IDENTIFYING LONG‐TERM PREFERENTIAL AND MATRIX FLOW RECHARGE AT THE FIELD SCALE J. S. Tyner, W. C. Wright, R. E. Yoder

ABSTRACT. Preferential recharge from agricultural production fields can lead to agrochemical contamination of groundwater due to short‐circuiting of the soil matrix. This research determines the relative amounts of preferential and matrix flow recharge from a no‐till agricultural production field. The conservative tracer potassium bromide was surface applied to a 0.37ha bermed field plot at a rate of 900 kg ha-1 as bromide. The site was instrumented to monitor precipitation and runoff from the bermed field plot. Three years after the bromide application, twenty‐one 3.65 m continuous soil cores were collected from the bermed plot and analyzed for bromide. Based solely on the shape of the bromide profiles, most of the 21 soil cores showed no apparent preferential flow. However, up to 83% of the infiltrated bromide mass was absent from individual soil cores, presumably due to preferential flow. Using a mass balance approach, we estimated that 58% of the recharge was due to preferential flow and 42% was due to matrix flow. Keywords. Bromide, Macropore flow, Matrix flow, Preferential flow, Recharge, Tracer.

A

riculture can contribute to nonpoint‐source water pollution by the surface and subsurface migration of nutrients and pesticides (Stone et al., 1998). Es‐ timating the subsurface transport of solutes is challenging, particularly if preferential flow is present. If ma‐ trix flow is the dominant infiltration mechanism, many sol‐ utes (including most pesticides and insecticides) will adsorb and degrade near the surface where conditions for biodegra‐ dation are optimal. However, preferential flow can enable the infiltrate to bypass much of the upper soil matrix and rapidly reach greater depths where biodegradation rates are typically much lower (Iragavarapu et al., 1998; Malone et al., 2004; Ray et al., 2004). Jaynes et al. (2001) applied bromide to a loam soil with drainage tiles and found that a non‐preferential flow model could predict the soil bromide residue profiles, but the model could not predict the rapid breakthrough and concentration of bromide to the tile drains. Malone et al. (2000) concluded that the contaminant transport model GLEAMS could predict herbicide concentrations in 30 × 30 × 30 cm blocks of no‐till silt loam soil, but predicted herbicide percolate concentra‐

Submitted for review in July 2006 as manuscript number SW 6582; approved for publication by the Soil & Water Division of ASABE in August 2007. The use of brand names does not imply endorsement by the University of Tennessee. The authors are John S. Tyner, ASABE Member Engineer, Associate Professor, and Wesley C. Wright, ASABE Member Engineer, Research Associate, Department of Biosystems Engineering and Environmental Science, University of Tennessee, Knoxville, Tennessee; and Ronald E. Yoder, ASABE Member Engineer, Professor and Head, Department of Biological Systems Engineering, University of Nebraska, Lincoln, Nebraska. Corresponding author: John S. Tyner, Department of Biosystems Engineering and Environmental Science, University of Tennessee, 2506 E. J. Chapman Dr., Knoxville, TN 37996‐4531; phone: 865‐974‐7130; fax: 865‐974‐4514; e‐mail: [email protected].

