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 ISI 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 HighWire
Right arrow Citing Articles via ISI Web of Science (12)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Miao, Z.
Right arrow Articles by Capri, E.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Miao, Z.
Right arrow Articles by Capri, E.
Agricola
Right arrow Articles by Miao, Z.
Right arrow Articles by Capri, E.
Related Collections
Right arrow Agricultural Pesticides
Right arrow Pesticides
Right arrow Coupled Flow/Transport Models
Published in J. Environ. Qual. 32:2189-2199 (2003).
© ASA, CSSA, SSSA
677 S. Segoe Rd., Madison, WI 53711 USA

TECHNICAL REPORTS

Organic Compounds in the Environment

Simulating Pesticide Leaching and Runoff in Rice Paddies with the RICEWQ–VADOFT Model

Zewei Miaoa, Mark J. Cheplickb, Martin W. Williamsb, Marco Trevisana, Laura Padovania, Mara Gennaric, Aldo Ferrerod, Francesco Vidottod and Ettore Capri*,a

a Istituto di Chimica Agraria ed Ambientale, Università Cattolica del Sacro Cuore, 29100 Piacenza, Italy
b Waterborne Environmental, Inc., 897-B Harrison Street, S.E. Leesburg, VA 20175
c DACPA, Sezione di Scienze agrochimiche, University of Catania, Italy
d Dipartimento Agronomia, Selvicoltura e Gestione del Territorio, Università degli Studi, Torino, Italy

* Corresponding author (ettore.capri{at}unicatt.it).

Received for publication October 7, 2002.

   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES
 
There is a current need to simulate leaching and runoff of pesticide from rice (Oryza sativa L.) paddies for assessing environmental impacts on a valuable agricultural system. The objective of this study was to develop a model for determining predicted environmental concentration (PEC) in soil, runoff, and ground water through the linkage of two models, rice water quality model (RICEWQ) and vadose zone transport model (VADOFT), to simulate pesticide fate and transport within a rice paddy and underlying soil profile. Model performance was evaluated with a field data set obtained from a 2-yr field experiment in 1997 and 1998 in northern Italy. The predictions of amount of pesticide running off from the paddy field and accumulating in the paddy sediment were in agreement with measured values. Leaching into the vadose zone accounted for approximately 19% of the applied dose, but only a small amount of chemical (<0.1%) was predicted to reach ground water at a 5-m depth due to sorption and transformation in the soil. The permeability of the soil and the water management practices in the paddy field were shown to have a strong influence on pesticide fate. These factors need to be well characterized in the field if model predictions are to be successful. The combined model developed in this work is an effective tool for exposure assessments for soil, surface water, and ground water, in the particular conditions of rice cultivation.

Abbreviations: DAT, days after treatment • EU, European Union • FOCUS, Forum for the Coordination in the Use of Models • LD-PE, low density polietyl • NP, nodal point • PEC, predicted environmental concentration • RE, reduction of error test • RICEWQ, rice water quality model • RMSE, root mean squared error • VADOFT, vadose zone transport model


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES
 
CONCENTRATIONS OF PESTICIDE RESIDUES in excess of the allowable European maximum residue (0.1 µg L-1) have been reported in surface and ground water drinking supplies in areas cropped with rice in Europe (Capri et al., 1999; Charizopoulos and Papadopoulou-Mourkidou, 1999; Gomez de Barreda, 1999; Cerejra, 2000; Villholth et al., 2000). In 14 studies reviewed by Capri et al. (1999) thiocarbamate herbicides (molinate [S-ethyl hexhydro-1 H-azepine-1-carbothiate], dimepiperate [S-(1-methyl-1-phenylethyl)-1-piperidinecarbothioate], thiobencarb [4-Chlorobenzyl diethylthiolcarbamate], and thiocarbazyl [S-(phenylmethyl)bis(1-methylpropyl)carbonmothiate), bentazone [3-isopropyl-1H-2,1,3-benzothiadiazin-4(3H)-one 2,2-dioxide], and oxadiazon [2-tert-Butyl-4-(2,4-dichloro-5-isopropoxyphenyl)-1,3,4-oxadiazolin-5-one] were frequently found at concentrations of 0.1 to 30 µg L-1 in both surface and ground water resources. In Greece, a monitoring study was performed in the most important rice area along the Axios River (15000 ha of rice). In the interregional program of the Ministry of Agriculture of Greece both molinate and propanil (3',4'-dichloropropionanilide, Rohm and Haas Co., Philadelphia, PA) were among the pesticides most frequently detected in the main drainage channel systems of the basin. Molinate was reported as the most frequently found pesticide in the Axios River Basin with the highest concentrations occurring during storm events (Charizopoulos and Papadopoulou-Mourkidou, 1999). In Spain, a monitoring study in the Albufera Natural Park reported detections of molinate (Gomez de Barreda, 1999).

