JEQ Grow Your Career With ASA
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF) Free
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Related articles in JEQ
Right arrow Similar articles in this journal
Right arrow Similar articles in Web of Science
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Citing Articles
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Gasser, M. O.
Right arrow Articles by Laverdière, M. R.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Gasser, M. O.
Right arrow Articles by Laverdière, M. R.
Agricola
Right arrow Articles by Gasser, M. O.
Right arrow Articles by Laverdière, M. R.
Related Collections
Right arrow Nitrogen
Right arrow Solute Transport Models
Right arrow Nutrient Management
Right arrow Water Pollution
Right arrow Vadose Zone Processes and Chemical Transport
Journal of Environmental Quality 32:1464-1473 (2003)
© 2003 American Society of Agronomy, Crop Science Society of America, and Soil Science Society of America

TECHNICAL REPORTS
Vadose Zone Processes and Chemical Transport

Predicting Nitrate Leaching under Potato Crops Using Transfer Functions

M. O. Gasser*, J. Caron, R. Lagacé and M. R. Laverdière

Département des sols et de génie agroalimentaire, FSAA, Université Laval, Québec, QC, Canada G1K 7P4

* Corresponding author (mogasser{at}grr.ulaval.ca)

Received for publication July 2, 2002.

   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 MATERIALS AND METHODS
 THEORY
 RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES
 
Nitrate leaching is a major issue in many cultivated soils. Models that predict the major processes involved at the field scale could be used to test and improve management practices. This study aims to evaluate a simple transfer function approach to predict nitrate leaching in sandy soils. A convective lognormal transfer (CLT) function is convoluted with functional equations simulating N mineralization, plant N uptake, N fertilizer dissolution, and nitrification at the soil surface to predict solute concentrations under potato (Solanum tuberosum L.) and barley (Hordeum vulgare L.) fields as a function of drainage water. Using this approach, nitrate flux concentrations measured in drainable lysimeters (1-m soil depth) were reasonably predicted from 29 Apr. 1996 to 3 Dec. 1996. With average application rates of 16.9 g m-2 of N fertilizer in potato crops, mean nitrate-leaching losses measured under potato were 8.5 g N m-2. Tuber N uptake averaged 9.7 g N m-2 and soil mineral N at start (spring) and end (fall) of N mass balance averaged 1.7 and 4.5 g N m-2, respectively. Soil N mineralization was estimated by difference (4.3 g N m-2 on average) and was small compared with N fertilization. Small nitrate flux concentrations at the beginning of the cropping season (May) resulted mainly from initial soil nitrate concentrations. Measured and predicted nitrate flux concentrations significantly increased at mid-season (July–August) following important drainage events coupled with complete dissolution and nitrification of N fertilizers, and declining N uptake by potato plants. Decreases in nitrate concentrations before the end of year (November–December) underlined the predominant effect of N fertilizers applied for the most part at planting acting as a pulse input of solute.

Abbreviations: CLT, convective lognormal transfer • pdf, probability density function


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
MATERIALS AND METHODS
 THEORY
 RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES
 
GROUND WATER NITRATE contamination is a concern for Quebec rural residents living in areas of sandy soils intensively cropped to potato and other crops demanding high N fertilizer inputs (Levallois et al., 1998). Risk assessment of ground water contamination caused by nitrogen fertilizer use and optimization of current agricultural practices with the intent of reducing nitrate leaching requires a detailed knowledge of nitrogen transformation and transport dynamics. Modeling is often used to simulate these processes and enhance our knowledge, but major difficulties in modeling arise from strong interactions between physical factors, plant uptake, and nitrogen cycling processes (Garnier et al., 2001). Another challenge arises when models are to be evaluated at the field scale where assessment of parameters represents a major task.

