12 24


medium Subsoil 10.755 39.2 1.0 0.0249 1.1689 1.34 12 24
Alluvial Loamy coarse Topsoil 60.00 40.3 2.5 0.0383 1.3774 1.31 8 22.2
sand
coarse Subsoil 70.00 36.6 2.5 0.0430 1.5206 1.31 8 22.2
5236 I. Marinov, A.M. Marinov

Given a rain or irrigation intensity P (cm/hr), the surface runoff QR (cm/hr) is calculated as:
QRðtÞ ¼ Pðt Þ−QIðtÞ ð8Þ
The water content in the roots region, WL1 (L3/L2), and the water content in the unsaturated
region, without roots, WL2 (L3/L2), are given by:
Zz¼z1
WL1 ðtÞ ¼ θ ðz; t Þ dz ð9Þ
z¼0

Zz¼z2
WL2 ðtÞ ¼ θ ðz; t Þ dz ð10Þ
z¼z1

and we define total water content in the soil as WTOT = WL1+WL2. These hydraulic quantities
are the input to the biogeochemical model described below.

2.2 Biogeochemical Model

Multiple biogeochemical models have been developed to explain N processes in soils (for a
review, see Cannavo et al. 2008). The biogeochemical model we employ follows Geng (1988)
and Geng et al. (1996) and is presented here for completeness and illustrated in Fig. 2. The
model solves for nitrate (N-NO3-), ammonium (N-NH4+), and organic nitrogen (Norg) in the
root layer and nitrate in the below-root layer (N-NO3n–) in units of kg/ha. Sources of nitrogen
in the soil come from plant residues decay, microbial and animal material decay, and
fertilization. Nitrogen transformations occur by means of mineralization (ammonification),
immobilization (reorganization), nitrification and denitrification. We assume no soil-
atmosphere nitrogen exchange. The novel element of our model is that the transfer functions

ETP P (IRIG)

FORG FAM FNIT plant QR

QI

Norg NH4l +
NO3 - Np

roots WP
layer
z1 NH4s+ (QI/WL1)NO3- L1
WL1

z2 NO3n- QIR
L2 -
(QIR/WL2)(NO3n ) WL2

GW groundwater V QE

Fig. 2 The biogeochemical submodel


A Coupled Mathematical Model to Predict the Influence of Nitrogen 5237

between different nitrogenous forms (Fig. 2) depend on the system’s hydraulic parameters:
precipitation or irrigation intensity (P), surface runoff (QR), and the moisture retention
properties of the soil (WL1, WL2) calculated from the unsaturated flow model (Eqs. 9–10).
For the root layer (L1) we write:

d N−NO−3
¼ ð1− QR=PÞ FNIT −ðQI=WL1 ÞðN−NO3 − Þ þ ½kni =ð1 þ kd ÞðN−NH4 þ Þ
dt
 krn ðN−NO3 − Þ−B½ðN−NO3 ─ Þ=ððN−NO3 ─ Þ þ kab Þ fpp ’ ðt=TÞ (11)


d N–NHþ4
¼ FAM þ kam ðNorgÞ  ½kni =ð1 þ kd ÞðN−NH4 þ Þ  ½kra =ð1 þ kd ÞðN−NH4 þ Þ
dt
ð12Þ

dðNorgÞ
¼ FORG þ krn ðN−NO3 ─ Þ þ ½kra =ð1 þ kd ÞðN−NH4 þ Þ  kam ðNorgÞ ð13Þ
dt
and for the below-root layer (L2):