Detections can be attributed partly to the high intensity of rice production and partly to the geological conditions where the rice is cultivated. In Italy, 40% of the rice cropping area consists of sand and gravel-based soils. The intense use of water for irrigating the rice between February and August can exacerbate diffuse and point contamination of ground water and surface water. Therefore, environmental agencies are implementing intensive monitoring programs to evaluate the impact of agriculture on contiguous natural ecosystems. One outcome of these programs has been the restriction of use of several pesticides, including molinate and bentazone.

The use of validated models in determining PECs in ground water and surface water is becoming integrated into the regulatory process basis for assessing the potential environmental exposure and for optimizing monitoring programs for registration purposes. In recent years substantial advances have been made in the use of simulation models to predict pesticide behavior in the soil-water-crop continuum, resulting in the increased use of models by national or international state regulatory agencies (Capri et al., 1999; Vanclooster et al., 2000). Nevertheless, validated models applied to rice cultivation at the European Union (EU) level are still missing. In the USA, a pesticide runoff model for rice crops, RICEWQ, was developed for pesticide exposure assessment (Williams et al., 1999). The model calculates chemical dissipation in the paddy and surface water releases from overflow and drainage, but does not take into account chemical leaching, which can be a main dissipation source under some field conditions (Capri and Miao, 2002). There are several models available for the calculation of pesticide leaching in the soil, including VADOFT a vadose zone transport model contained within the USEPA's Pesticide Root Zone Model (Carsel et al., 1998). However, most of these models have been designed for use with non-flooded crops and have no facility to incorporate the upper boundary condition found in rice paddies.

In Europe, Forum for the Coordination in the Use of Models (FOCUS) (FOCUS, 2003) has developed a common system for conducting risk assessments for pesticides in rice at the lower tiers of the risk assessment. The system is especially intended for the inclusion of a substance in Annex 1 of the Directive 91/414. Unfortunately, the FOCUS group has not specified a validated higher tier model for calculating PEC values in surface water and ground water for pesticide use on rice.

The aim of this study was to evaluate a model formulated through the linkage of RICEWQ with VADOFT to predict pesticide fate in a rice paddy field, including the solute transport processes of leaching in the unsaturated (vadose) zone and runoff.


   MATERIALS AND METHODS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES
 
Description of the RICEWQ-VADOFT Model
Assessments were made to evaluate the dissipation, leaching, and runoff of agro-chemicals in aquatic systems. These assessments were conducted using an integrated model that linked two fate and transport models, RICEWQ version 1.6.2 and VADOFT.

RICEWQ simulates water and chemical mass balance associated with the unique flooding conditions, overflows, and controlled water releases that are typical in a rice cropping system (Williams et al., 1999). The model applies the principle of mass balance to simulate water volume changes in the paddy and chemical residues in three media of the rice paddy (rice foliage, water column, and benthic sediments) from the point of chemical application:

[1]
where {partial}C (mg kg-1) is the change in concentration over time {partial}t (s); {sum}Minflux (10-6 kg) and {sum}Moutflux (10-6 kg) are cumulative influx and outflow of chemical mass from the control volume; V (m3) (i.e., the rice paddy), and {sum}Mreact (10-6 kg) is mass transformation from all processes.