To overcome highly heterogeneous water infiltration conditions at the field scale, Jury (1982) proposed the use of a stochastic transfer function to model solute transport. Solute transport is assumed to be convective with a stochastic component expressing the variability of solute travel times or pathways in the soil. The transfer function model uses a simple probability density function (pdf) related to cumulative surface net applied water or drainage water flowing through a reference depth (Jury, 1982; Jury et al., 1986). Several studies have reported successful attempts to model flux and resident concentrations following nonreactive solute pulses using the convective lognormal transfer (CLT) function (Jury et al., 1982; Butters and Jury, 1989; Ellsworth et al., 1996; Vanderborght et al., 1996; Caron et al., 1999). Hence, solute movement could be adequately represented by simple analytical expressions of the pdf.

Very few studies report the use of transfer functions to model the behavior of solutes undergoing biological processes in a complex soil–plant system. In this case, the input concentration is not a constant and solute flux concentrations evolving at any soil depth or time are expressed as the convoluted product of an input concentration function with a solute transfer function (Jury, 1982; White and Magesan, 1991; White et al., 1998):

[1]
where Cex and Cent are the output and input concentrations (M/L3), respectively, I is net surface infiltration or drainage water (L), fl is the solute flux transfer pdf calibrated at reference depth l (1/L), and I' is a dummy variable for integration purposes (L).

Using a pdf approach, Heng et al. (1994) modeled sulfate leaching to subsurface drains undergoing different processes such as mineralization–immobilization, soil adsorption, oxidation, rainfall input, and plant uptake. In their model, Heng et al. (1994) assumed source–sink terms were variable surface inputs and initial soil sulfate concentrations were homogeneous throughout the soil profile. Using the same experimental observations, Heng and White (1996) assumed sulfate fertilizer applications were the only surface pulse inputs, while other source–sink terms were linearly combined as a single initial homogeneous soil concentration. White et al. (1998) modeled nitrate leaching under the same experimental conditions. Source–sink terms related to net mineralization and plant uptake were again assumed as homogeneous initial soil concentration values. Thus, boundary and initial value problems can be addressed in different ways to model solute transport undergoing biochemical processes (Jury and Scotter, 1994).

Under limited biological activity with no plant interference, Ellsworth et al. (1996) demonstrated that nitrate leaching followed bromide leaching dynamics in a fine sandy loam and the CLT model was more effective than the conventional convective–dispersive equation in simulating bromide and nitrate concentrations at the plot scale. However, while solute movement can be adequately modeled with a solute flux transfer pdf, Heng et al. (1994) pointed out that the weakness of the model lay in quantitatively simulating the dynamics of biological processes. Most important processes affecting soil mineral nitrogen in sandy soils cropped to potato in Quebec are related to N fertilizer applications, crop N uptake, soil N mineralization, and nitrate leaching (Gasser et al., 2002; Tran and Giroux, 1991). In Quebec, potato N fertilization rates range from 130 to 210 kg N ha-1 and average tuber N uptake is 70 kg N ha-1 (assuming 4.6 Mg ha-1 average commercial tuber dry matter yield and 15 kg N Mg-1 tuber dry matter production) (Gasser and Laverdière, 2000). Since N fertilizers are predominantly band-applied at planting under ammonia forms, dissolution and nitrification of ammonium fertilizers may play an important role in the timing of nitrate concentrations developing in the soil (McInnes and Fillery, 1989). Cumulated N uptake by potato crops can be modeled using asymptotic functions, relating plant N uptake to time or cumulated heat units (Gasser, 2000).

Soil N mineralization depends on mineralizable N availability and abiotic conditions such as soil temperature and soil moisture governing mineralization kinetics. Laboratory incubations are used to determine the influence of soil temperatures on mineralization kinetics (Ellert and Bettany, 1992). Other studies have successfully related soil organic N or organic C contents to the N mineralization potential of incubated soils (Simard and N'dayegamiye, 1993; Cabrera and Kissel, 1988), but only a few studies have attempted to compare simulated results with field measurements (Cabrera, 1993; Houot et al., 1989).

Using the CLT model, White et al. (1998) simulated long-term trends in nitrate leaching in mole and pipe drainage assuming linear source–sink terms related to soil N nitrification and plant N uptake. Integrating sink terms such as plant N uptake in the transfer function model represents one of the major difficulties for modeling nitrate leaching under annual crops such as potato, since plant N uptake is delayed in time compared with input concentrations related to fertilizer applications, soil N mineralization, and nitrification, causing net negative input concentrations at some periods.

