| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
a Dep. of Civil Eng., Univ. of Mississippi, 208 Carrier Hall, P.O. Box 1848, University, MS 38677-1848
b USDA-ARS, National Soil Tilth Lab., 2150 Pammel Dr., Ames, IA 50011-4420
c Bayer CropScience Res. Park, 17745 S. Metcalf Ave., Stilwell, KS 66085
d USDA-ARS, 2150 Centre Ave., Building D, Suite 200, Ft. Collins, CO 80526
* Corresponding author (gafox{at}olemiss.edu)
Received for publication February 20, 2004.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSAPPENDIXREFERENCES |
|---|
Abbreviations: EF, express fraction • ISO, isoxaflutole • MUUF, map unit use file soil database • NOF, normalized objective function • RZWQM, Root Zone Water Quality Model
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSAPPENDIXREFERENCES |
|---|
Pesticide movement to drains is commonly simulated with macropore flow (Chen and Wagenet, 1992; Saxena et al., 1994; Kumar et al., 1998; Villholth and Jensen, 1998; Villholth et al., 1998). However, researchers understand little about the connectivity of macropores with subsurface drainage systems. Immediate breakthrough of solutes and pesticides in subsurface drainage indicates extraordinarily efficient transport by preferential pathways, hypothesized to be the result of a few drain-connected macropores by Villholth et al. (1998) and Shipitalo and Gibbs (2000). A field study in Ohio indicated that macropores created by earthworm burrows directly transfer water to subsurface drains (Shipitalo and Gibbs, 2000). Infiltration rate correlated to distance from the subsurface drains: the rate at which water entered earthworm burrows declined with the log of distance from the subsurface drains. Direct transfer of water was limited to a distance of 0.5 m on either side of the drain (Shipitalo and Gibbs, 2000). A field and modeling study in Denmark observed extremely fast breakthrough of tracer generated by a few, drain-connected macropores (Villholth and Jensen, 1998).
The objective of this research is to investigate the interrelationship between preferential flow through macropores and artificial subsurface drainage on conservative tracer and pesticide transport. The Root Zone Water Quality Model (RZWQM) was used to simulate flow from a subsurface drained field in Allen County, Indiana, after application of potassium bromide (KBr) tracer and isoxaflutole (ISO), the active ingredient in BALANCE herbicide, during the 2000 growing season. This research evaluates RZWQM's ability to predict chemical transport to subsurface drains using minimal calibration and proposes a modification to RZWQM to account for macropores directly connected with subsurface drains.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSAPPENDIXREFERENCES |
|---|
The Green and Ampt (1911) equation is used to calculate infiltration rates into the soil profile. The model simulates subsurface drainage using the integrated Hooghoudt equation developed by Bouwer and van Schilfgaarde (1963). This equation predicts the rate of fall of the water table in relation to the subsurface drain:
 | [1] |
, vs. h relationships:
 | [2] |
 | [3] |
r is the residual water content (cm3 cm–3), s is the saturation water content (cm3 cm–3), C2 is the second intercept on the conductivity curve, N1 is the first exponent on the conductivity, and B is given by the following expression:
 | [4] |
The model routes precipitation that exceeds the infiltration rate into macropores based on a flow capacity limit determined by Poiseuille's law (Malone et al., 2003). The model then evenly distributes water entering into macropores among the number of effective macropores per unit area:
 | [5] |
The volume of effective macropores per unit volume of soil is macro* (cm3 cm–3), as defined in detail by Malone et al. (2003), and rp is the average macropore radius (cm). Flow is sequentially routed downward through continuous macropores in 1-cm increments and allowed to laterally infiltrate into surrounding, unsaturated soil based on a radial infiltration rate in macropores. This radial infiltration rate is dependent on a lateral sorptivity reduction factor that represents impedance of infiltration by organic material on macropore walls or macropore wall compaction (Malone et al., 2001).
The RZWQM simulates chemical transport through a variety of mechanisms. Rainfall transports chemicals from the soil to overland flow. The model simulates chemical washoff from plant foliage and mulch. The RZWQM routes water and chemicals through macropores and allows reactions through chemical partitioning with soil surrounding the macropores. The model assumes a linear isotherm between chemical adsorbed to soil and chemical in solution (Malone et al., 2001).