The pesticide mass-balance equations in water, sediment, and foliage subecosystems are listed as follows:

[2]

[3]

[4]
where {partial}MW (mg), {partial}MS (mg), and {partial}MF (mg) are the change in chemical mass in water, sediment, and foliage, respectively, over time ({partial}t) (s); MWapp (mg) is the mass of the applied pesticide not lost to drift and arriving at the water surface; MFapp (mg) is the mass of the applied pesticide intercepted by foliage; Mwash(mg) is the mass washed off from foliage; MW deg (mg), MS deg (mg), and MF deg (mg) are the masses of pesticide degraded in water, sediment, and foliage, respectively; MWtran (mg), MStran(mg), and MFtran (mg) are the masses of metabolite formed by transformation of parent compound in water, sediment, and foliage, respectively; Mvolat(mg) is the mass volatilized across the air-water interface; Mout (mg) is the mass lost in overflow or drainage; Mseep (mg) is the mass lost in seepage; Mbed (mg) is the mass transfer to bed sediment by direct partitioning; Msetl (mg) is the mass transfer to sediment by particulate settling; Mresus (mg) is resuspended mass; Mdifus (mg) is the mass diffusion between the water and sediment; and Mharv (mg) is the mass of pesticide removed after harvest (this mass may be removed from the ecosystem, left alone and available for wash off, or be applied to bed sediment).

VADOFT performs one-phase one-dimensional transient or steady-state simulations of downward water flow and chemical solute transport in variably saturated porous media (Carsel et al., 1998). The code employs the Galerkin finite-element technique to approximate the governing equations for water flow and chemical transport with spatial discretization (expressed as nodal points [NPs]) performed using linear elements. It allows for a wide range of nonlinear flow conditions, and handles various transport processes, including hydrodynamic dispersion, advection, linear equilibrium sorption, and first-order decay. The VADOFT code solves the Richards' equation, the governing equation for infiltration of water in the vadose zone:

[5]

Where {Psi} is the pressure head (m), K is the saturated hydraulic conductivity (m s-1), krw is the relative permeability (dimensionless), z is the vertical coordinate (m), t is time (s), and {eta} is an effective water storage capacity (m-3) defined as:

[6]
where Ss is specific storage (m-3), Sw is water saturation (dimensionless), and {Phi} is the effective porosity (dimensionless).

The governing equation for one-dimensional transport of a nonconservative chemical solute species in a variably saturated soil takes the form:

[7]
where D is the apparent dispersion coefficient (m2 s-1), c is the solute concentration (µg L-1), {theta} is the volumetric water content (m3 m-3) ({theta} = {Phi}Sw), the vertical Darcy velocity (m s-1) R is the retardation coefficient (dimensionless), and {lambda} is the first-order decay constant (s-1). Note that D is defined as:

[8]
where {alpha}L is the longitudinal dispersivity (m); and D* is the effective molecular diffusion coefficient (m2 s-1).

As for water balance, RICEWQ and VADOFT both use a water-balance model to calculate the water balance in the paddy and soil profile, respectively:

[9]
where the change in storage ({partial}S) (m3) over time ({partial}t) (s) is equal to the cumulative sum of inflow sources ({sum}I) (m3 s-1), minus the cumulative sum of outflow ({sum}O) (m3 s-1), and:

[10]
where {partial}Si (m3) is the moisture storage of Soil Layer i over time ({partial}t) (s), {sum}Ii(m3 s-1) is soil water inflow into Soil Layer i, and {sum}Oi(m3 s-1) is the water moisture outflow from Soil Layer i.

The models were integrated by transferring water and pesticide flux predicted as seepage by RICEWQ as prescribed boundary condition loadings into VADOFT. The top 5 cm of the profile was represented by the active sediment layer in RICEWQ. The remainder of the soil profile was represented as multiple compartments in VADOFT. The bottom of the active sediment layer is the crossed interface between two subsystems represented by RICEWQ and VADOFT. As only one pivot link between the two submodels, the water and chemical seepage out of paddy sediment predicted by RICEWQ becomes the water and chemical input of VADOFT.

RICEWQ is driven by daily weather data and operates at a subdaily time step to obtain the daily decay, seepage, runoff, and leaching amount by integration. When the soil moisture in the paddy exceeds field capacity, seepage to VADOFT occurs. As the paddy dries, soil moisture can decrease down to the wilting point through evapotranspiration. Pesticide residues in seepage water interact with the bed sediment through sorption and degradation.

The models are not implicitly coupled. That is, seepage is not dependent on the status above and below the interface at a given time. Water and chemical flux across the interface is one-dimensional. However, upward movement of soil water and chemical are accounted for in VADOFT.

Herein, the term seepage refers to water and chemical percolating from paddy sediment into the vadose zone (i.e., mass transfer from RICEWQ to VADOFT), and leaching refers to downward movement of water and chemical within the vadose zone or from vadose zone to ground water. For a full description of both models, RICEWQ and VADOFT, the reader is referred to Williams et al. (1999), Carsel et al. (1998), and Capri and Miao (2002).