Although the use of transfer functions implies linearity of the convoluted product, any combination of nonlinear functions describing input concentrations and solute transfer can be used, as long as the probability that a solute entering the soil surface at time t–t' and exiting at time t is unaffected by other processes during its transport and is independent of its time of entrance (Jury and Roth, 1990; White and Magesan, 1991). Since most major nitrogen transformation processes related to microbial activity, plant uptake, and fertilizer dissolution occur in the vicinity of the soil surface and solute transport is relatively fast in sandy soils, nitrate leaching could be modeled using the transfer function approach with input concentrations related to different source–sink terms entering at the soil surface. The objective of this paper is to evaluate the pertinence of such a simple approach using realistic functions describing the different nitrogen source–sink terms with the CLT function to predict nitrate leaching under transient water conditions in cultivated sandy soils.


   MATERIALS AND METHODS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
THEORY
 RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES
 
Site Description and Instrumentation
This study was conducted in five fields cultivated by potato farmers of the Portneuf region located near Quebec City, Canada (latitude between 46°45' and 46°50'; longitude between 71°35' and 72°00'). Details of the experimental sites and their instrumentation are given elsewhere (Gasser, 2000; Gasser et al., 2002). Briefly, the soils under investigation were either Morin or Pont Rouge sands (Humo ferric Podzol, Humic Haplorthod) developed on medium to coarse sands of fluvio-marine or deltaic origin (Raymond et al., 1976). The cultivated layer (0–0.25 m) of these soils was loamy sand to sandy loam, while underlying horizons were medium to coarse sands with variable gravel content.

Fifteen drainable lysimeters (three per field), the size of a rectangular reservoir (1-m2 surface by 1-m maximum depth), were installed during July 1995. Construction and installation is described in more detail in Gasser et al. (2002). Following significant rainfall events (>10 mm), drainage volume was manually recorded in each drainable lysimeter on 20 occasions in 1996, starting at the end of snow thaw on 29 April until 12 December when soil froze. A water sample (20 mL) was collected in each lysimeter on every event and immediately frozen for storage and transport to the laboratory. Samples were thawed before nitrate and ammonium analysis with Technicon (Tarrytown, NY) automated procedures that included repeated analysis of samples to check for error (Keeney and Nelson, 1982). Ammonium concentrations (>0.25 mg NH4–N L-1) in drainage water were only detected on a few occasions.

Field operations and rates of N fertilizers applied are reported in Table 1 . Four fields (Fields 1, 2, 3, and 4) were planted with potato and one (Field 5) was seeded to barley. Nitrogen fertilizers were for the most part band-applied at planting and ammonium fertilizers represented 72 to 83% of applied N at planting for potato. Smaller rates of N fertilizers were applied as ammonium nitrate in Fields 3, 4, and 5, approximately 30 d after planting. Potato fields were row-cultivated and hilled with a disk cultivator, creating a height of approximately 30 cm between furrow bottom and potato ridge. Potato fields were harvested in September, but due to poor growing conditions, the barley field was not harvested.


View this table:
[in this window]
[in a new window]
  		Table 1. Field operations schedule and N fertilizers applied as ammonium, nitrate, and urea in five fields.

 
Plant N uptake was monitored seven times during the cropping season by harvesting plants growing 5 to 10 m away from lysimeters every 10 d following emergence. To determine N exported at harvest, tubers from three to four plants grown on each lysimeter were harvested on rows 110 cm long (91-cm interrow spacing) before producers harvested the field. Fresh plant tissues were separated in tuber, foliage, and root tissues and weighed. Subsamples were weighed, oven-dried at 60°C, and weighed again to determine dry matter content. Subsamples of plant tissues were milled and sieved at 0.5 mm to determine total plant N with an automated Kjeldahl digestion procedure in a Kelfoss apparatus (Horwitz, 1980, Methods 7B.01 to 7B.04).