tions were typically 0.8% to 17% of those measured. Köhne and Gerke (2005) applied bromide to a loam soil with drain‐ age tiles and were unable to model the concentrations in tile drain effluent using a non‐preferential (CDE type) transport model. However, using a preferential (mobile‐immobile type) model, they yielded satisfactory results for the tile ef‐ fluent bromide concentrations. Larsson and Jarvis (1999) found that the dual‐porosity flow model MACRO, which takes into account preferential flow, could satisfactorily depict long‐term nitrate leaching. Simulation with this model of a substantial precipitation event occurring soon after fertilizer application increased ni‐ trate leaching by 34% when compared to a simulation with no preferential flow. By conducting simulation analysis with the GLEAMS model, Iragavarapu et al. (1998) concluded that matrix flow was not adequate to predict the leaching depth of surface‐applied bromide. Yoder (2001) surface applied bromide to a bermed plot with 23 shallow monitoring wells installed in the surrounding 10 ha field. Bromide was detected in the majority of shallow monitoring wells within two to three months after being ap‐ plied. The first detection of bromide in a shallow monitoring well occurred at a distance of 150 m only nine weeks follow‐ ing application. The rapid transport of bromide from the bermed area to the shallow wells indicated significant sub‐ surface water movement via preferential flow, yet the per‐ centage of the bromide traveling in the preferential domain was unknown. The objective of this study was to quantify the relative amounts of preferential and matrix flow recharge by applying a mass balance approach to a surface‐applied bromide tracer using the same plot as Yoder (2001). We also attempted to de‐ termine if preferential flow could be discerned from analyz‐ ing the shape of soil core bromide profiles without knowledge of the applied bromide mass.

Transactions of the ASABE Vol. 50(6): 2001-2006

E 2007 American Society of Agricultural and Biological Engineers ISSN 0001-2351

2001

MATERIALS AND METHODS SITE DESCRIPTION A 10 ha water transport research field site was developed in 1991 at the Ames Plantation located near Grand Junction, Tennessee, and resides within Major Land Resource Area 134 (Southern Mississippi Valley Silty Uplands). Within the 10 ha field site, a 0.37 ha bermed plot was constructed to iso‐ late the surface water movement from the remainder of the field site. The elevation of the field site is 150 m above mean sea level, and the soil is a Quaternary Age loess over alluvial deposits (Hardeman, 1966). The loess is approximately 1.7m thick and is composed of three depositional units: Peo‐ ria, Roxana (Farmdale), and Loveland (youngest to oldest). In most locations at the field site, differences between the Peoria and Roxana loess deposits could not be determined by chemical or morphological data (Livingston, 1993). The al‐ luvial deposits consist of quartz sands and intermittent clay lenses (Russell and Parks, 1975). The interface between the loess and underlying alluvial deposits temporally perches water after moderate precipitation events (Freeland et al., 2002). A permanent water table resides at a depth of approxi‐ mately 15 m. The field site was double cropped with winter wheat and soybeans from 1993 to 1995, cotton from 1996 to 1998, and corn from 1999 to 2001. A winter wheat cover crop was planted in the fall of 2001, followed by soybeans in 2002. During the term of this study, no harvesting was conducted within the bermed plot. Runoff (Q) from the bermed plot was measured with a 0.6m H‐flume. An ISCO model 3230 bubble meter moni‐ tored flume stage and triggered an ISCO model 3700 sampler to collect flow‐proportional runoff samples. A tipping‐ bucket rain gauge measured precipitation (P). POTASSIUM BROMIDE APPLICATION Granular potassium bromide (KBr) was surface applied to the bermed plot at a rate of 300 kg ha-1 as bromide in the spring of 1993 (Yoder et al., 1998). In February 2000, a con‐ tinuous 40 mm dia. × 3 m long soil core was collected from the bermed plot. Laboratory analysis of the soil core revealed that only a small amount of bromide persisted in the soil pro‐ file: an average concentration of 4.5 mg L-1 and a maximum concentration of 8 mg L-1 at a depth of 2.7 m. Additionally, bromide had not been detected in any of the monitoring wells for more than a year. A second application of granular KBr was applied to the bermed plot on 6 April 2000 at a rate of 900kg ha-1 as bromide. To ensure uniform distribution of the KBr, the plot was subdivided into a grid with 64 subsections, and equal amounts of KBr were broadcast by hand onto each subsection. SOIL SAMPLE COLLECTION AND ANALYSIS On 5 February 2003, continuous 3.65 m long soil cores were collected at 21 locations (fig. 1) within the bermed plot using a truck‐mounted Giddings S‐M hydraulic soil sampling machine with a 45 mm dia. × 910 mm long slotted soil tube. A portion of the soil samples were oven dried to measure the volumetric water content profile from each core. Bromide concentrations from the soil pore water were measured by mixing 50 g of field moist soil with 40 mL of deionized water to produce slurries that were shaken for 1 h and allowed to settle for 18 h. A 10 mL aliquot of supernatant

