Quantifying the Impact of Preferential Flow on Field-scale Chemical Transport Under Two Steady-state Conditions in Silt Loam Soil

T. J. Gish, K.-J. S. Kung*, J. Posner, G. Bubenzer, C. S. Helling, E. J. Kladivko, and T. S. Steenhuis, T. J. Gish, Hydrology Lab. and C. S. Helling, Weed Science Lab., USDA-ARS, BARC-W, 10300 Baltimore Blvd., Beltsville, MD 20705-2350.

K.-J. S. Kung, Dep. Soil Science, J. Posner, Agronomy Dep., and G. Bubenzer, Dep. Biological System Engineering, Univ. of Wisconsin-Madison, Madison, WI 53706-1299.

E. J. Kladivko, Dep. Agronomy, Purdue Univ., West Lafayette, IN.

T. S. Steenhuis, Dep. Agric. and Biological Engineering, Cornell University, Ithaca, NY 14850.

Introduction

To ensure high food/fiber production from limited arable lands, fertilizers and pesticides play essential roles in modern agriculture. After extensive use, however, some of these agrichemicals have resulted in nonpoint groundwater contamination. The general public demands not only high-quality, low-cost food and fiber but also clean and pristine environments. Tremendous efforts have been focused on alleviating the undesirable trade-off between production of food/fiber and deterioration of water quality during the past three decades. Each year, federal (USDA/USGS/EPA) and state (DNR) agencies, private industries, and research institutes together spend hundreds of millions of dollars to monitor the potential adverse effects of agrichemicals on environments. Generally, there is substantial scatter and variability in the results - making scientific interpretation and policy development difficult. Groundwater and surface water contam­ination represents one of our nation's most significant, long-term problems.

Results from many lab experiments based on chemical transport in homogenized soil matrix demonstrated that soil should be an excellent filtering system. This is why many highly toxic chemicals have been approved by regulatory agencies for agricultural use. However, research results from field experiments indicate that chemicals can bypass the vast soil matrix and be transported to groundwater expeditiously (Helling and Gish, 1991; Flury, 1996). There have been many papers published during the last two decades to demonstrate that preferential flow was the major reason for non-point groundwater contamination by pesticides. Nevertheless, the scope and magnitude of preferential flow as well as its impact on and relevance to field-scale contaminant transport have not been quantified and resolved. This was mainly because the conventional sampling protocols approved by the US-EPA were not effective.

Both pesticides and the conventional sampling protocols were developed more than 50 years ago when preferential flow and its impact on contaminant transport were not understood at all. The conventional sampling protocols implicitly assume that contaminants move through the entire soil matrix. Therefore, samples are collected at random locations and results are averaged to determine the patterns of overall deep leaching. Because the hydraulically active preferential flow pathways generally occupied a very small fraction of the entire soil matrix, most randomly collected samples would not detect the transport through preferential pathways. A sample that happens to be near a preferential pathway will recover much more than those located away from a pathway (Ju et al., 1997). This partly explains the substantial scatter and variability in sampling results - making scientific interpretation and policy development difficult.

To accurately determine the mass flux of pesticide transport in an unsaturated soil, one must identify and sample the preferential pathways (Ju et al., 1997). However, there are no tools that can nondestructively and 3-D map and visualize these complex preferential flow paths. In a tile-drained field, regardless of how insidious/complex the preferential flow paths may be, some contaminants that reach shallow groundwater will drain out in the tiles as long as the water table is above the tile line. Sampling tile drain was the only alternative method to directly measure total mass flux of chemical leaching under large field-scale conditions. Nevertheless, Flury (1996) pointed out that using tile drain as a monitoring protocol has its intrinsic drawback. Unless tiles were closely spaced, some chemicals reaching the shallow watertable would inevitably be transported either vertically into regional groundwater or laterally (parallel to the tile) away from the field, when the chemicals were applied to the entire field. Under both conditions, some chemicals reaching a shallow groundwater table would not drain out from tiles. This suggested that monitoring tile drain would underestimate the total mass flux. Furthermore, chemical breakthrough time and pattern collected from tile drain would reflect the convoluted chemical transport through both unsaturated and saturated zones. In order to use results from tile drain to predict contaminant transport in an unsaturated soil profile, one must deconvolute the chemical breakthrough time and pattern to extract information on the transport through the unsaturated zone. This has been an extremely difficult task complicated by the fact that a shallow groundwater table would fluctuate during a precipitation event.