Soils surrounding drainable lysimeters (2-m radius adjacent to lysimeters) were sampled with a hand auger (7-cm diameter) on 10 occasions between 29 Apr. and 18 Nov. 1996 at five depths (0–15, 15–30, 30–60, 60–90, and 90–120 cm). In potato fields, samples were taken 15 to 30 cm from the center of the hill. In the barley field, samples were taken randomly. Composite samples were made of six subsamples taken at 0- to 15- and 15- to 30-cm depths and two subsamples for the 30- to 120-cm depths. Nitrate and ammonium in soil samples were extracted with a 2 M KCl solution (1:5 solution), shaken for 1 h, and filtered on #411 VWR (West Chester, PA) paper filters, before nitrate and ammonium analysis using Technicon automated procedures (Keeney and Nelson, 1982). Nitrate and ammonium masses (g m-2) and concentrations (g m-3) were calculated using measured soil gravimetric water content, gravel (>4 mm) content, and bulk density for each layer and lysimeter location (Gasser et al., 2002). From soils sampled at 0- to 15- and 15- to 30-cm depths on 29 Apr. 1996, initial soil organic C content was determined using the modified Walkley–Black procedure and total N content was determined using an automated Kjeldahl digestion procedure in a Kelfoss apparatus (Horwitz, 1980, Methods 7B.01 to 7B.04) (Table 2) .


View this table:
[in this window]
[in a new window]
  		Table 2. Initial soil properties on 29 Apr. 1996 at 0- to 30- and 0- to 90-cm depths.

 

   THEORY
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 THEORY
RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES
 
Transfer Function Model Approach under Transient Water Regime
To use the transfer function model approach under a transient water regime and normal field conditions, Jury and Roth (1990) stipulate that the following conditions apply: (i) pdfs of the solute travel times are related to cumulative drainage or net infiltrated water, but are independent of water flux; (ii) capillary rise of water is negligible or does not generate any important solute transport to the surface; (iii) the transfer function should be corrected for variations in soil water content, when pdf is calibrated under a certain soil water content; and (iv) water flux may affect the transfer function in well-structured soils, since wet pore geometry may vary. In coarse-textured soils, it is hypothesized that contrasting water fluxes should less affect pore sizes and solute transport.

To model variable input concentrations, Eq. [1] can be summed for each increment of cumulative drainage {Delta}I:

[2]

Under a transient water regime, cumulative drainage I is a function of varying water flux Jw at time t:

[3]
where {Delta}t is the time interval. Also, to transform Eq. [2] into a travel-time pdf representing transient outflow concentrations under nonsteady state water flow, the drainage pdf must be adjusted to water flux Jw at time t–t' to preserve unity of the pdf:

[4]

Input concentration from source terms is the mass of solute applied M (M/L2) per infiltrating water flux [L3/(L2 T)] at time t:

[5]

Since infiltrating water flux Jw(0,t–t') is estimated as precipitation minus evapotranspiration in the model (Jury et al., 1990), it closely matches drainage water flux Jw(z,t–t') in the soil profile over a short period of time, and Eq. [2] simplifies to:

[6]

An initial (t = 0) uniform solute concentration in soil produces an additional outflow concentration at z given by (Jury et al., 1990; White et al., 1998):

[7]
where C0 is the average initial value of the solute concentration in the soil solution to depth z. Using summation, Eq. [7] translates to:

[8]

[8]

The appropriate pdf for conservative solute movement through a soil may be stochastic–convective in nature, such as the lognormal function (Jury and Roth, 1990). The lognormal flux concentration pdf at depth z is given by the equation:

[9]
where µ and {sigma} are the mean and standard deviation of the distribution of I values, respectively, and l is a calibration depth (Jury, 1982).