Site Details and Measurements
The field study was conducted in 1997 and 1998 on a rice field of about 35 ha located in the northern part of the East Sesia zone in Italy (Fig. 1 ; Municipality of Barengo: 45° 33'30'' N, 8°31'15'' E). The agronomic practices applied in the study area were representative of the cultivation techniques usually applied in the northern Italian rice fields.



View larger version (27K):
[in this window]
[in a new window]
  		Fig. 1. Site map of the study area and layout of the test paddy showing the position of inlet and outlet floodgate and the piezometers (Ferrero et al., 2001).

 
A specific paddy field of about 2 ha (test paddy) within the study area was considered. The soil, a clayey paleosoil with fluvioglacial flood deposits, near the surface was sandy (54% sand), and became even more coarsely textured with depth down to 60 cm (Table 1).


View this table:
[in this window]
[in a new window]
  		Table 1. Soil texture of the scenario in the paddy (Ferrero et al., 2001).

 
The test substance, the herbicide cinosulfuron (1-(4,6-dimethoxy-1,3,5-triazin-2-yl)-3-[2-(2-methoxyethoxy)phenylsulfonyl]urea, was applied in the 35-ha study area (test paddy included) to fields flooded with 16 cm of water at the second leaf stage of the crop on 24 Apr. 1997 and 19 Apr. 1998. Immediately after the treatment, water circulation in the test paddy was stopped for 14 and 22 d, in 1997 and 1998, respectively. Additional interruption of water circulation occurred from 22 to 30 and from 52 to 57 d after treatment (DAT) in 1997 and from 30 to 37 and 52 to 65 DAT in 1998. When the water was circulating, the water depth in the field ranged between 9 and 11 cm.

Herbicide residues were determined in samples periodically collected from paddy water, ground water, and the paddy sediment to a 5-cm depth. For paddy sediment, when the paddy was flooded, the active sediment was usually saturated and well mixed with water (Ferrero et al., 2001). Paddy water samples were collected in the test paddy in the top 5 cm of the water layer by directly filling 1-L low-density polietyl (LD-PE) flasks. Two water samples were taken before and immediately after treatment, at 2 DAT, and then weekly for approximately 2 mo. Ground water was collected only in 1998, from two piezometer pipes and two wells. Six 1-L subsamples were taken from each location in 1-L LD-PE flasks (Bracco, Milan, Italy) on each occasion. The piezometer pipes were 7-m deep and were installed in the test paddy 10 m from each end of the southern embankment of the test paddy (Fig. 1). Ground water samples from the piezometers were collected before the treatment and at 8, 22, 29, 36, 43, 59, 57, 64, 90, and 131 DAT. The two wells were about 60-m deep, and were situated uphill and downhill about 250 and 700 m from the test paddy, respectively. The rice fields between the wells and the test paddy had never previously been treated with cinosulfuron. Samples from the wells were collected before the treatment and at 29, 57, 90, and 131 DAT.

Model Parameterization and Assumptions
Tables 2 and 3 summarize the input variables/parameters required by the RICEWQ-VADOFT model. Crop practice parameters (e.g., crop emergence and harvest date, pesticide application date and rate, and maximum area of coverage of crop), water management parameters (date and paddy water depth of initial and terminal irrigation, depth of paddy outlet, maximum irrigation and the drainage rate, and depth of active sediment layer), and pesticide properties (sediment partition coefficient and pesticide degradation half-life in water) were obtained from the field and laboratory measurements (Ferrero et al., 2001). Additional parameters were estimated from literature (Cheng, 1990; Tomlin, 1994; Williams et al., 1999; Ferrero et al., 2001; International Rice Research Institute, 1978; Luppi and Finassi, 1981; Yoshida, 1981; Mikkelsen and Dedatta, 1991; Jury et al., 1991; Dust et al., 2000; Jarvis et al., 2000; Capri et al., 2001; Miao et al., 2001; Miao et al., 2003) calculated from pedotransfer functions. The field capacity of bed sediment (cm3 cm-3), wilting point (cm3 cm-3), and bulk density (Mg m-3) were derived from Baumer-ASW/EPIC with soil parameter estimation (SOILPAR) software (Acutis and Donatelli, 2003) according to observed soil physical and chemical data, while the initial soil-moisture content (cm3 cm-3) was set to field capacity. We assumed that the depth of the paddy sediment of test plots was uniform. The initial paddy water depth was set to 10 cm, the depth of the active sediment was assumed to be 5.0 cm, and the ground-water depth was 5.14 m (Fig. 2) .