d N  NO−3n
¼ ðQI=WL1 ÞðN−NO3 − Þ  ðQIR=WL2 ÞðN−NO3n − Þ; ð14Þ
dt
where parameters kam, kni, kra, krn are the rate coefficients of ammonification, nitrification, and
immobilizations from ammonium and nitrate, respectively, obtained by model calibration.
Dimensionless distribution coefficient kd is a constant here, but can more generally depend on
soil density and moisture content. B is the maximum nitrate plant uptake during its life cycle.
FNIT, FAM, FORG are the prescribed nitrate, ammonium and organic nitrogen fertilizer applica-
tion rates at the soil surface (kg/ha/day). The biochemical model parameters are chosen as
kd=0.35, kam =0.002 month-1, kra=0.1 month-1, kni=0.2 month-1, krn=0.1 month-1, kab=0.4 kg
ha-1, B=240 kg ha-1/year, following the detailed biogeochemical model calibration performed
by Geng (1988) and Geng et al. (1996).
Nitrate will only penetrate the soil dissolved in water, so given an applied nitrate fertilizer
rate FNIT, only a percent proportional to the effective water penetrating the soil (rather than
leaving the system via runoff), 1-QR/P, will actually enter through the soil surface (RHS of
Eq. 11). If there is no runoff, QR=0 and the entire applied nitrate fertilizer enters the soil. The
ammonium and organic fertilizers are not soluble so the prescribed fertilizer application rates
(FAM, FORG) contribute fully to nitrogen cycling in the top-soil layer.
The total ammonium content in soil from Eq. (12) can be written as [N–NH+4 ]=[N–
NH+4 ]a +[N–NH+4 ]s, where [N–NH+4 ]a and [N−NH+4 ]s are respectively the nitrogen content in
the ammonium adsorbed phase and solved phase in kgN/ha and the following holds: [N–
NH+4 ]a =kd[N–NH+4 ]s =kd[N–NH+4 ]/(1+kd).
The nitrogen uptake by crops increases with nitrogen availability, following a Monod type
curve (last term in Eq. 11), where B is the maximum nitrate uptake by the plant during the life cycle,
kab is a half saturation coefficient characteristic for the crop, t is time since germination started and
0
0 1 d F pp ð t 0 Þ
f pp ð t=T Þ ¼ ⋅ ; t 0 ¼ t=T : ð15Þ
T d t0
Here F’pp is a prescribed function of cumulated nitrogen uptake (day-1) over the life span of
the crop, T, assumed to be the same over all soil-climate conditions for a given crop; the value
5238 I. Marinov, A.M. Marinov

for corn comes from Watts and Hanks (1978). Geng et al. (1996) showed that this model
simulates the N cycle and N leaching reasonably well for both wheat and corn crops under a
variety of climate and soil conditions. Equations (11)-(14) are solved numerically in Matlab,
using inputs from the hydraulic model for unsaturated soil.

2.3 Water Transport in Groundwater Model

We consider an homogeneous, unconfined aquifer with constant speed vx, in the horizontal x
direction (vy =0,vz =0). Under these conditions the partial differential equation that governs
three-dimensional transport of a conservative constituent in groundwater, considering advec-
tion, dispersion and fluid sources reduces to:
   
∂C ∂ ∂C ∂ ∂C
R ¼ Dxx  vx C þ Dzz þ Sr ð16Þ
∂t ∂x ∂x ∂z ∂z

in the x-z plane and an equivalent equation in the x-y plane. C (x,z,t) [ML-3] is the time and
space variable pollutant concentration (Fig. 3), Sr is an additional source and R is the
retardation factor (Fetter 1992). Nitrate N-NO3n leached from the soil above the aquifer is
imposed as a boundary condition for C at each time step. Dxx is the dispersion coefficient of the
pollutant in the x direction (longitudinal) and Dzz is the dispersion coefficient in the vertical z
direction given by Dxx ¼ αL ! v ; Dzz ¼ αT ! v .
We assume a 300 m wide corn field and we consider the dispersal of the leached nitrogen
pollutant into a 1000 m wide and 10 m deep section of an aquifer. We assume a longitudinal
dispersivity equal to (1/10) of the field length αL =30 m, and a vertical dispersivity
αT =(1/100 ) αL = 0.3 m (Gelhar et al. 1992). The model integrates equation (16) using the
Alternating Directions Implicit Procedure.