Estimation of Model Parameters
Initial Soil Nitrate Concentrations
On 29 Apr. 1996, the initial amount of soil nitrate present to a depth of 90 cm in soils adjacent (2-m radius) to the 15 lysimeters ranged from 0.2 to 2.4 g N m-2 (Table 3) . Lowest soil nitrate masses were measured on Field 5 where nearly no nitrate was detected in the subsurface horizons (30–90 cm). Initial soil volumetric water contents to a depth of 90 cm ranged from 0.10 to 0.21 m3 m-3 with higher water volumes present in the surface horizons (0–30 cm). Initial soil NO-3 concentrations C0 were calculated on a volumetric water content basis at each soil depth (0–15, 15–30, 30–60, and 60–90 cm). Soil depth averaged C0 value for each lysimeter ranged from 1.8 to 24 g N m-3 and mean standard deviation of C0 was less than 44% of mean value of C0, except on Field 5 where only very low concentrations of nitrate were present in the subsurface horizons (Table 2). Therefore, C0 was assumed approximately constant with depth for all lysimeters, except on Field 5.


View this table:
[in this window]
[in a new window]
  		Table 3. Inorganic N mass balance at the 0- to 90-cm depth between start (29 Apr. 1996) and end (3 Dec. 1996) of monitoring period.

 
Nitrogen Source Terms
Nitrogen source terms related to fertilizer applications and mineralization–immobilization of soil organic matter were modeled as time-dependent surface (z = 0) inputs M (Eq. [6]). Nitrogen fertilizer inputs were divided into nitrate and ammonium inputs according to Table 1. We assumed nitrate fertilizer inputs were instantaneously dissolved at the soil surface (pulse inputs in the model). Presence of ammonium masses (g m-2) in soil was assumed to solely result from dissolution of ammonium fertilizers, whereas possible plant ammonium uptake was neglected (Fig. 1) . Under this assumption, time-distributed nitrogen inputs Nf(t) from ammonium fertilizer sources are shown to follow a lognormal pdf:

[10]
where NH4f is ammonium fertilizer input (Table 1), and µ and {sigma} are field-specific parameters fitted to normalized ammonium masses measured in soil (0- to 30-cm depth) using nonlinear regression (Table 4) (Gasser, 2000). As illustrated in Fig. 1, sidedressing small amounts of ammonium fertilizers in Fields 3 and 4 had only a small effect on the distribution of ammonium masses measured in soils after sidedressing.



View larger version (31K):
[in this window]
[in a new window]
  		Fig. 1. Evolution of normalized ammonium masses in soil during cropping season and fitted lognormal probability density function (pdf) of time-distributed nitrogen inputs [Nf(t)]. Planting date (P) and fertilizer sidedressing date (S) are indicated with arrows.

 

View this table:
[in this window]
[in a new window]
  		Table 4. Parameter values for lognormal solute transfer function, lognormal NH4 fertilizer dissolution function, and plant N uptake function.

 
Soil N mineralization (Nm) during the monitoring period was determined from an inorganic N mass balance (Table 3):

[11]
where Nse is soil mineral N in autumn at the end of the monitoring period, Np is plant N uptake exported at harvest with potato tubers or immobilized in the barley stubble, Nl is leached N, Nss is soil mineral N in spring at the start of the monitoring period, and Nf is mineral N fertilizer applied.

Cumulative N mineralization was modeled using a first-order equation that included soil temperature, soil total N, and organic C as variables:

[12]
where Nm is soil N mineralization, N is soil total N, C is soil organic C, T(t) is daily mean soil temperature, and a = 0.0043, Q10 = 4.06, c = 5.30, and d = 0.96 are mineralization parameters established under laboratory conditions with similar sandy soils (Gasser, 2000). Figure 2 illustrates the fitted function (Eq. [12]) to cumulative N mineralization patterns of four sandy soils sampled in spring following previous sod and potato crops, and incubated at three constant temperatures.



View larger version (31K):
[in this window]
[in a new window]
  		Fig. 2. Cumulative N mineralization patterns of four sandy soils sampled in spring following previous sod and potato crops, and incubated at three constant temperatures (4, 14, and 24°C).

 
Daily soil N mineralization rates were simulated using the first derivative of cumulative N mineralization Nm(t) (Eq. [12]). Soil N mineralization Nm was estimated on each lysimeter from N mass balance (Table 3), daily mean soil temperatures were recorded at a 20-cm depth by an automatic weather station located less than 1 km from one field, and soil total N and organic C were measured on each lysimeter (Table 2).