Using multiple piezometers to monitor the fluctuation of shallow watertable during precipitation events in the Till Hydrology Farm of the Iowa State University, it was found that the watertable close to the tile line barely changed and always had the steepest gradient (Dr. Dan Jaynes, personal communication, 1997). This suggested that chemical transport occurred mostly in the unsaturated zone if a chemical was applied only to a narrow strip at soil surface close to the tile line. In other experiments independently conducted in the Till Hydrology Farm of the Iowa State University by Dr. Dan Jaynes and the Willsboro Farm of the Cornell University by Dr. Tammo Steenhuis, respectively, results showed that, after soil profile was healed, contaminant transport through preferential flow paths was not caused by the installation of the tiles.

Results from field experiments by Kung et al. (2000) demonstrated that applying chemicals at a narrow strip near the tile could quantify the impact of preferential flow paths on contaminant transport. Most importantly, they found that preferential flow paths of natural soils have a wide pore spectrum. Preferential flow paths with larger pores would become hydraulically active only when the profile becomes wetter during a precipitation event. This suggested that not all preferential flow paths are simultaneously active and the impact of preferential flow paths on contaminant transport is dictated by rain intensity and duration. Because the volumetric flow in a pore is proportional to the fourth power of the pore radius, in order to predict and prevent contaminant transport through preferential flow paths, it is important to comprehend when a preferential flow would become hydraulically active. The soil pore size spectrum is one of the most fundamental soil properties.

The small and large pores are formed very differently. The small matrix pores are always interconnected and generally self-similar. The spectrum of the small pores is a static property of a soil profile and measurement made at several locations can accurately represent the spectrum of the primary pores of an entire soil horizon with the same soil texture. The larger structural pores is much more complex and have very dynamic formation and destruction cycles. As a result, the size spectrum of large pores measured at several random locations can not be extrapolated to represent the spectrum of the entire soil profile. Nor can a measurement be extrapolated temporarily. For this reason, it is difficult to directly measure the field-scale size spectrum of the large structural pores, which constitute the preferential flow paths in finer soils.

No methodology has been ever developed to holistically measure the field-scale size spectrum of large secondary soil structural pores. The hydraulic conductivity has been traditionally used to describe the lumped effect of all the soil pores on water movement. From the 1960's, soil physicists have spent tremendous effort to directly measure soil hydraulic conductivity and use statistical approach to estimate the 3-D nature of soil hydraulic conductivity. But, results show that, even the field-scale 3-D soil hydraulic conductivity is too complex to be measured by conventional instruments, which simply could not encompass the spatial and temporal scales of the secondary soil structural pores. The lack of methodology to non-destructively quantify the size spectrum of the large secondary soil structural pores has frustrated all efforts to correctly predict contaminant transport, let alone to prevent groundwater contamination by agrichemicals. As a result, agrichemicals have been increasingly criticized by environmentalists; many people have gradually become reluctant to acknowledge the essentialness of agrichemicals.

When a process was too complex to identify/measure all involved parameters, indirect method was often used to find relationship between input and outcome. For example, Transfer Function approach was developed in hydrology for watershed-scale storms hydrograph to predict flooding and later adopted to estimate contaminant transport in unsaturated soils (Jury, 1982). Furthermore, inverse algorithms have been developed and commonly used in many areas such as complex signal processing and artificial intelligence.

The objective of the study was to collect accurate chemical breakthrough patterns so that it would be possible to use these patterns to deconvolute and indirectly quantify the field-scale pore size spectrum of preferential flow paths under different rain intensities. Under transient infiltration condition like those of Kung et al (2000), soil matrix potential gradients could cause lateral water movement and solute transport among inter-connected pores. This made it difficult to use indirect method to estimate the soil pore spectrum by using accurately measured tracer breakthrough patterns. In order to measure soil pore spectrum relevant to field-scale contaminant transport, the complexity of lateral dispersion associated with transient flow must be first eliminated from the experimental procedure. In other words, tracer breakthrough pattern must be collected under steady state conditions under different infiltration rates.

Materials and Methods

Field experiments were conducted in summer of 1999 at the Walworth County Farm in Elkhorn, Wisconsin. The site is located within the Southern Wisconsin and Northern Illinois Drift Plain. The Lake Michigan and Green Bay glaciers retreated from this area about 10,000 years ago and tall grass prairie began to dominate from five to six thousand years ago (Curtis, 1959) until the arrival of European farmers. Then, the area has gone through an evolution of primarily continuous small grains, to dairy, to cash grain corn/soybean systems. Generally there is a silt loam surface horizon (0-35 cm, organic matter 3%), followed by a clay-loam or sandy clay-loam B-horizon (35-65 cm), and underlain by glacier well-mixed gravelly till. The average depth to compacted glacial till is between 80 and 130 cm. The soils in the site are classified as Pella silt loam soils. These prairie-derived silt loam soils covers the southeast of the Wisconsin into the northern Illinois and represent one of the most productive agricultural lands in this region.