2002

Figure 1. Diagram of the 0.37 ha bermed plot and 21 soil‐core sampling locations (small circles).

was collected from the settled slurries using a syringe and passed through a 0.45 mm Acrodisc IC syringe filter. The su‐ pernatant was analyzed for bromide using an IonPac AS9‐HC analytical column in conjunction with a Dionex DX‐100 liq‐ uid ion chromatograph with a suppressed conductivity detec‐ tor (APHA, 2005). BROMIDE MASS BALANCE A control volume was defined that encompassed the area within the bermed plot and from the soil surface to the soil core depth (3.65 m). A bromide mass balance was calculated from 6 April 2000, the date of the second bromide applica‐ tion, to 5 February 2003, the date of the soil core collections. Figure 2 shows a cross‐section of the control volume where Bapplied is the mass of bromide applied to the plot, BQ is the mass of bromide that exited with runoff, Binfil is the mass of bromide that infiltrated, BT is the mass of bromide that was transported back to the surface by transpiration, Bpref is the mass of bromide that exited the plot from the preferential flow domain, Bmatrix is the mass of bromide that exited the plot from the matrix flow domain, and Bsoil is the bromide mass remaining within the control volume. The crops planted in the bermed plot likely took up some of the applied bromide each year through transpiration. Ira-

Figure 2. Illustration of the control volume and flow paths: Bapplied is the mass of applied bromide, BQ is the mass of bromide in runoff, Binfil is the mass of infiltrated bromide, BT is the mass of bromide that was trans‐ ported back to the surface by transpiration, Bpref is the mass of bromide that exited the plot by preferential flow, Bmatrix is the mass of bromide that exited the plot by matrix flow, and Bsoil is the bromide mass remaining within the control volume.

TRANSACTIONS OF THE ASABE

Bin fil = Bapplied − BQ

(2)

and the remainder of the infiltration enters the matrix flow domain: B I matrix = soil = 1 − I pref (3) Bin fil A version of equation 3 that includes recycling of bromide for a single year can be described by equations that sum the contributions to Bsoil : Bsoil = Bin fil I matrix − BT + BT I matrix

(4)

BT = Bin fil I matrix Recyc

(5)

where Recyc is the fraction of the bromide within matrix flow that is taken up by plants and subsequently released at the sur‐ face. The left, middle, and right terms on the right side of equation 4 relate to: (1) the bromide that initially enters the matrix flow domain, (2) the bromide extracted by transpira‐ tion, and (3) the transpiration derived bromide that re‐enters the matrix domain, respectively.

Vol. 50(6): 2001-2006

0% bromide recycling 15% bromide recycling 30% bromide recycling

100

I matrix and I pref (%)

gavarapu et al. (1998) found that plant uptake ranged from 5% to 10% in the first year of the applied 197 kg ha-1 of bro‐ mide for a variety of crops including corn, soybean, alfalfa, and oats. Kung (1990) found that potato plants took up half of the 111 kg ha-1 of applied bromide, and Schnabel et al. (1995) found that ryegrass removed an average of 15% of bromide when the bromide application rate was 42 kg ha-1. However, in this study, no bromide was removed from the bermed plot due to plant uptake because no harvest was con‐ ducted. Therefore, bromide taken up by plants was subse‐ quently re‐deposited at the surface as the biomass decomposed. Scorza Júnior et al. (2004) documented this process occurring on winter wheat grown on a clay soil in the central Netherlands. This cycling of a portion of the bromide may have caused an apparent retardation in the long‐term vertical bromide velocity within the matrix flow domain. It also gave additional opportunity for recycled bromide to en‐ ter the preferential flow domain. Bromide that exited the control volume as runoff (BQ ) was calculated from the product of the measured runoff volumes and the flow‐proportional samples that were collected and measured for bromide. The bromide that infiltrated the soil surface entered either the matrix flow domain or the macro‐ pore flow domain. Additionally, there was some transport of bromide between these two domains. Because macropores are large and only filled with water during and shortly after precipitation events, any uptake of bromide by transpiration was likely from the matrix flow domain. Given the depth of the soil cores and the shape of the mea‐ sured bromide concentration profiles (presented in the Re‐ sults section), we concluded that the loss of bromide from the base of the control volume was almost entirely due to prefer‐ ential flow. We also assumed that all bromide within the pref‐ erential flow domain exited the control volume. Therefore, the bromide remaining in the control volume during the core sampling was within the matrix flow domain. The fraction of infiltration that enters the preferential flow domain can be calculated from: Bin fil − Bsoil (1) I pref = Bin fil