Field Experiments
The research site was a 30.4-ha isolated field located in Maumee Township, Allen County, Indiana, approximately 1.6 km east of the town of Woodburn (Fig. 1)
. The field consists of Hoytville silty clay (fine, illitic, mesic Mollic Epiaqualfs) with slope less than 2%. The field contained systematic subsurface drain lines (polyurethane pipes) with a 10-m spacing running east–west that connect to a drain system emptying at the northeast and southeast corners of the field. The subsurface drains were at a depth of 0.9 to 1.2 m. No-till agricultural management practices were used at the site. Crop history records indicate that corn was planted at the test site during 1998, followed by soybean [Glycine max (L.) Merr.] in 1999, and then corn again in 2000.
|
|
|
|
|
Drain flow was measured using an Isco Model 730 Bubble Flow Meter (Isco, Lincoln, NE). The level of water in the subsurface drains and flow was determined based on the pipe diameter, smoothness, and slope. Flow data were missing for high-flow events due to flooding of the sampling site. Therefore, available flow data were the regression limbs of subsurface drainage after storm events. Multiple water samples from the subsurface drains were collected daily starting 5 d after KBr application for the duration of the growing season. The water and soil samples were analyzed to determine the extent of leaching to the drainage system. The drain sampling points were selected where the subsurface drain systems empty into the collector ditches. The drain outlets were modified to allow use of an automated sampling device. A sample pickup was placed where the drains were first exposed at the ditch bank. An Isco model 6700 sampler (Isco, Lincoln, NE) was installed at each drain and programmed to collect a water sample every 80 min (18 samples every 24-h period). Each day the filled sample bottles were collected and new empty bottles placed in the automated sampler. The freshly collected samples were labeled, dated, and placed in frozen storage.
Tracer samples were analyzed using an ion chromatograph (USEPA, 1993). Soil samples were extracted with 100 mL of HPLC-grade water after thawing and then centrifuged. An aliquot of the extract was then diluted with HPLC-grade water before analysis by ion chromatography. The limit of detection of the method is 0.1 mg L–1. Water samples were analyzed for residues of isoxaflutole using a liquid chromatography/mass spectrometry system (Smitley et al., 2000). The limit of quantification is 0.010 µg L–1. The limit of detection is 0.003 µg L–1. Samples were fortified with 13C labeled standards of the three analytes. Residues of isoxaflutole were extracted from water using a RP-102 resin cartridge (Applied Separations, Allentown, PA), and then removed with acetonitrile–methanol. All residue analysis was accomplished by LC-MS-MS (Applied Separations, Allentown, PA) using a C-8 column. The typical analytical set included 4 to 20 field samples, at least one untreated control sample and at least one fortified procedural recovery sample. Each set of sample injections was bracketed by calibration standard injections of appropriate analyte concentrations.
RZWQM Modeling
The soil profile was divided into five layers for input into RZWQM: 0 to 15 cm, 15 to 30 cm, 30 to 107 cm, 107 to 152 cm, and 152 to 296 cm. The model was capable of deriving critical soil hydraulic parameters (i.e., porosity, conductivity, field capacity, and water retention parameters) based on minimum input of the particle-size distribution and bulk density. Therefore, particle-size distribution and bulk density from the soil samples were input into the model to generate a first estimate of the soil properties. These soil samples suggested a uniform soil property distribution with depth in the profile. Additional soil parameters were available from the Map Unit Use File (MUUF) soil database (Baumer et al., 1987; Rawls et al., 2001) for Hoytville silty clay. Soil hydraulic data can be generated by MUUF for use in different hydrologic–water quality simulation models. Data from the soil database compared reasonably well to laboratory measurements of soil texture, organic matter, and bulk density from the field samples. The surface crust conductivity and soil macroporosity parameters including the macropore radius, effective macroporosity, and lateral sorptivity reduction factor were obtained from selected literature. The same macropore parameters were input for all depths. Management practices for the field were entered into the model including the applications of KBr and ISO. Weather data were input based on measured hourly data from the weather station.