A 160 m by 72 m field plot was selected. The field has around 1% to 3% slope and has been under no-till corn-soybean-wheat rotation for 12 years. Tiles with uniform 18-m spacing at around 1 to 1.2 m depth were installed in the field in early 1970's. Manhole at the down gradient of the field was installed in 1997. Flow rate of the tile drain was continuously recorded by using a submerged pressure transducer (with resolution of 75 mV per 1 cm water head) to measure water height in a flume with a 15o V-shaped, sharp-edged notch.

The soybean was just emerged when the experiments started. A steady 4 mm h-1 irrigation is applied to two 16 m by 30 m areas near the three central tile lines through two rows of solid-state sprinklers (Fig. 1). Each row of sprinklers had 3 nozzles. Between the two rows of solid-state sprinklers, there was a 3.5 m by 24 m shed in the center. The shed was 1.5 m high and made of aluminum frame and corrugated PVC wall/roof, which had about 90% light transparency. The long side of the shed was parallel to the tile line offset 0.3 m. There were 8 carefully calibrated nozzles inside the shed. These nozzles were oscillating inside the shed, similar to the design of Ghodrati et al (1990). However, the nozzles were 2.4 m apart and mounted on a trolley and oscillating along the long side of the shed. This design offered around 90% to 95% Christianson uniformity inside the shed under all climatic conditions, while that of the outside nozzles varied from 20% to 85% depending primarily on the wind conditions. The irrigation rate inside the shed was 3.1 mm h-1. The shed was 15 m from the manhole. Water collected from the roof of the shed during natural precipitation events was diverted away from the field by gutters.

The tile flow was continuously monitored after irrigation started. After the tile reaches steady-state outflow, a pulse of conservative tracer, potassium bromide (2.35 kg Br), was applied from the nozzles inside the shed. During the tracer application, the irrigation rates at both inside and outside were maintained constant. No tracers were applied to outside. Water samples from tile drain were manually collected once every 2 minutes during the first 2 hours after tracer application. Then, water samples were collected once every 6 minutes during the next 10 hours. Water samples were collected once every 15 minutes for the next 12 hours and then decreased to every 30 minutes for the next 12 hours. Then, water samples were collected once every hour for the next 3 days and once every two hours for the next 10 days. Flow rate of the tile drain was continuously monitored.

The irrigation was terminated at 22 days after tracer application. The entire field was under free drain for a week. Then, four 10-cm diameter soil cores were collected inside the shed and soil samples from each core were collected every 7.5 cm to 75 cm depth. At this time, the soybean reached about 25 cm height. Mower was used to cut the soybean to about 10 cm height on a 5 m by 40 m strip along to the other side of the tile line beside the shed. The entire irrigation system was then moved laterally to the other side of the tile line such that the shed was again 0.3 m offset from the tile.

The experiment was repeated, except the irrigation intensity inside the shed was reduced to 0.89 mm h-1, while that of the outside was maintained at 4 mm h-1. After the tile drain reached steady state, a second pulse of conservative tracer, penta-fluorobenzoic acid (PFBA 1.02 kg) was applied from the nozzles inside the shed. Again, water samples were collected with identical frequency after tracer application. The experiment was terminated at 14 days after tracer application.

This experiments were design to collect the field-scale breakthrough patterns of conservative tracers so that the impact of preferential flow on chemical transport in an unsaturated soil profile could be quantified and compared under two different steady state infiltration rates. Tracers recovered from tile drain had traveled through unsaturated soil profile and saturated zone. Kung et al (2000) showed that by applied tracer only to a narrow strip close to tile line would minimize the tracer travel time in the saturated zone. Nevertheless, as irrigation rate decreased during the second tracer experiment, the overall shape of water table between two adjacent tiles would drop. This could influence the tracer travel time in the saturated zone and make the comparison of breakthrough patterns of two tracer less meaningful. We minimized this error by maintaining the same outside irrigation intensity at 4 mm h-1 to assure that the overall shape of water table between two adjacent tiles was approximately the same during the two tracer experiments. On the other hands, the two tracers were applied to two sides of a tile line and traveled through different soil profiles. The main reason that we moved the shed was because, about a week after tracer application, a thin layer of algae started to grow on soil surface inside the shed. Nevertheless, because the tile was installed almost 30 years ago, it was plausible to assume that the soil on two sides of a tile line had identical physical properties. There was no surface ponding or runoff inside the shed through both experiments.