3 Designing Our Model Experiment: Soil Characteristics and the Irrigation-Fertilization


Schedule

Our new coupled model (CMAPS) is used to analyze nitrogen addition via fertilization in an
irrigated corn field (Fig.1a) of size (300 m/300 m). The irrigation system Cazasu is part of the
agricultural region Balta Brailei in the S-E Plains of Romania and has been active since 1973.
The soils in the region are of the Chernozem type (loamy sand) and alluvial soils and the
aquifer contains silt and fine sand. The aquifer has a thickness of 3–15 m and is found at
depths of 1.5–5 m below the surface. The aquifer has velocities of the order 5–10 m/day and
flows in the NW to SE direction into the Danube River. Regional average precipitation and
evapotranspiration are 470–560 mm/year and 670–740 mm/year, respectively. The region is
favourable to corn production.
Soil properties for the two standard types of soils from our agricultural area are shown in
Table 1; we will refer to these soils as “medium” and “coarse” from here onward. Each soil
profile contains a topsoil layer (top 20 cm) and a subsoil layer (everything below 20 cm).
A goal of our agricultural management is to establish an irrigation and fertilization
regime such that (1) the water and nitrate reserve in the root layer (L1) are larger
than the plant requirement at all times and (2) the aquifer pollution with nitrate is
minimal.
Detailed meteorological and water crop demand data typical for the area are shown in
Table 2. Monthly irrigation applied (Irrigation schedule I1) takes into account precipitation
A Coupled Mathematical Model to Predict the Influence of Nitrogen 5239

Fig. 3 Irrigation amount P and soil water content in the top layer (WL1), bottom layer (WL2) of soil and in total
soil column (WTOT), in cm, for medium soil: a irrigation schedule I1, b irrigation schedule I2; for coarse soil: c
irrigation schedule I1, d irrigation schedule I2

(PR), potential evapotranspiration (ETP) and optimal water plant need (WP): I1 = PR - (ETP+
WP). WP is characteristic for our particular regional climate and the type of corn planted and

Table 2 Meteorological, water crop demand data, and irrigation schedules I1 and I2

J 1 2 3 4 5 6 7 8 9 10 11 12

Month of the year IV V VI VII VIII IX X XI XII I II III


PR mm/(m2month) 65 52 76 62 54 50 50 50 30 30 35 60
ETP mm/m2month) 50 56 85 90 90 65 50 50 30 20 20 35
WP mm/(m2month) 45 56 111 192 174 90 0 0 0 0 0 0
I1 mm/(m2month) 30 60 120 220 210 105 0 0 0 10 15 25
I1 m3/(ha month) 300 600 1200 2200 2100 1050 0 0 0 100 150 250
I2 mm/(m2month) 60 120 240 440 420 210 0 0 0 20 30 50
I2 m3/(ha month) 600 1200 2400 4400 4200 2100 0 0 0 200 300 500
5240 I. Marinov, A.M. Marinov

follows (Lazaroiu A et all 2008). Irrigation schedule I2 has identical timing with I1 but uses
twice the amount of water at all times.
Nitrogen consumption is 200–240 kg/ha for a corn production of 10,000 kg/ha (Lazaroiu A
et all 2008). We thus propose two fertilization scenarios. Schedule F1 corresponds to an
addition of 100 kg/ha ammonium and 100 kg/ha nitrate once a year on the first day the corn is
planted, corresponding to April 1st of each year. Schedule F2 uses the same total amount of
fertilizer as F1 but distributes it over the year, based on the detailed needs of the corn crop
through its various life stages (Lazaroiu A et all 2008), as detailed in Table 1 ESM.