The model was calibrated by adjusting the modified Brooks-Corey parameters (Brooks and Corey, 1964), vertical saturated conductivity, and lateral saturated conductivity until model predictions matched observed subsurface drainage collected during the experiments from Day 120 to 190 of the 2000 growing season. Potassium bromide transport through the soil matrix and macropores was then simulated. The current released version of RZWQM simulated macropore flow to cease at the water table. Therefore, the model assumed no connection between macropores and subsurface drains. The RZWQM was modified to include a contributing area parameter, or express fraction, relating the percentage of macropores in direct hydraulic connection to subsurface drains. Observed Br concentrations were compared with predicted concentrations both before and after model modification. Model evaluations were based on linear regressions of predicted vs. measured subsurface drainage and concentrations. An additional test based on an acceptance criterion as quantified by a normalized objective function (NOF) was utilized (Hession et al., 1994; Kornecki et al., 1999). This function is the ratio of the standard deviation of differences (STDD) to the overall mean (Xa) of the observed parameter:
 | [6] |
|
RESULTS AND DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSAPPENDIXREFERENCES |
|---|
|
|
Table 5 presents the calibration values for hydraulic conductivity and Brooks-Corey parameters. Vertical, saturated conductivities were reasonable compared with the MUUF soil database that suggested hydraulic conductivities between 0.1 and 1.0 cm h–1. Lateral, saturated conductivities were assumed three times greater than the vertical conductivity, similar to the default RZWQM parameters. The bubbling pressure head and particle-size distribution index in the calibrated model resemble a loam soil much more than a silty clay soil. In fact, the primary factor influencing model predictions of subsurface drainage was the bubbling pressure head, especially for the surface layer. Adequate subsurface drainage was not simulated in RZWQM without reducing the surface layer bubbling pressure head. Requiring a lower bubbling pressure head than expected follows a pattern exhibited in studies with RZWQM. For example, Malone et al. (2003) required a lower bubbling pressure head with RZWQM for the soil surface layer than measured experimentally from soil samples. Determining the validity of such parameters requires experimental verification from soil samples. These parameters were not measured during soil sample collection. An alternative verification was possible with the Rosetta model (Schaap and Leij, 1998), which estimated soil moisture characteristic curves and unsaturated hydraulic conductivity curves based on measured soil particle-size distribution and bulk density using the van Genuchten (1980) relationships. Predicted unsaturated hydraulic conductivity derived using the van Genuchten (1980) relationship for the soil samples compared reasonably well with the modified Brooks-Corey relationship (Brooks and Corey, 1964) based on calibrated soil parameters in RZWQM.
Model Evaluation
Figures 3 and 4
compare predicted Br and ISO concentrations for the calibrated, hydrologic model vs. the observed Br concentrations from the subsurface drains. Observed Br and ISO concentrations indicate rapid delivery of the tracer and pesticide to the subsurface drains. Subsurface drainage contained Br as soon as 5 d after application and ISO as soon as 3 d after application. The model met the acceptance criterion for Br (NOF = 0.8) but not ISO (NOF = 1.3). A linear regression between predicted and measured Br concentration resulted in a slope of 0.42, intercept of 1.30, and R2 = 0.25. The model underpredicted concentrations following rainfall events soon after application of the KBr tracer. For example, observed Br concentrations in the two samplers on Day 131 ranged between 16.0 and 18.5 mg L–1 compared with a predicted concentration of 4.6 mg L–1. The model underpredicted Br concentration relative to the average observed concentration until Day 141. The pesticide breaks down rapidly such that only one peak in concentration was observed during the study period. Concentrations in the two samplers ranged between 1.3 and 1.4 µg L–1 on Day 131. The model failed to capture this significant peak in predicting a concentration of approximately 0.1 µg L–1. A linear regression between predicted and observed ISO concentrations resulted in a slope of <0.1.
|
|
Model Modification
The original macropore component of RZWQM routes the generated surface runoff downward through the macropore until it meets the water table. Lateral infiltration and mixing with soil surrounding the macropores can occur as water flows through the macropores. If water and chemical reach the top of the water table, then the water table is raised from this available water one layer at a time. As this available water is added to the layer immediately above the water table, the chemical is also mixed into that layer matrix. After mixing, the model raises the water table 1 cm and then fills the next layer until routing all the water. This results in a concentration bulge at the top of the water table as verified by the Br simulations.
The RZWQM model was modified by including a contributing area parameter relating the percentage of macropores in direct hydraulic connection to subsurface drains. When macropore flow first reaches the top of the water table, a macropore–subsurface drain express fraction (EF) of the water and chemical is routed directly into the subsurface drain through macropores. The EF is proportional to the available water and chemical that reaches the soil depth of subsurface drainage. The remaining water and chemical is allowed to fill and mix with the layers above the water table, as described above. The modified model routes water from a percentage of the macropores near subsurface drains directly into the drains to simulate macropore–drain connectivity. The EF will influence pesticide transport when there are macropores present and excess water is available on the soil surface during an infiltration time step. With this assumption, water will be available as an EF when water either reaches the drain directly (i.e., water table below subsurface drain) or the top of the water table (i.e., water table above subsurface drain). In the former, the macropore water front is decreased by the EF fraction and the remainder is added to the top of the water table. In the later, the macropore water front reaches the water table before the subsurface drain and then the model abstracts chemical as a continuous column of water above the subsurface drain that is instantaneously mixed and transported. This modification stems from the hypothesis that even if the average water table height is above the drain, macropore flow can occur directly into drains. The water table peaks halfway between drains and recedes to the drain depth at the drains. In cases where the soil matrix above the drains may be saturated for a short period of time because of capillary tension in soil, the macropores should still be open because the tension is very low in a 3-mm diameter macropore. Even if there is water in the macropore, mixing should allow flow from the surface directly into the subsurface drain.