Results and Discussion

Figure 2 shows that tile flow as a function of time after the first tracer application. The flow rate was extremely steady around 200 mL s -1 during the first 30 hours after tracer application. Although the irrigation intensity was maintained constant, the tile flow did have diurnal fluctuation because of the evaporation and transpiration. At the 35 h and 60 h after tracer application, there were heavy and short thunderstorms, which cause sharp increase of tile drain. After the second precipitation, the irrigation outside the shed was shut off because water started to pond on the soil surface outside the irrigation shed and surface runoff occurred. Nevertheless, because metal flashing was installed around the shed, there was no runoff into the shed.

After the precipitation, because it was difficult to decide when to turn the outside irrigation back on, the tile flow dropped to 130 mL s -1. The irrigation outside the shed was again shut off at around 4.5 days after tracer application because of a forecasted local thunderstorm with high intensity. However, the storm never occurred. This event caused tile flow to drop and deviate from steady state flow. The generator failed at 7.5 days. This again caused tile flow to drop and deviate from steady state flow. After 10 days, the outside irrigation was accidentally shut down twice.

Although there is fluctuation in tile drain, Fig. 2 shows that bromide had distinct breakthrough pattern. Initial arrival was detected in tile drain only 16 minutes after its application. The saturated hydraulic conductivity of homogenized soil samples taken from the soil profile ranged from 3 to 20 mm h-1. The irrigation inside the shed was very uniform and its intensity was less than the saturated conductivity of the soil. The quick initial breakthrough showed that bromide was leached down through preferential flow paths. In another tracer experiments, Kung et al. (2000, 2000a) applied a benzoic acid six hours after a steady-state irrigation with 3 mm h-1 intensity in the Clermont silt loam of the South East Purdue Agricultural Center of Purdue University in Butlerville, Indiana and the Rhinebeck variant loam of the Willsboro Research Farm of the Cornell University in Willsboro, NY, where tile drain was used as monitoring protocol. The results showed that the breakthrough of benzoic acid tracer in the tile occurred in 18 and 12 minutes after tracer application, respectively. Their sampling interval was every six minutes. It was rather surprising to find that the bromide breakthrough time in the Pella silt loam in southeastern Wisconsin was very similar to those in Indiana and NY. The similar initial breakthrough time in three different sites indicated that the three sites have similar preferential flow pathways. This confirmed the hypothesis proposed by Kung et al. (2000a) that preferential flow pathways were not dictated by the soil textures and soil origins.

The bromide mass flux shown in Fig. 2 quickly increases two orders of magnitude within 6 hours after tracer application. The mass flux maintained essentially constant from 6 to 60 hours after tracer application. Afterward, bromide mass flux started to tail off gradually until 10 d after tracer application. The total bromide mass recovered from the tile was 85.1% of the total mass applied. The quick peaking and the high total mass recovered from tile again showed that most of the bromide was leached down through preferential flow paths.

In the second tracer experiment, the flow rate of tile drain fluctuated around 140 mL s -1 and again had slight diurnal because of the evaporation and transpiration (Fig. 3). There were 4 major natural precipitation events occurred from 90 to 300 h after tracer application and each event caused sharp increase of the tile flow hydrograph. This was beyond our control. Because the shutting off and turning on of the outside irrigation was better controlled during the second tracer experiment, the tile flow never dipped below the steady state rate of 140 mL s -1.

Figure 3 shows that there is very low amount PFBA detected sporadically in tile drain from10 minutes to 3 hours after application. Because the background PFBA concentration was essentially zero and the concentrations from samples collected during this period were well above the detection limit of 10 ug L-1, we considered the breakthrough during this period was real. This early breakthrough suggested that preferential flow paths also participated in the transport of PFBA under a lower steady state infiltration rate of 0.89 mm h-1. Theoretically, because the irrigation rate during the first tracer experiment was about 3 times higher than the second experiment, the bromide should have an earlier breakthrough time than PFBA. However, the PFBA arrived at only 10 minutes after application (vs. 16 minutes for bromide). This was mainly because the PFBA had a much lower detection limit of 10 mg/L as compared to 0.5 mg/L for bromide.