4 Results and Discussion

In summary, 8 scenarios are considered over a 3 year period for two different irrigation
schedules (I1 and I2), two fertilization strategies (F1 and F2), and two texture classes of soil
characteristic for the region: coarse and medium. For each scenario we simulate and analyse
water seepage through soil, nitrogen biogeochemistry in soil, nitrate leaching to the water table
and nitrate dispersion in the aquifer.
Figures 3, 4, 5 and 6, Table 3 and Tables 2ESM -6ESM summarize our results.
Table 3 shows the total annual N leaching to the aquifer, calculated as the product of
N-NO3n- concentration and water discharge out of soil layer L2, for each of the
3 years and as a total amount over the 3 years. Of the total fertilizer applied, the
annual loss via leaching varies from 5.6% (for medium soil, I1, F2) to 53% (coarse
soil, I2, F1) depending on the irrigation and fertilization type, with the losses
increasing over time. In agreement with previous work (e.g. Gardenas et al. 2005),
we observe both higher seasonal leaching and higher aquifer nitrate pollution for
coarse-textured soils (Tables 2ESM-3ESM) compared to the medium soils, because the
texture of the coarse soil does not restrict downward movement of water containing
nitrate. Additionally, immediate loss of water from the top layer of the coarse soil
(Fig. 3c, 3d) means that WL1 can decrease below the minimum water amount
recommended for the corn crop, Cmin. This is not the case for the medium soil,
(Fig. 3a, 3b) where WL1 is sufficient for the crop at all times in our simulations, i.e.,
WL1 > Cmin. We calculated these critical water values as Cmin= Cw+1/2 (Cf -Cw) for
medium soils and Cmin= Cw+1/3 (Cf -Cw) for coarse soils, where Cw and Cf are the
wilting point and field capacity representative for each soil type (Table 1).
Our simulations suggest that medium soils are superior for corn production, ensuring less
wilting, higher yield and less aquifer pollution.
Increased irrigation (compare I2 with I1) increases the water content of both soil layers
and the total nitrate leaching to the aquifer (Table 3). Larger irrigation both (a) increases
the net amount of nitrate transferred to the aquifer and (b) dilutes the nitrate concentra-
tion. It is not clear apriori which of these effects will win. For our particular site and
model setup the dilution effect wins, such that for a given fertilization schedule and soil
type, a more modest irrigation discharge (I1) is associated with higher subsoil nitrate N-
NO3n concentration, higher nitrate concentration in the aquifer (Tables 2–3 ESM), and
lower total nitrate transferred to the aquifer.
Fertilization at the beginning of the plant life cycle (scenario F1) tends to increase both
seasonal nitrate leaching and nitrate concentration in the aquifer, while fertilization throughout
the plant life cycle (F2) reduces the potential for leaching and aquifer pollution for all soil types
and irrigation regimes. Fertilization scenario F1 is inadequate, as plant nutrient demand is
small when the plants are small and hence nitrate can immediately leak out of the root layer or
A Coupled Mathematical Model to Predict the Influence of Nitrogen 5241

Fig. 4 Concentration of nitrate (N-NO3), ammonium (N-NH4) in the top root layer of the soil, nitrogen plant
demand (Nplant) and well as nitrate below the root layer (N-NO3n) in kgN/ha for three scenarios. Nplant is the
ideal nitrogen uptake of the plant and is given by equation 15
5242 I. Marinov, A.M. Marinov

Fig. 5 Net leakage of nitrate (N-NO3n) from the soil into the aquifer in kg/(ha day) for coarse and medium soil,
for all fertilization and irrigation schedules analysed
A Coupled Mathematical Model to Predict the Influence of Nitrogen 5243

Fig. 6 Concentration of the nitrate pollutant in the aquifer in mg/L at the end of year 1 (continuous red line) and
at the end of year 3 (dashed line) for: a the worst scenario: coarse soil, fertilization F1 and Irrigation I1, and for b
the best scenario with lowest nitrate concentration in groundwater: medium soil, fertilization F2 and Irrigation I2.
The pollution source is at z=10m (watertable), between x=200 m and x=500 m. Longitudinal dispersivity is αL
=30m. Sensitivity of model results to αL is shown in Figs. 9ESM and 10ESM

be converted into organic nitrogen (Norg). As a consequence, nitrate supply in the root layer
will become insufficient for the plant at the next growth stage, resulting in small crop yield.
By contrast, timing the fertilization to correspond to the period of maximal plant need (F2)
increases the plant uptake and crop yield, and minimizes nitrate leaching to the ground in the
first year (compare Figs. 4a and e).
The maximum amount leached and hence the worst case from the point of view of the
aquifer pollution is fertilization F1 and Irrigation schedule I2 for coarse soil (Fig.5c). In this
case 53% (101.81 kg/ha) of the N fertilizer added is lost to the aquifer in year 1, and similar
amounts in years 2 and 3. The least amount leached (Table 3) and hence the lowest
5244 I. Marinov, A.M. Marinov