View this table:
[in this window]
[in a new window]
  		Table 2. Summary of main input parameters and variables of rice paddy used by RICEWQ.

 

View this table:
[in this window]
[in a new window]
  		Table 3. Soil physical and hydraulic parameters used for VADOFT.

 


View larger version (31K):
[in this window]
[in a new window]
  		Fig. 2. Schematic of the modeled system for the RICEWQVADOFT model.

 
The simulation interval represented the entire growing season of the paddy from 1 April to 10 September, as based on crop management practices. Paddy rice emerged on 6 April and 12 April in 1997 and 1998, respectively. Maturation/harvest occurred on 9 Sept. 1997 and 4 Sept. 1998. The dose concentrations were 39.5 and 41.5 µg L-1 in 1997 and 1998, respectively (Ferrero et al., 2001)

Within the soil core, two horizons were assumed: a first horizon (5–55 cm) sandy loam soil and a second horizon (55–514 cm) gravel-based subsoil (Table 3, Fig. 2). The soil compartment size (nodal space {Delta}z [cm]) and time step value ({Delta}t) (s) were set according to the Peclet number and Courant number criteria:

[11]
where {alpha}L is the longitudinal dispersion (cm). In the absence of site-specific {alpha}L values it is recommended that the dispersivity be chosen as one-tenth of the distance of the flow path or {alpha} = 0.1xv, where xv is the thickness of the vadose zone (cm) (Carsel et al., 1998). In the simulation, transient water flow was assigned to modules within VADOFT relating to both water flow and chemical transport (Fig. 3) . The seepage rate was determined by calibration to field experimental data (Ferrero et al., 2001). The decay coefficient of the soil porous material was obtained from soil half-life measurement in the laboratory. For both horizons, the same value of Darcy velocity (cm d-1) was used (Carsel et al., 1998) and the retardation coefficient of soil profiles was calculated with the formula:

[12]
where R is the retardation factor (dimensionless), Kd (cm3 g-1) is the distribution coefficient, {rho} is the soil bulk density (g cm-3), and {theta}s (cm3 cm-3) is the saturated water content.



View larger version (29K):
[in this window]
[in a new window]
  		Fig. 3. Simulated water flow chart of the RICEWQ and VADOFT models. 1, paddy water input including irrigation and precipitation; 2, paddy runoff; 3, evapotranspiration, 4, the bottom of the active paddy sediment layer; 5, seepage and infiltration; 6, water inflow into vadose zone of soil profile; and 7, water outflow from Soil Layer i.

 
RICEWQ requires daily precipitation (cm) and daily evaporation (cm) during the simulation period. In the RICEWQ model, evapotranspiration is assumed equal to pan evaporation (Williams et al., 1999), which is a valid assumption for an aquatic environment (Linsley and Franzini, 1979). The potential evapotranspiration was calculated with the Penman-Monteith approach using RadEst Version 3.00 (Donatelli et al., 2000), which uses daily rainfall, maximum and minimum temperatures, and relative humidity.

An accurate representation of the water balance (the combination of seepage, evapotranspiration, rainfall, irrigation, and overflow and controlled drainage) is important for accurately predicting chemical fate and transport (Capri and Miao, 2002). Some difficulties were encountered in representing the water management practices that occurred during the field study in which inlet and outlet floodgates were opened at the same time. RICEWQ as currently configured does not allow both irrigation and drainage to occur simultaneously. To conform to actual water exchange processes, several combinations of irrigation and outflow conditions were examined. The best calibration was achieved with irrigation initiating at a depth of 9 cm and overflow occurring at a depth of 11 cm (Fig. 2).

Model performance was objectively assessed by comparing the degree of agreement of the model with measurements (mass concentration in the sediment, paddy water, and ground water) in the data sets.


   RESULTS AND DISCUSSION
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
CONCLUSIONS
 REFERENCES
 
Water Balance
The water balance in the paddy is summarized in Table 4. The majority of the total applied water was lost from evapotranspiration (approximately 56–61%) with surface water outflow and net percolation accounting for roughly 6 to 10% and 29 to 39% of applied water, respectively.