Nitrogen Sink Term
Daily N uptake by potato plants was subtracted from source terms M entering the soil surface (Eq. [6]). Figure 3 illustrates cumulative N uptake by potato plants during the cropping season and final tuber N uptake at harvest. Cumulative N uptake by potato plants and final tuber N uptake were modeled using a lognormal pdf:

[13]
where µ and {sigma} are parameters fitted to cumulative plant N uptake during the growing period. Since final tuber N uptake is lower than maximal N uptake by the whole potato plant during the season, the use of a lognormal bell-shaped function enables the simulation of the release of N accumulated in potato roots and foliage at the end of the cropping season. Field-specific parameters µ and {sigma} are reported in Table 4 and lysimeter-specific total tuber N uptake Np is reported in Table 3. Total barley N uptake on Field 5 was estimated as 3 g m-2 to represent N immobilization in barley stubble. Daily plant N uptake or release at the end of the season was calculated using the first derivative of cumulative plant N uptake, Np(t).



View larger version (31K):
[in this window]
[in a new window]
  		Fig. 3. Evolution of cumulative plant N uptake during cropping season and fitted lognormal probability density function (pdf) of cumulative plant nitrogen uptake [Np(t)]. Final plant N uptake corresponds to tuber N uptake. Arrows indicate planting date (P), fertilizer sidedressing date (S), and harvest date (H).

 
Solute Transfer Function
The parameters for the lognormal solute transfer pdf (Eq. [9]) were calibrated with bromide breakthrough curve experiments made on the same drainable lysimeters. Most of the information concerning the calibration process of the solute transfer function has been reported in Gasser et al. (2002). Briefly, pulses of 10 g Br- m-2 were hand-sprayed in May 1996 on soil surfaces centered over each lysimeter. The Br- flux concentrations were measured subsequently in lysimeters at a 1-m soil prediction depth z following significant drainage events (15 occasions) from June to November 1996. Lognormal solute transport parameters (µ and {sigma}) were calibrated using a 30-cm reference depth l (Eq. [9]) to compare with previously published parameter values using the same calibration depth l. Since Br- and NO-3 flux concentrations were measured at the same 1-m soil prediction depth z, dispersive effects related to calibration depth should not be an issue here. Finally, solute lognormal transfer parameters calibrated for each field are reported in Table 4.


   RESULTS AND DISCUSSION
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 THEORY
 RESULTS AND DISCUSSION
CONCLUSIONS
 REFERENCES
 
Table 3 reports N mass per surface attributed to each balance component at the end of the monitoring period over the 15 drainable lysimeters. On average, nitrate leaching losses measured under potato were 8.5 g N m-2, tuber N uptake averaged 9.7 g N m-2, mass of N fertilizers applied averaged 16.9 g N m-2, and soil mineral N at start (spring) and end (fall) of N mass balance averaged 1.7 and 4.5 g N m-2. Mass of leached N and tuber N uptake represents on average a slightly greater mass than N fertilizer applied. Net soil N mineralization can explain this gain in soil mineral N that was leached and also the greater mass of soil mineral N at the end of the monitoring period. On average, net soil N mineralization under potato is estimated at 4.3 g N m-2, which is comparable with other estimates of N mineralization for sandy soils in Quebec (Simard and N'dayegamiye, 1993; Tran and Giroux, 1991). However, this component shows much greater variation between lysimeters, which could result from unmeasured processes such as denitrification and immobilization. Error propagation linked to measured components in Eq. [11] (estimated as the square sum of measurement error terms related to N components in Eq. [11]) could also explain a great part of this variation (Gasser, 2000).