Table 3 Total (time integrated) nitrate leached from the soil to the aquifer in (kg/ha) as a function of soil type,
irrigation regime and fertilization schedule

Time interval (days) Soil Texture - Coarse Soil Texture - Medium

Fertilization-F1 Fertilization-F2 Fertilization-F1 Fertilization-F2

Irrig.I1 Irrig.I2 Irrig.I1 Irrig.I2 Irrig.I1 Irrig.I2 Irrig.I1 Irrig.I2

0-365 65.74 101.81 29.75 44.69 20.39 44.68 11.19 23.46


366-731 73.66 100.53 44.90 60.02 27.71 45.28 24.49 42.96
732-1096 83.92 106.56 51.39 64.77 29.29 43.18 32.74 50.84
1-1096 223.32 308.90 126,.04 169.48 77.39 133.14 68.42 117.26

aquifer pollution and our best “recommended” scenario overall is fertilization F2, irriga-
tion I1 for medium soil (Fig.5f), where only 5.6% of the N fertilizer added is lost in the
first year.
Repeated application of fertilizers in an identical pattern for 3 years results in an accumu-
lation of N-NO3 in soil from 1 year to the next, and increases in total soil nitrate leaching
(Fig. 5, Table 3) and nitrate concentration in groundwater (Fig. 6, Table 2 ESM). Interestingly,
this happens while the water content in the soil layers does not increase from one year to
another (Fig. 3). Thus, the additional leaching to the groundwater is due to a gradual
accumulation of nitrate in the soil (relative to the uptake capacity of the plant), rather than
to an increased flow of water through soil (Fig. 4; Fig.1 ESM to Fig. 8ESM).
For the area analyzed, we conclude (Table 6 ESM) that medium soils are more suitable for a
sustainable corn production than coarse soils, ensuring both higher yield (and less wilting) and
less nitrate pollution of the aquifer.
Regardless of soil type and irrigation schedule, “gradual” fertilization scenario F2 (nitrogen
application timed with the plant needs) is clearly better than F1 (fertilization done only when
the crop is planted) for both higher crop yield and for minimizing aquifer pollution (Fig. 4 and
Figs.1 ESM to Fig. 8ESM).
Figure 6 and Fig. 9ESM show the concentration of the nitrate pollutant in the aquifer after
1 year (continuous red line) and after 3 years (dashed line) for coarse (Fig.6a) and for medium
soil (Fig.6b). In agreement with the increased nitrate leaching from soil over time, we note
increased aquifer pollution over time, with broad aquifer areas reaching and exceeding the
acceptable drinking water level, 10 mg/l for the coarse soil under the F1-I1 scenario.
Dispersivity coefficients for aquifers are notoriously difficult to estimate (e.g. Gelhar et al.
1992). We perform a sensitivity analysis to see the response of the aquifer pollutant plume to a
potential error in longitudinal and vertical dispersivities (αL, αT). Results are shown in
Figs. 9ESM-10ESM and Tables 4ESM-5ESM. Decreasing (increasing) the vertical
dispersivity coefficient decreases (increases) the penetration of nitrate in the aquifer. The
aquifer nitrate concentration is mostly insensitive to our dispersivity assumption in the top
5m of the aquifer, but becomes increasingly sensitive with depth below 5m.