75

I pref

50

I matrix

25

0 0

25

50 B soil / B infil (%)

75

100

Figure 3. Comparison of Ipref and Imatrix with varying amounts of re‐ cycled bromide.

Inserting equation 5 into equation 4 and solving for Imatrix yields: I matrix =

(

)

Bin fil Recyc − 1 + Bsoil

2

(

Bin fil 1− Recyc Bsoil

2

2 Bin fil Recyc

)2

+

4 Bin fil Recyc Bsoil

(6)

We solved equation 6 for Imatrix over a range of conditions (fig. 3). A given ratio of Bsoil to Bmatrix (x‐axis) dictates (eq.3) the relative amount of preferential flow (y‐axis). Three sets of solutions are presented representing Recyc equal to 0%, 15%, and 30%. Any error in predicting Imatrix due to neglecting BT is at a maximum, but still small, when Imatrix and Ipref are equivalent. NUMERICAL MODELING The water and bromide transport beneath the plots was nu‐ merically modeled using Hydrus 1‐D version 2.02 (Simunek et al., 1998). This model solves the Richards equation for one‐dimensional transport of water and solutes in variably saturated media and includes a sink term to account for water lost by evapotranspiration. This version of Hydrus 1‐D does not provide a means to account for macropore flow. Modeling was completed to determine if the measured bromide profiles could be predicted using a matrix‐flow only model (no ma‐ cropore flow) and to estimate Bmatrix given that assumption. The water flux at the upper boundary condition was set to a constant value of (P - Q) × Imatrix . We assumed that the Q at individual locations was equal to the average Q measured from the entire bermed area. The lower boundary condition was set as a unit potential gradient. The concentration of bro‐ mide within the infiltrating water was set at a constant value for the first day of the simulation, such that the total mass of infiltrated bromide was equal to the mass of bromide mea‐ sured in the soil core. For the remainder of the simulation, the infiltrating water had no bromide. Removing the water and bromide represented by the runoff and preferential flow al‐ lowed the plots to be modeled as purely matrix flow. The modeled amount of evapotranspiration (ET) was opti‐ mized such that the depth of the modeled and measured bro‐ mide peaks overlapped. Two‐thirds of the root uptake took place in the upper 0.45 m, and the remainder was less than

2003

α (cm-1)

8.1 × 10-5 5.8 × 10-5

0.016 0.016

Table 1. Values used to model bromide transport. n Dispersivity P (-) (cm) (cm d-1) 1.6 1.3

8 16

0.95 m deep. Since the actual Q at any location was not truly known, any error in the estimate in Q caused a similar, but op‐ posite signed, error in the estimate of ET. Bromide uptake and subsequent release by plants was not included. Unsaturated soil hydraulic variables including sat‐ urated hydraulic conductivity (Ks ) and van Genuchten a and n were initially estimated based on the soil textures of the cores. Since the flux beneath the root zone was defined by the upper boundary condition minus the root uptake, further modification of the unsaturated hydraulic variables served only to match the modeled and measured soil moisture pro‐ files. The dispersivities were fit to optimize the width of the bromide peak (table 1) and are reasonable given the 3.65 m modeled depth (Lallemand‐Barres and Peaudecerf, 1978).