From 7 to 90 hours after tracer application, there was steady detection of very low amount PFBA. Although the PFBA breakthrough started earlier, the total PFBA mass recovered during the first 90 hours was only 0.16% of the total PFBA mass applied (vs. 66.6% for bromide during the same 90-hour period). This suggested that, although early breakthrough of PFBA indicated that the preferential flow paths participated under 0.89 mm h-1 infiltration rate, its impact on the total mass loss of a contaminant was negligible. This confirmed observation by Kung et al. (2000) that, when a soil profile became wetter during a prolonged precipitation event, more and more preferential flow pathways became hydraulically active.

It is around 90 hours after PFBA application when a main PFBA breakthrough starts to emerge (shown in Fig. 3). This main breakthrough curve peaked at around 240 h after tracer application. This slow peak was not observed in the Br breakthrough curve. This suggested that most of the PFBA was transported through soil matrix pores and the preferential flow pathways used to transport bromide during the first irrigation event were not hydraulically active during the second irrigation.

The total PFBA mass recovered from the tile during the entire sampling period was 16.8% of the total mass applied. Note that, however, when the irrigation was terminated, the breakthrough of PFBA is still gradually tailing off as shown in Fig. 3. As a result, the total PFBA mass leached out could not be accurately calculated. About twice amount of bromide mass was applied as compared to that of the PFBA. However, the mass flux of the PFBA at peak was about 16 times less than that of the bromide. The lower mass flux at peak and much broader breakthrough pattern together suggested that most of the PFBA had been transported through the soil matrix pores.

To confirm this assessment, the 1-D steady-state analytical solution of convection-dispersion equation by Jury et al (1991) was used to fit the PFBA mass breakthrough pattern. There are three parameters in the analytical solution, i.e., velocity V, apparent dispersion D, and the soil profile depth L. The V was derived from irrigation rate (0.89 mm h-1) divided by the effective soil water content q, which was 32% (measured from 4 soil core samples taken from the top 30 cm at the end of the experiment). The depth to tile drain was 0.95 m. The D was not measured during the experiment and was, therefore, adjusted to achieve the best fit.

The fitted analytical result is shown as a solid line in Fig. 3. It was found that when D was 5 x 10 -8 m2 s-1 would achieve the best fit. This value was close to many lab measurements of D based on homogenized silt loam soils. The perfect fit between the measured breakthrough pattern and the analytical solution suggested that, under a steady state rate of 0.89 mm h-1, vast majority of the tracer was transported in a field soil through the small matrix pores, which had little difference between homogenized lab condition and undisturbed field condition. Therefore, if the second tracer were reactive instead of conservative, it would have been most likely adsorbed and degraded and not reached the water table.

Summary and Conclusions

Tile drain monitoring facility was used to examine chemical leaching patterns under two steady state infiltration conditions. Results from the first tracer breakthrough pattern showed that chemical arrival occurred within 16 minutes after tracer application and the peaking of breakthrough was very sharp. This demonstrated that preferential flow paths played significant role on contaminant transport in a field soil profile. This breakthrough pattern was ideal to be used in indirect method to assess the soil hydraulic properties.

Results from the second tracer experiment showed that the main breakthrough did not occur until 90 hours after tracer application and the breakthrough reached peak slowly. This main breakthrough pattern coincided with an analytical solution based on 1-D convective-dispersive transport. This suggested that the spectrum of matrix pores under undisturbed field condition was similar to that of the homogenized soil. There was sporadic fast breakthrough. However, the total mass leached out before 90 hours was negligible. This indicated that the preferential flow paths were not hydraulically active and had little impact on the contaminant transport when the intensity of a precipitation event was below a certain value.

Literature Review

Everts, C.J., R.S. Kanwar, E.C. Alexander Jr., and S.C. Alexander. 1989. Comparison of tracer mobilities under laboratory and field conditions. J. Environ. Qual. 18:491-498.

Flury, M. 1996. Experimental evidence of transport of pesticides through field soils - A review. J. Environ. Qual. 25:25-45.

Gachter, R, J.M. Ngatiah, and C. Stamm. 1998. Transport of phosphate from soil to surface waters by preferential flow. Environ. Sci. & Tech. 32(13):1865-1869.

Gaynor, J.D., and W.I. Findlay, 1995. Soil and Phosphorus loss from conservation and conventional tillage in corn production. Journal of Environmental Quality, 24:734-741.

Ghodrati, M., F.F. Ernst, and W.A. Jury. 1990. An automated spray system for application of solutes to small field-plots. Soil Sci. Soc. Am. J. 54:287-290.