5 Conclusions

We propose a new coupled model which combines a hydraulic module describing water
transport in unsaturated soils, a biogeochemical module describing nitrate dynamics in soils,
and an aquifer module describing nitrate dispersion in groundwater. The model can be used to
A Coupled Mathematical Model to Predict the Influence of Nitrogen 5245

develop best management practices; i.e. to achieve high crop yield while minimizing ground-
water nitrate pollution. The model can be adapted for specific regions of the world if soil water
retention curve, hydraulic conductivity for unsaturated soil, velocity and longitudinal
dispersivity for the groundwater are known.
Traditionally, the agricultural management of N has focused on the amount of fertilizer
applied (e.g., [SWRCB] State Water Resources Control Board 2013). Our simulations indicate
that a strategy that also includes water management will be more effective in reducing N
loading to groundwater, in line with recent studies (Letey and Vaughan 2013). Under-irrigation
causes poor yields as plants can simply wilt. Over-irrigation (schedule I2) increases the net
nitrate leakage to the groundwater (Fig. 5), while lowering the nitrate concentration in the root
regions and restricting plant growth. An intermediate trade-off between the two is necessary.
We show that optimizing N fertilization schedule to correspond to the corn life cycle
reduces groundwater contamination. Medium soils are superior for corn production, as their
texture restricts downward movement of water and nitrate, ensuring more uptake of water and
nitrate by the plants and thus more yield, as well as less aquifer pollution. Intensely fertilized
and irrigated coarse soil areas with shallow aquifers (e.g. Balta Brailei, Romania) are partic-
ularly prone to groundwater contamination. Finally, we find higher irrigation resulted in lower
N concentrations in both soil and aquifer, but much larger net leaching to the aquifer.
Measurements of nitrate leaching in agricultural soils and aquifer nitrate pollution in
Romania are sporadic and have not yet been catalogued in a useful way for scientific use.
We recommend surveying nitrate accumulation in the unsaturated zone and in rural wells in
areas deemed prone to nitrate contamination, to assess the impact of intensive agricultural
practices on nitrate leaching and groundwater contamination and provide a comparison dataset
for modelling attempts.

References

[SWRCB] State Water Resources Control Board (2013). Recommendations Addressing Nitrates in Groundwater.
Report to the Legislature. 20 Feb 2013, 70 p.
Antonopoulos VZ (2001) Simulation of water and nitrogen balances of irrigated and fertilized corn-crop soil. J
Irrig Drain Eng 127:77–83
Cannavo P, Recous S, Parnaudeau V et al (2008) Modeling N dynamics to assess environmental impacts of
cropped soils. Adv in Agron 97:131–174
Cassman KG, Dobermann A, Walters DT, Yang HS (2003) Meeting cereal demand while protecting natural
resources and improving environmental quality. Ann Rev Environ and Resour 28:316–358
Costa JL, Massone H, Martinez D, Suero EE, Vidal CM, Bedmar F (2002) Nitrate contamination of a rural
aquifer and accumulation in the unsaturated zone. Agric Water Manag 57:33–47
Fetter CW (1992) Contaminant hydrogeology. Macmillan Publishing Company, New York, pp 54–73
Gardenas AI, Hopmans JW, Hanson BR et al (2005) Two-dimensional modeling of nitrate leaching for various
fertigation scenarios under micro-irrigation. Agric Water Manag 74(3):219–242
Gelhar LW, Welty C, Rehfeldt KR (1992) A critical review of data on field-scale dispersion in aquifers. Water
Resources Res 28(7):1955–1974
Geng QZ (1988) Modélisation conjointe du cycle de l’eau et du transfert des nitrates dans un système
hydrologique. Dissertation, ENSM de Paris
Geng QZ, Girard G, Ledoux E (1996) Modeling of nitrogen cycle and nitrate transfer in regional hydrogeologic
systems. Ground Water 34(2):293–304
Greenwood DJ, Zhang K, Hilton HW et al (2010) Opportunities for improving irrigation efficiency with
quantitative models, soil water sensors and wireless technology. J Agric Sci 148:1–16
Hanson BR, Simunek J, Hopmans JW (2006) Evaluation of urea-nitrate fertigation with drip irrigation using
numerical modelling. Agric Water Manag 80:102–113
5246 I. Marinov, A.M. Marinov