RESULTS AND DISCUSSION Precipitation after the 2000 bromide application was be‐ low average, which may have limited the amount of bromide transported from the bermed area as runoff (fig. 4). Heavier precipitation occurred from 2001 to 2003, but it is unlikely that there was still bromide remaining on the surface avail‐ able for transport within runoff. Measured bromide pore‐water concentrations from the soil cores collected in 2003 are presented in figure 5. The ma‐ jority of the bromide concentration profiles have a single peak of bromide at a depth ranging from approximately 1 to 2 m and a maximum concentration ranging from 52 to 314 mg L-1. Although it appears from the profiles that the 3.65 m sampling depth captured the bromide plume in all but a few cores, a subsequent bromide mass balance shows that much of the infiltrated bromide (Bapplied - BQ ) is absent from the profiles. Some of the bromide profiles, particularly those on the eastern portion of the plot (cores 17 to 21), show a large bromide peak near the surface with a low concentration of bromide well in advance of the peak. This may indicate a situ‐

0.337 0.337

Q (cm d-1)

Imatrix (-)

ET (cm d-1)

0.107 0.107

0.41 0.72

0.05 0.15

ation in which the recharge moves preferentially to depth and then enters the matrix‐flow regime. Figure 6 shows the mass of bromide measured in the cores per the mass of bromide infiltrated (Bsoil /Binfil ). Up to 83% of the infiltrated bromide is absent from the profiles. Figure 5 shows that the five cores collected from the eastern portion of the plot (white circles in fig. 6) had large bromide masses located at shallow depths. This indicates that the matrix flow velocity was low and little bromide was lost to preferential flow or runoff relative to the center and western portions of the site. The mean and standard deviation of bromide profiles for the five eastern cores are plotted along with the remainder of the bromide profiles in figure 7. It appears that the two groups of bromide profiles are from distinct populations. A particle size analysis from five soil cores was conducted to discern any differences in soil texture between the populations, but none were noted. The average amount of sand, silt, and clay in the upper 2 m was 23%, 61%, and 16%, respectively. The standard deviation of sand, silt, and clay between cores was 5%, 10%, and 9%, respectively. Two bromide profiles (soil cores 6 and 17) that typified the two populations were selected for modeling the bromide transport by matrix flow only (figs. 8 and 9). The modeled bromide profiles matched the measured bromide profiles well even though preferential flow was not present in the model. The exception was the base of the profile in figure 9, which has a higher measured than modeled bromide con‐ centration. This may indicate movement of bromide from preferential flow paths into the adjacent soil matrix. Because the spatial variability of infiltration and runoff across the plot could not be determined, variables related to the bromide mass balance were calculated as averages from the entire plot. Precipitation averaged 1.23 m year-1 and resulted in 0.39 m year-1 of runoff. Of the 334 kg of bromide applied to the plot, 5.0 kg exited as runoff. A total of 137 kg of bromide was present in the soil profile and was assumed

200

4000 Precipitation Runoff Cumulative precipitation

160 Event -Based P and Q (mm)

Historical average cumulative precipitation

3500 3000 2500

120 Missing data

80

2000

Bromide applied

Cores collected

1500

Cumulative P (mm)

Core 6 Core 17

Ks (cm s-1)

1000 40 500 0 Jan. 2000

0 July 2000

Jan. 2001

July 2001

Jan. 2002

July 2002

Jan. 2003

Figure 4. Event‐based precipitation and runoff. Note that following bromide application, the precipitation rate was below the mean precipitation rate.