View this table:
[in this window]
[in a new window]
  		Table 4. Summary of measured and simulated water balance in the paddy field.

 
Input Analysis
Two key parameters, seepage (infiltration rate) and NP value, were identified as significant variables with respect to chemical leaching predictions (Fig. 4 and Table 5). A sensitivity analysis was performed by increasing the seepage rate from the observed value, 0.23 to 0.90 cm d-1 based on values reported in literature ranging from 0.02 to 1.56 cm d-1 for similar paddy field conditions (International Rice Research Institute, 1978; Luppi and Finassi, 1981; Yoshida, 1981; Mikkelsen and Dedatta, 1991), and reducing the NP to 31 (empirical estimation reducing the number of soil layers). The increase of the seepage rate and the decrease of the number of NPs were done in an attempt to generate more leaching to match the experimentally measured concentrations in piezometers and wells. In the initial condition, when NP and seepage rate were given values of 81 and 0.23 cm d-1, the chemical leaching was limited to a depth of 2.86 m with the highest concentrations occurring at a depth of 10 cm (Fig. 4a and 4b). Changing NP resulted in qualitative and quantitative changes in chemical leaching. When NP was changed from 81 to 31, chemical leaching occurred at the bottom of the soil column (Table 5), but the highest concentrations still occurred at a depth of 10 cm (Fig. 4c,d). The reason is that with a lower NP value, the soil column is divided into fewer wider layers. Numerical dispersion is increased, so the distance traveled by the leading edge of the pesticide increases. However, this does not affect the location of the maximum concentration. NP values have no impact on chemical runoff.



View larger version (27K):
[in this window]
[in a new window]
  		Fig. 4. The modeled pesticide concentration in soil water down the soil profile on 31 Aug. 1997 (129 d after treatments [DAT]) and 1998 (134 DAT). (a) seepage rate = 0.23 cm d-1(observed) and nodal point (NP) = 81 in 1997, (b) seepage rate = 0.23 cm d-1 (observed) and NP = 81 in 1998, (c) seepage rate = 0.23 cm d-1 (observed) and NP = 31 in 1998, and (d) seepage rate = 0.90 cm d-1 (estimated) and NP = 31 in 1998.

 

View this table:
[in this window]
[in a new window]
  		Table 5. The effect of seepage rate and nodal point (NP) value on chemical mass balance in 1998.

 
When the seepage rate was increased to 0.9 cm d-1, predicted concentrations at the bottom of the soil core were closer in magnitude to observed measurements in ground water (Fig. 5 and Table 5). However, under these conditions, the predicted concentrations in runoff did not match the observed concentrations (Table 6). Overall, the best calibration to experimental data was achieved with NP set to 31 and seepage rate set to 0.23 cm d-1.



View larger version (19K):
[in this window]
[in a new window]
  		Fig. 5. The predicted daily chemical leaching concentration change at the 92-cm depth in (a) 1997 and (b) 1998. Note: BT—before the treatment.

 

View this table:
[in this window]
[in a new window]
  		Table 6. The predicted vs. observed pesticide runoff of the paddy field.

 
Pesticide Mass Balance
With calibrated conditions (NP = 31 and seepage = 0.23 cm d-1), dissipation in the paddy field (total chemical dissipation in paddy water, sediment, and crop foliage) reached 33 g ha-1 (80.5% of applied active ingredients) in 1998, and runoff estimations were 129 mg ha-1, accounting for 0.31% of applied active ingredient (Table 5). The predicted herbicide concentrations in outflow are very close to the observed (Table 6). The Pearson's correlation (SPSS 9.0) between the predicted and observed values is significant at the 0.95 level in 1997 and 1998. With a seepage rate of 0.90 cm d-1, predicted concentrations in runoff water are substantially lower than observed concentrations (Table 6).

Figure 6 illustrates the predicted and measured concentration of cinosulfuron in paddy sediment. The mean observed values of sediment concentrations are close to the predicted values at nearly every time point. However, the wide variation in the measured data makes it difficult to critically evaluate the performance of the model.



View larger version (27K):
[in this window]
[in a new window]
  		Fig. 6. The predicted vs. observed daily chemical mass in paddy sediment in (a) 1997 and (b) 1998 and their standard deviation.