Hopmans JW, Bristow KL (2002) Current capabilities and future needs of root water and nutrient uptake
modelling. Adv Agron 77:104–175
Hutson JL (2000) LEACHM: Model Description and User’s Guide. School of Chemistry, Physics, and Earth
Science, The Flinders University of South Australia, Adelaide, South Australia. Ground Water Modeling
Center, Riverside, CA
Jego G, Martinez M, Antiguedad I, Launay M, Sanchez JM, Justes E (2008) Evaluation of the impact of various
agricultural practices on nitrate leaching under the root zone of potato and sugar beet using the STICS soil-
crop model. Sci of the Total Environ 394:207–221
Kumazawa K (2002) Nitrogen fertilization and nitrate pollution in groundwater in Japan: Present status and
measures for sustainable agriculture. Nutr Cycl in Agroecosyst 63:129–137
Kundu MC, Mandal B (2009) Agricultural Activities Influence Nitrate and Fluoride Contamination in Drinking
Groundwater of an Intensively Cultivated District in India. Water, air and soil pollut 198:243–252
Lazaroiu A et all. (2008) Porumbul. Sa valorificam cat mai bine potentialul de productie ridicat al hibrizilor
PIONEER, Editura Fundatiei Culturale Gheorghe Marin Speteanu, Bucuresti (in Romanian)
Letey J, Vaughan P (2013) Soil type, crop and irrigation technique affect nitrogen leaching to groundwater. Calif
Agr 67(4):231–241. doi:10.3733/ca.E.v067n04p231
Marinov AM, Diminescu MA (2008) Experimental research and mathematical modelling of soil and ground-
water contamination with nitrogen compounds. Proc. of the IX-th. Int. Conf. on Water pollution, WIT
Transactions on Ecology and the Environment, WIT Press. Vol 111:115–126. doi:10.2495/WP080121
Nolan BT, Ruddy B, Hitt K (2006) Vulnerability of shallow groundwater and drinking-water wells to nitrate in
the United States. Environ Sci Technol 40:7834–7840
Oenema O, Van Liere L, Schoumans O (2005) Effects of lowering nitrogen and phosphorus surpluses in
agriculture on the quality of groundwater and surface water in The Netherlands. J Hydrol 304:289–301
Rinaldi M (2001) Application of EPIC model for irrigation scheduling of sunflower in Southern Italy. Agric
Water Manag 49:185–196
Rocha EO, Calijuri ML, Santiago AF, Campos de Assis L, Alves LGS (2012) The Contribution of Conservation
Practices in Reducing Runoff. Soil Loss, and Transport of Nutrients at the Watershed Level, Water Resour
Manage 26:3831–3852. doi:10.1007/s11269-012-0106-1
Schubert C, Knobeloch L, Anderson H, Warzecha C, Kanarek M (1997) Nitrate- contaminated drinking water
followback study. Submitted to the WI Department of Natural Resources and the WI Groundwater Co-
ordinating Council. Department of Preventive Medicine, University of Wisconsin-Madison and the WI
Department of Health and Family Services, 17p.
Su X, Wang H, Zhang Y (2013) Health Risk Assessment of Nitrate Contamination in Groundwater: A Case
Study of an Agricultural Area in Northeast China. Water Resour Manage. doi:10.1007/s11269-013-0330-3
Tilman D, Cassman KG, Matson PA, Naylor RL, Polasky S (2002) Agricultural sustainability and intensive
production practices. Nature 418:671–677
Van Genuchten MT (1980) A closed-form equation for predicting the hydraulic conductivity of unsaturated soils.
Soil Sci Soc Am J 44:892–898
Ward MH, Mark SD, Cantor KP, Weisenburger DD, Correa-Villasenor A, Zahm SH (1996) Drinking water
nitrate and the risk of non-Hodgkin’s lymphoma. Epidemiol 7:465–471
Watts DG, Hanks RJ (1978) A soil-water-nitrogen model for irrigated corn on sandy soils. Soil Sci Soc Amer
Proc 42:492–499

View publication stats

You might also like