The EF is input into RZWQM with the macroporosity parameters (soil physical property parameter). The EF is specified as the decimal percentage of effective macropores in direct connection with subsurface drains. Approaches that are more mechanistic are needed to derive EF but deterministic equations do not currently exist for analyzing macropore–subsurface drain connectivity as functions of soil and hydraulic conditions. Therefore, the model modification is an initial attempt to account for directly connected macropores and subsurface drains.
Bromide and ISO transport using the modified model were sensitive to EF, with greater subsurface drainage concentrations predicted as EF increased. Including an EF improved the response of the model in predicting immediate breakthrough of the tracer and pesticide. Model results with an EF = 0.02 or 2%, where the value of EF was determined by fitting the data to the measured Br concentrations, are presented in Fig. 3 and 4. An express fraction of 2% did not influence the simulated subsurface drainage. However, large values of EF will influence simulated drainage volumes. The first two simulated peaks in Br and first simulated peak in ISO matched the observed concentrations reasonably well. The modified model overpredicted ISO on Day 140. Predicted Br and ISO concentrations at Day 131 (i.e., the first peak) were 14.2 mg L–1 and 1.7 µg L–1, respectively. A linear regression between predicted and observed Br and ISO concentrations resulted in slopes of 0.62 and 0.81, intercepts of 0.70 and 0.01, and R2 of 0.43 and 0.60, respectively. The normalized objective functions after model modification were 0.7 for Br and 0.9 for ISO.
Fast breakthrough by drain-connected macropores mainly influenced initial breakthrough when concentration differences between the macropore and matrix flow domains were large. Therefore, simulated peaks in Br after Day 143 remained low compared with observed concentrations. The discrepancies in later peaks were the result of the model transporting more Br through the soil matrix as compared with measured concentrations. Consequently, less Br was available at the soil surface for transport through macropores during rainfall events. Figure 5a demonstrates that observed concentrations in the soil for depths between 0 and 15 cm match predictions by RZWQM. However, the model appeared to predict a Br wave or pulse through the soil profile as indicated by the increases in concentration in the lower soil layers (15–65 cm) as shown in Fig. 5b–d. Observed data do not indicate such a concentration wave but a constant Br concentration between 15 and 65 cm below the soil surface. In fact, observed Br concentrations below the upper soil layer (0–15 cm) remained near 2 mg L–1 throughout the tracer study.
|
|
CONCLUSIONS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSAPPENDIXREFERENCES |
|---|
In general, the modified model better represents the possible interrelationship between macropores and subsurface drains to capture early peaks in tracer and pesticide concentrations following application. Further research should be conducted to test the modified model under different climate, soil, subsurface drain configuration, and chemical applications. In addition, the degree of direct connectivity between macropores and subsurface drains needs to be characterized using approaches that are more mechanistic. A constant percent of macropore flow (e.g., 2%) directly entering drains is used in this manuscript. This is a very simplified assumption, and further investigation into directly connected macropores may indicate that the express fraction should change with varying saturated-unsaturated flow conditions above the tile drain. At this time, however, our understanding of the dynamics of the express fraction under different antecedent water content and rainfall intensities is insufficient.
|
APPENDIX |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSAPPENDIXREFERENCES |
|---|
C, coefficient relating the fall midway between the drains to the average fall of the water table
C1, saturated hydraulic conductivity (m s–1)
C2, second intercept on the conductivity curve
De, effective depth of the impermeable layer below the drain center (m)
f, drainable pore-space
h, matric suction head (cm)
K, saturated hydraulic conductivity of the soil (m s–1)
K(h), unsaturated hydraulic conductivity (m s–1) as a function of matric suction head, h (cm)
L, distance between the drains (m)
mo, mt, initial height and height at time t of the water table above drain centers midway between drains (m)
macro*, volume of effective macropores per unit volume of soil (cm3 cm–3)
N1, first exponent on the conductivity
N2, exponent for the conductivity curve
nmacro*, number of effective macropores per unit area
NOF, normalized objective function
rp, average macropore radius (cm)
S, bubbling pressure (cm)
STDD, standard deviation of differences
t, time (s)
xi, ith observed value
Xa, mean of the observed parameter
yi, ith predicted value
, soil water content (cm3 cm–3)
r, residual water content (cm3 cm–3)
s, saturation water content (cm3 cm–3)
|
ACKNOWLEDGMENTS |
|---|
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSAPPENDIXREFERENCES |
|---|
Related articles in JEQ:
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| 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 | |||