 
Figure 7 shows the simulated rate of cinosulfuron seepage into the sediment over time. The rate of cinosulfuron seepage into the vadose zone increased until DAT 15 and 29 during 1997 and 1998, respectively, reaching a maximum of 199 and 231 mg ha-1 d-1, respectively. After these dates, mass loadings into the vadose zone began to decrease (Fig. 7a). By DAT 78, the cumulative seepage mass into the vadose zone reached 6920 and 7710 mg ha-1 for 1997 and 1998, respectively, accounting for 17.5 and 18.6% of applied chemical, respectively (Fig. 7b).



View larger version (20K):
[in this window]
[in a new window]
  		Fig. 7. Seepage mass out of paddy sediment to vadose zone: (a) daily and (b) cumulative.

 
Water and chemical mass seeping into the active sediment layer may not equal the mass entering the vadose soil column due to chemical sorption and water storage. The predicted amount of cinosulfuron seeping out of the paddy sediment and into the vadose zone was 18.1 and 19.2% of the applied dose in 1997 and 1998, respectively (the results for 1998 are shown in Table 5). If the paddy is drained, water can evaporate out of the sediment until the wilting point is reached. If water is added to the paddy, soil moisture increases. When soil moisture exceeds field capacity, seepage from the sediment into the underlying vadose zone will occur. Therefore, the chemical masses in the paddy sediment and chemical partitioning to sediment have a strong influence on chemical seepage and leaching to vadose zone.

The predicted amounts of herbicide remaining in the soil column at the end of the growing season for 1997 and 1998 were very low, accounting for 0.6 and 0.5% of applied a.i., respectively. Most of the herbicide mass entering the vadose zone decayed in the soil profile during the growing season. The proportion of chemical decay to the chemical inflow mass into vadose zone was 96.7 and 97.2% in 1997 and 1998, respectively.

The maximum concentration in the soil water was predicted to occur on 31 August at a depth of 10 cm for 1997 (DAT 129) and 1998 (DAT 134) (7.35 and 6.9 µg L-1, respectively, (Fig. 4). The maximum simulated depth of cinosulfuron was 92 cm, where the concentration in soil water reached 2 x 10 -5 µg L-1 in both years at DAT 131. The flux of cinosulfuron from the bottom of the soil column into ground water was insignificant. The predicted concentrations at the depth of ground water were practically zero (Table 5). Ferrero et al. (2001) found that concentrations in ground water were above detection limits, and suggested that the detections might have been caused by the large area of rice cultivation rather than by cinosulfuron applications to the paddy during the field experiment. Since cinosulfuron had not been applied in the study area, this conclusion seems unlikely. However, the ground water contamination may be explained by the high stoniness of the soil, which may have caused hydrological effects (e.g., preferential flow) not accounted for in this parameterization of VADOFT.

The RICEWQ-VADOFT model was calibrated with the 1997 dataset, while the predictions in 1998 were used to validate the calibrated model. Two statistical indices, root mean squared error (RMSE) and the reduction of error test (RE), were used to evaluate the model performance. Of them, RMSE calculated from observed and predicted data was used to express the overall fit of the model simulation:

where Pi are the predicted values, Oi the observed values, Ô the average of the observed values, n the number of observations. With a perfect fit, the value would be zero (Loague and Green, 1991).

RE was used to indicate the calibration quality:

[14]
where Yi and Yi are the actual and estimated data for the verification period, and c is the mean of the actual data in the calibration period. Unfortunately, there is no formal significance test for RE. As a guide, an RE > 0.0 refers that model performance for the verification period is better than for the calibration period, while an RE < 0.0 means that the model predictions for the verification period are worse than for the calibration period. Thus, a negative RE is reason for rejecting the validity of the model (Cook et al., 1987).

For the verification of the predicted pesticide runoff in Table 6, RMSE for two scenarios with seepage rates of 0.23 and 0.90 cm d-1 are 155.11 and 232.12 in 1997, 139.86 and 245.65 in 1998, respectively, which illustrated that the predictions fit the observation reasonably well, particularly for the scenario with seepage rate of 0.23 cm d-1. The reduction of error tests for the two scenarios with seepage rates of 0.23 and 0.90 cm d-1 are 0.99 and 0.75, respectively. Both are >0.0. The performance of the model in 1998 is better than with the 1997 dataset.