Use of the Convective Lognormal Transfer Function to Predict Nitrate Leaching
Measured and predicted nitrate flux concentrations evolving at a 1-m soil depth in each lysimeter are compared in Fig. 4 . Predicted nitrate fluxes during the earliest part of the monitoring period, between day of the year (DOY) 121 (2 May) and DOY 200 (18 July), were generally in good agreement with measured values, although in some lysimeters values were overestimated (e.g., Lysimeters 8 and 11). Therefore, initial nitrate concentrations Co measured in soils surrounding lysimeters (Table 2) fairly well represented initial conditions in lysimeters, because these concentrations were those contributing to early N leaching. In general, measured nitrate concentrations increased after DOY 200 (18 July) and decreased before DOY 300 (26 October), underlying the predominant effect of N fertilizers applied for the most part at planting, acting as a pulse. High nitrate concentrations, close to 100 mg NO3–N L-1 measured in Lysimeters 10 and 13, were well predicted, but were underestimated in Lysimeters 4 and 12 where higher measured concentrations were recorded. Low nitrate concentrations were overestimated by the transfer function model in Lysimeters 3, 5, and 6. Overestimation of N fertilizer inputs caused by potato rows seeded on the edges of Lysimeters 5 and 6 could explain low concentrations measured in these lysimeters, which in turn resulted in poor simulations. Table 3 also reports negative net N mineralization on these lysimeters, which supports the possibility that fertilizers were not directly applied over these lysimeters. Overall, predicted nitrate concentration patterns were closely matched to observed patterns in nine lysimeters (1, 4, 7, 8, 10, 12, 13, 14, and 15), but were less acceptable in five lysimeters (2, 3, 5, 6, and 11).



View larger version (30K):
[in this window]
[in a new window]
  		Fig. 4. Measured and predicted nitrate flux concentrations in drainable lysimeters at a 1-m soil depth.

 
Divergence between measured and predicted nitrate flux concentrations was examined throughout the monitoring period for each field using the relative mean bias error:

[14]
where n is number of lysimeter observations in each field, Cm is measured flux concentration, and Cs is predicted flux concentration.

Relative mean bias error was overall most pronounced in Field 2 (Lysimeters 4, 5, and 6) compared with other fields (Fig. 5) . As explained earlier, overestimation of N fertilizer inputs on Lysimeters 5 and 6 due to misaligned potato rows and banded fertilizer may have caused an overprediction of nitrate fluxes. On the other fields, predicted nitrate fluxes seemed slightly overestimated during the earliest part of the monitoring period, but this overestimation decreased during the rest of the period.



View larger version (23K):
[in this window]
[in a new window]
  		Fig. 5. Relative bias error on predicted nitrate flux concentrations evolving at a 1-m soil depth under four potato fields (1, 2, 3, and 4) and one barley field (5) in 1996.

 
Using the transfer function model and measured drainage volumes, predicted cumulative nitrate masses leached during the whole year were calculated and compared with measured values (Fig. 6) . A linear regression with intercept equal to zero (R2 = 0.64, p < 0.001) indicates a slope near unity (0.97) between predicted and measured values. Considering all lysimeters under potato fields, the model overestimated mean annual nitrate N losses of 8.5 g m-2 by 0.5 g m-2. Deviations greater than the root mean square error (1.96 g m-2) were observed on Lysimeter 3 (Field 1), Lysimeter 5 (Field 2), and Lysimeter 12 (Field 4).



View larger version (21K):
[in this window]
[in a new window]
  		Fig. 6. Measured and predicted cumulative NO3–N leaching mass flux at a 1-m soil depth under four potato fields (1, 2, 3, and 4) and one barley field (5) in 1996.

 
To illustrate the functions implied in this model, input variables and parameters were averaged from potato crops over drainable lysimeters, and nitrogen daily fluxes from positive source terms (fertilizer N and soil N mineralization) and negative sink terms (N tuber uptake and leached N) were simulated (Fig. 7) . Average values of N components under a potato crop were used (Table 3). For this example and since ammonium fertilizers represent more than 68% of N fertilizers applied, N fertilizer fluxes were simulated solely as ammonium applied at planting that undergoes dissolution and nitrification (Eq. [11]).



View larger version (26K):
[in this window]
[in a new window]
  		Fig. 7. Simulated N mass fluxes under average conditions for potato production in 1996.

 
Figure 7 illustrates that soil N mineralization is on average a small N input (positive) compared with N fertilizers and that plant N uptake was delayed in time compared with N fertilizer inputs. As explained earlier, plant N uptake was modeled using the first derivative of a bell-shaped lognormal distribution to account for plant N uptake during the earlier part of the monitoring period (negative output) followed by releases of organic N from crop residues at the end of the period (positive input).