Helling, C.S. 1992. Pesticides, agriculture, and water quality. In Proc. 2nd Stockholm Water Symp, August 1992, Stockholm, Sweden.

Helling, C.S., and T.J. Gish. 1991. Physical and chemical processes affecting preferential flow. p. 77-86. In T.J. Gish and A. Shirmohammadi (ed.) Preferential flow. Proc. Natl. Symp., 16-17 Dec. 1991, Chicago, IL. Am. Soc. Agric. Engr., St. Joseph, MI.

Jaynes, D.B. 1994. Evaluation of fluorobenzoate tracers in surface soils. Groundwater. 32:532-538.

Ju, S-H., and K-J.S. Kung. 1997. Impact of Funnel Flow on Contaminant Transport in Sandy Soils Numerical Simulation. Soil Sci. Soc. Am. J. 61:416-427.

Ju, S-H., K-J.S. Kung, and C. Helling. 1997. Simulating impact of funnel flow on contaminant sampling in sandy soils. Soil Sci. Soc. Am. J. 61:409-415

Ju, S-H., and K-J.S. Kung. 1997. Impact of Funnel Flow on Contaminant Transport in Sandy Soils: Numerical Simulation. Soil Sci. Soc. Am. J. 61:416-427.

Jury, W.A. 1982. Simulation of solute transport using a transfer function model. Water Resour. Res. 18:363-368.

Kanwar, R.S., J.L. Baker, and D.G. Baker. 1988. Tillage and split N-fertilization effects on subsurface drainage water quality and corn yield. Trans. ASAE. 31(2):453-460.

Kladivko, E.J., G.E. Van Scoyoc, E.J. Monke, K.M. Oates, and W. Pask. 1991. Pesticide and nutrient movement into subsurface tile drains on a silt loam soil in Indiana. J. Environ. Qual. 20:264-270.

Kladivko, E.J., J. Grochulska, R.F. Turco, G.E. VanScoyoc, and J.D. Eigel. 1999. Pesticide and nitrate transport into subsurface tile drains of different spacings.

J. Environ. Qual. (In Press) Kung, K-J.S. 1990. Preferential flow in a sandy vadose zone: 1. Field observation & 2. Mechanism and implications.. Geoderma. 46:51-71.

Kung, K-J.S., T.S. Steenhuis, T. Gish, E. Kladivko, G. Bubenzer, and C.S. helling. 2000. Impact of preferential flow on the transport of adsorptive and non-adsorptive tracers. Soil Sci. Soc. Am. J. (In press)

Kung, K_J.S., T.S. Steenhuis, T. Gish, E. Kladivko, L. Gehring, G. Bubenzer, and C.S. Helling. 1998. Quantify the water dynamics of preferential flow: I. Variant clayey loam at Cornell site. Soil Sci. Soc. Am. J. (In press)

Logan, T.J., G.W. Randall, and D.R. Timmons. 1980. Nutrient content of tile drainage from cropland in the north central region. North Central Regional Publ. no. 268. OARDC Res. Bull. no. 1119. OARDC, Wooster, OH

Muir, D.C., and B.E. Baker. 1976. Detection of triazine herbicides and their degradation products in tile-drain water from fields under intensive corn (maize) production. J. Agric. Food. Chem. 24:122-125.

Richards, T.L., and T.S. Steenhuis. 1988. Tile drain sampling of preferential flow on a field scale. In P.F. Germann (ed.) Rapid and far reaching hydrological processes in the vadose zone. J. Contam. Hydrol. 3:307-325.

Roth, K., W.A. Jury, H. Fluhler, and W. Attinger. 1991. Transport of chloride through an unsaturated fields soil. Water Resour. Res. 27:2533-2541.

van Genuchten, M. Th., and W. J. Alves. 1982. Analytical solutions of the one-dimensional convective-dispersive solute transport equation. Techn. Bull. 1661, ARS, USDA, 151 p.

Zucker, L.A. and L.C. Brown. 1998. Agricultural Drainage: water quality impacts and subsurface drainage studies in the midwest. MSEA Bulletin 871. Ohio State University Extension.

*Email: kung@calshp.cals.wisc.edu

Home | Who we are | Research | Publications | Links | Site map | Contact us

University of Wisconsin-Madison
College of Agricultural and Life Sciences
If you have any difficulty accessing this page, please contact our web master.

Site hosted by UW-Madison Center for Integrated Agricultural Systems