2004

TRANSACTIONS OF THE ASABE

Figure 5. Profile of measured bromide concentration in the 21 soil cores collected in the bermed plot. The vertical axis of each profile is soil core depth (0 to 3.75 m), and the horizontal axis is bromide concentration (0 to 325 mg L-1). Sampling locations are approximate. 0 0.5

Depth (m)

1 1.5 2 2.5 3

Western and central cores Five eastern cores

3.5 4

0

100

200 -1 Bromide (mg L )

300

Figure 7. Mean and one standard deviation of the two populations of bro‐ mide profiles. A lognormal distribution was assumed to calculate the stan‐ dard deviation. 0

to be moving largely in the matrix flow domain, since it had not yet exited the control volume. Of the 329 kg of bromide that infiltrated the surface, 192 kg exited the control volume, presumably as preferential flow, since the slow‐moving bromide matrix flow plumes were captured in the upper 3.65 m. Table 2 presents the averages of measured and calculated bro‐ mide mass balance parameters.

Vol. 50(6): 2001-2006

1 Depth (m)

Figure 6. Gray‐scale contours represent the percentage of the bromide re‐ covered within the soil cores (Bsoil /Binfil ). Black and white circles repre‐ sent soil core sampling locations. Black contour lines are elevation above sea level (m).

2 Measured Modeled

3

4

0

20

40

60 80 -1 C (mg L bromide)

100

120

Figure 8. Measured and modeled bromide profile from core 6.

2005

0

Depth (m)

1

2 Measured Modeled

3

4

0

50

100

150

200

250

300

C (mg L -1 bromide)

Figure 9. Measured and modeled bromide profile from core 17. Table 2. Results of bromide mass balances within bermed area from 6 April 2000 to 5 February 2003. Bapplied 334 kg bromide BQ 5.0 kg 192 kg bromide Bpref Bsoil 137 kg bromide 58% Ipref Imatrix 42%

This research demonstrates the potential difficulty of dis‐ cerning the presence and magnitude of preferential flow from analysis of solute concentration profiles. If we had not known the mass of tracer applied, it would have been very difficult to recognize the magnitude of preferential flow that was present.

CONCLUSIONS A potassium bromide tracer was applied to the surface of a 0.37 ha bermed plot. After three years, twenty‐one 3.65 m long continuous soil cores were collected and analyzed for bromide concentration. Most of the bromide profiles ap‐ peared to demonstrate classic matrix flow, with the exception that up to 83% (average of 58%) of the infiltrated bromide mass was absent from the profiles due to preferential flow. Numerical modeling of the bromide transport within the cores was successfully performed without constraint on the modeled mass of bromide applied and without inclusion of preferential flow. This exercise demonstrated that it would be difficult to determine the existence of the preferential flow using tracer concentrations as measured in vertical soil cores without prior knowledge of the applied bromide mass. Since sites are often investigated without knowing how much com‐ pound was released, estimating the released mass of com‐ pound and rate of transport using data gleaned from only soil cores might lead to overly conservative estimates if preferen‐ tial flow is also prevalent and unobserved. ACKNOWLEDGEMENTS The Tennessee Agricultural Experiment Station and the Hobart Ames Foundation provided the financial support for this study. The Ames Plantation is owned and operated by the Hobart Ames Foundation for the benefit of the University of Tennessee.

REFERENCES APHA. 2005. Standard Methods for the Examination of Water and Wastewater. 21st ed. A. D. Eaton, L. S. Clesceri, E. W. Rice, A. E. Greenberg, and M. A. H. Franson, eds. Washington, D.C.: American Public Health Association.