Even though nitrate flux concentrations exhibited various bell-shaped patterns (Fig. 5), nitrate leaching (mass flux negative output) was very much influenced by drainage water flux as shown in Fig. 7. Significant nitrate leaching occurred when plant N uptake started to decline at around DOY 200 (18 July). The use of catch crops to reduce nitrate leaching in this situation would be ineffective, since a great part of leaching occurred under the presence of the potato crop. However, interrow nitrate catch crops could be a feasible alternative in other crops, such as cereals.

Reducing N fertilizer use by synchronizing fertilizer N release with plant N uptake seems the only viable alternative to reduce nitrate leaching in crops such as potato. For instance, the optimal peak for fertilizer N release should be delayed in time to closely match maximal plant N uptake at around DOY 195 (13 July). Sidedressing N fertilizers during the cropping season and use of slow-release N fertilizers have been promoted to reduce nitrate leaching, but significant reductions have been measured only when N fertilizer use was significantly reduced (Prunty and Greenland, 1997; Saffigna et al., 1977). Also, sidedressing N fertilizer in nonirrigated potato crops has been shown to produce inconsistent results in terms of potato yields (Porter and Sisson, 1993). Under rainfed conditions, rainfall distribution and soil water availability not only control nitrate leaching but also influence plant N uptake and availability of N fertilizers. To adequately test various fertilization scenarios, the present model would need to include some interactions between soil water availability, plant growth, and fertilizer dissolution and availability.


   CONCLUSIONS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 THEORY
 RESULTS AND DISCUSSION
 CONCLUSIONS
REFERENCES
 
Overall, the use of four functional equations to describe the dynamics of main N processes occurring in this soil–plant system seems appropriate to model nitrate leaching. Initial soil nitrate concentrations measured in spring were homogeneously distributed in the soil profile and the analytical solution to the CLT function for this condition produced reasonable estimates of early nitrate flux concentrations at a 1-m soil depth. Combining soil N mineralization, fertilizer N dissolution–nitrification, and plant N uptake into one input function at the soil surface appeared to be a simplified but reasonable assumption that allowed the estimate of nitrate flux concentrations at a 1-m soil depth when used in a convoluted product with the CLT function. Use of a bromide tracer to parameterize the CLT function also appeared to adequately represent the leaching process in these soils. Although some biological reactions implying denitrification or plant N response to N fertilizer concentrations and soil water availability were omitted, the nitrate leaching process was reasonably described with such an approach, indicating the importance of nitrate leaching during drainage events in these soils.


   ACKNOWLEDGMENTS
 
Funding for this research was provided by the Natural Sciences and Engineering Research Council of Canada, the Canadian Pork Council, and the Fédération des Producteurs de Porcs du Québec. Gratitude is expressed to M. Arfaoui, D. Marcotte, D. N'Sengiyumva, and D. Tse Bi Tra for technical assistance.


   REFERENCES
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 THEORY
 RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES
 


Related articles in JEQ:

This Issue in Journal of Environmental Quality

JEQ 2003 32: 1167-1172. [Full Text]  




This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF) Free
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Related articles in JEQ
Right arrow Similar articles in this journal
Right arrow Similar articles in Web of Science
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Citing Articles
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Gasser, M. O.
Right arrow Articles by Laverdière, M. R.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Gasser, M. O.
Right arrow Articles by Laverdière, M. R.
Agricola
Right arrow Articles by Gasser, M. O.
Right arrow Articles by Laverdière, M. R.
Related Collections
Right arrow Nitrogen
Right arrow Solute Transport Models
Right arrow Nutrient Management
Right arrow Water Pollution
Right arrow Vadose Zone Processes and Chemical Transport

HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
The SCI Journals Agronomy Journal Crop Science
Journal of Natural Resources
and Life Sciences Education		 Vadose Zone Journal
Soil Science Society of America Journal Journal of Plant Registrations The Plant Genome