In a previous paper (Miao et al., 2003), an uncertainty analysis with a Monte Carlo stochastic approach and stepwise regression was performed for identification of the reliability of model prediction and screening of crucial process variables. The results suggested that the runoff predictions conform to a log normal distribution with a short right tail, and that the model runoff uncertainty is acceptable and the model estimation is reliable at field scale due to its narrow spread of the uncertainty distribution (Fig. 8a) and the center position of the measured runoff values in the prediction uncertainty distribution (47% of exceedence of model estimations over the field measurement [Miao et al., 2003; Warren-Hicks et al., 2002]). However, the contribution of parameters to prediction uncertainty (root of uncertainty [Keller et al., 2001]) is varied. Spatial parameters including Kd and VBIND (mixing sediment depth to allow pesticide direct partitioning to bed [cm]) together with management parameters like the date and rate of application, and climatic conditions are the main contributors to the uncertainty, so these parameters should be carefully parameterized (Fig. 8b).



View larger version (27K):
[in this window]
[in a new window]
  		Fig. 8. The paddy runoff uncertainty distribution and parameter contribution to uncertainty of the RICEWQ-VADOFT model: (a) the uncertainty distribution (400 iterations) and (b) parameter contributions to model uncertainty.

 

   CONCLUSIONS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
REFERENCES
 
This study investigated the capability of the RICEWQ-VADOFT model to simulate pesticide fate in a rice paddy, including leaching and runoff. The models were parameterized to simulate a field data set obtained in a 2-yr field experiment in 1997 and 1998 in northern Italy. The results demonstrate that the RICEWQ and VADOFT models in combination can predict runoff and dissipation in paddy water. However, work is still required to accurately predict leaching at the bottom of the soil column. Water management and seepage rate are site-specific model parameters that have a strong influence on the predicted fate of the pesticide. Nodal spacing is also an important parameter to define to ensure accurate descriptions of dispersion within the soil column.

From the European registration point of view, detailed scenario definitions are needed to correctly establish model inputs. At a low tier, two to three scenarios should be selected to represent the variability in site characteristics of the European rice areas in Greece, Italy, France, Spain, and Portugal. Once the representative scenarios were defined, the model would be an effective tool also in higher-tier risk assessment for calculating the predicted environmental concentration in water and soil as well as for risk mitigation.

From a scientific point of view, a desirable model enhancement would be to simulate water management strategies with simultaneous irrigation and drainage. An implicit coupling between the two models would be desirable to represent the influence of pressure head gradients on infiltration rates and transient upward flow.

Additional model testing is recommended to fulfill the requirements of a full model evaluation. For example, future field experiments should use a Br tracer together with shallow monitoring wells to better define the movement of water and non-reactive chemicals in the saturated zone beneath rice paddies.


   ACKNOWLEDGMENTS
 
This study was carried out within the framework of the Italian national project of MIUR 2001-2002 (Herbicide Fate in Paddy Rice). The constructive comments of anonymous reviewers and editor of the journal on the original manuscript are appreciated.


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


Related articles in JEQ:

This Issue in Journal of Environmental Quality

JEQ 2003 32: 1931-1938. [Full Text]  



This article has been cited by other articles:


Home page
 Vadose Zone JHome page
D. G. Karpouzas, E. Capri, and E. Papadopoulou-Mourkidou
Basin-Scale Risk Assessment in Rice Paddies: An Example Based on the Axios River Basin in Greece
Vadose Zone J., March 8, 2006; 5(1): 273 - 282.
[Abstract] [Full Text] [PDF]

Home page
 J. Environ. Qual.Home page
Z. Miao, M. Trevisan, E. Capri, L. Padovani, and A. A. M. Del Re
Uncertainty Assessment of the Model RICEWQ in Northern Italy
J. Environ. Qual., November 1, 2004; 33(6): 2217 - 2228.
[Abstract] [Full Text] [PDF]

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 ISI 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 HighWire
Right arrow Citing Articles via ISI Web of Science (12)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Miao, Z.
Right arrow Articles by Capri, E.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Miao, Z.
Right arrow Articles by Capri, E.
Agricola
Right arrow Articles by Miao, Z.
Right arrow Articles by Capri, E.
Related Collections
Right arrow Agricultural Pesticides
Right arrow Pesticides
Right arrow Coupled Flow/Transport Models

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