2006

Freeland, R. S., D. J. Inman, R. E. Yoder, and J. T. Ammons. 2002. Technical note: Detecting vertical anomalies within loessial soils using ground‐penetrating radar. Applied Eng. in Agric. 18(2): 264‐264. Hardeman, W. D. 1966. Geologic map of Tennessee (West sheet). Nashville, Tenn.: State of Tennessee, Department of Conservation, Division of Geology. Iragavarapu, T. K., J. L. Posner, and G. D. Bubenzer. 1998. The effect of various crops on bromide leaching to shallow groundwater under natural rainfall conditions. J. Soil Water Cons. 53(2): 146‐151. Jaynes, D. B., S. I. Ahmed, K.‐J. S. Kung, and R. S. Kanwar. 2001. Temporal dynamics of preferential flow to a subsurface drain. SSSA J. 65(5): 1368‐1376. Köhne, J. M., and H. M. Gerke. 2005. Spatial and temporal dynamics of preferential bromide movement towards a tile drain. Vadose Zone J. 4(1): 79‐88. Kung, K.‐J. S. 1990. Influence of plant uptake on the performance of bromide tracer. SSSA J. 54(2): 975‐979. Lallemand‐Barres, P., and P. Peaudecerf. 1978. Recherche des relations entre la valeur de la dispersivite macroscopique d'un milieu aquifere, ses autres caracteristiques et les conditions de mesure, etude bibliographique. Bulletin, Bureau de Recherhes Géologiques et Miniéres 3/4: 277‐287. Larsson, M. H., and N. J. Jarvis. 1999. A dual‐porosity model to quantify macropore flow effects on nitrate leaching. J. Environ. Qual. 28(4): 1298‐1307. Livingston, R. L. 1993. Soil and geomorphic relationships of two water quality landscapes on the Ames Plantation. MS thesis. Knoxville, Tenn.: The University of Tennessee, Department of Plant and Soil Science. Malone, R. W., M. J. Shipitalo, L. W. Douglass, L. B Owens, T. C. Nelson, R. C. Warner, and M. E. Byers. 2000. Assessing herbicide movement using soil samples versus percolate samples. Trans. ASAE 43(2): 343‐348. Malone, R. W., M. J. Shipitalo, and D. W. Meek. 2004. Relationship between herbicide concentrations in percolate, percolate breakthrough time, and number of active macropores. Trans. ASAE 47(5): 1453‐1456. Ray, C., T. Vogel, and J. Dusek. 2004. Modeling depth‐variant and domain‐specific sorption and biodegradation in dual‐permeability media. J. Contam. Hydrol. 70(1‐2): 63‐87. Russell, E. E., and W. S. Parks. 1975. Stratigraphy of the outcropping upper Cretaceous, Paleocene, and lower Eocene in western Tennessee. Bulletin 75. Nashville, Tenn.: State of Tennessee, Department of Conservation, Division of Geology. Schnabel, R. R., W. L. Stout, and J. A. Shaffer. 1995. Uptake of a hydrologic tracer (bromide) by ryegrass from well and poorly‐drained soils. J. Environ. Qual. 24(5): 888‐892. Scorza Júnior, R. P., J. H. Smelt, J. J. T. I. Boesten, R. F. A. Hendriks, and S. E. A. T. M. van der Zee. 2004. Preferential flow of bromide, bentazon, and imidacloprid in a Dutch clay soil. J. Environ. Qual. 33(4): 1473‐1486. Simunek, J., M. Sejna, and M. T. van Genuchten. 1998. HYDRUS 1D software package for simulating the one‐dimensional movement of water, heat, and multiple solutes in variably saturated media. Version 2.02. Riverside, Cal.: USDA‐ARS U.S. Salinity Laboratory. Stone, K. C., P. G. Hunt, M. H. Johnson, and T. A. Matheny. 1998. Nitrate-N distribution and trends in shallow groundwater on an eastern coastal plains watershed. Trans. ASAE 41(1): 59‐64. Yoder, R. E., D. D. Tyler, C. R. Mote, M. E. Essington, J. T. Ammons, T. C. Mueller, and D. C. Yoder. 1998. Vadose and groundwater quality research on tilled and no‐tilled paired watersheds. In Proc. 1998 ASCE Intl. Water Resources Engineering Conf., 1254‐1259. St. Joseph, Mich.: ASAE. Yoder, R. E. 2001. Field‐scale preferential flow at textural discontinuities. In Proc. 2nd Intl. Symposium on Preferential Flow, 65‐68. St. Joseph, Mich.: ASAE.

TRANSACTIONS OF THE ASABE