| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
a Federal Office for Water and Geology, Hydrological Survey, 3003 Bern, Switzerland
b Swiss Federal Institute for Environmental Science and Technology (EAWAG), 8600 Dübendorf, Switzerland
c Partners GmbH, Blaufahnenstrasse 14, CH-8093 Zurich, Switzerland
d Swiss Federal Institute of Technology, Department of Soil Protection, Grabenstrasse 3, 8952 Schlieren, Switzerland
* Corresponding author (andreas.kohler{at}bwg.admin.ch)
Received 23 April 2002.
ABSTRACT
Leaching of fertilizer and pesticide compounds from agricultural lands into ground and surface waters is a major environmental problem. To study this phenomenon, a bromide tracer experiment was performed on a 1.6-ha area of a tile-drained arable field to determine the contribution of preferential flow to solute leaching, and to investigate the factors triggering the onset of such events. During the 2 yr following the application of Br-, concentration of the drainage effluent exhibited a pattern of peaks coinciding with the discharge peaks. The concentration peaks, on top of a smooth base flow, were interpreted as being the contribution of preferential flow. The separation of peak and base flow indicated that 73% of the Br- leached during the 2 yr of study was exported through preferential flow. A simple leaching bucket model was able to accurately predict the occurrence of solute peaks in the drain discharge. The analyses indicated that preferential flow originated primarily at the boundary between top- and subsoil as soon as the topsoil became sufficiently saturated. Thus, low intensity, high duration events could also trigger preferential flow.
Abbreviations: EAPV, equilibrium above-drain pore-water volume
EXPORT OF FERTILIZERS and pesticides from agricultural lands into ground and surface waters is a major environmental problem. Transport of these components by preferential flow processes through macrostructures bypassing the soil matrix has been recognized as an important flow mechanism. Recent studies show that the occurrence of preferential flow due to heavy rainfall events is the rule rather than the exception in many soils (Flury and Flühler, 1994; Jaynes et al., 2001; Lennartz et al., 1999; Stamm et al., 1998). Evidence of preferential solute transport in field soils has been provided by solute profiles or dye patterns after infiltration (Dekker and Ritsema, 1996; Forrer et al., 2000) and small plot experiments where the seepage water is intercepted by a drain line (Bronswijk et al., 1995; Magesan et al., 1995). Subsurface drainage systems may themselves aggravate the problem by short-circuiting the flow pathways from the subsoil to surface waters.
Preferential flow means that transport is channeled through pathways with high velocity and limited exchange with slower pathways. The hydraulic interaction between the two domains is far from local equilibrium (Flühler et al., 1996; Forrer et al., 1999). Such channeling may be due to macropores (Beven and Germann, 1982), nonuniformity of water repellency (Ritsema and Dekker, 1995), or hydraulic conditions causing instability of infiltration fronts (Glass et al., 1989; Kung, 1990). Factors influencing the triggering of preferential flow are rain intensity, amount of rainfall, soil moisture antecedent to the rainfall event, roughness of the soil surface, texture of surface soil, and water repellency (Bouma, 1990).
The interaction between preferential flow and subsoil drains is rather complicated and has not been thoroughly studied. While in some cases the flow through preferential paths may directly feed a drain, in other cases preferential paths may have no bearing on the exfiltration processes into the drains. For the later cases, most of the water captured by a drain may have already moved some distance through the saturated zone. Only a small portion of the water delivered by the preferential paths, if any, may directly enter a drain after seeping through the unsaturated zone (Jury, 1975). Mixing during the travel distance through the saturated zone may effectively remove any concentration pattern of preferential transport in the drainage discharge.
In this study, the role of preferential transport in the export of solutes via the tile drain system was investigated. The area studied was in a former wetland now used for intensive agriculture in the vicinity of Zurich, Switzerland. For this purpose a Br- tracer solution was sprayed onto a 1.6-ha field, and the tracer concentration and the flow rate of the drainage discharge during the following 2 yr were monitored. The goals of the study were (i) to assess the occurrence and contribution of preferential transport in solute export with the drainage water and (ii) to determine the conditions of initial soil moisture, rainfall intensity, and cumulative amount of rainfall in a storm event required to trigger preferential transport into the drains.
Although complex programs with macropore components such as MACRO (Jarvis, 1994) are available for such analysis, a common problem of these models is over-parameterization and the difficulty in finding the true parameters. Modeling at different levels of complexities is thus important for a better understating of the physical processes at work. For this reason we opted in this study to use two simple bucket models for the analysis.
MATERIAL AND METHODS
Experimental Site
The study area depicted in Fig. 1 is situated in the valley of the Furtbach creek, north of Zurich, in a former wetland. The area was drained in 1920 and reclaimed for agriculture. The field drains consisted of clay tubes lying at a depth of 1 m below soil surface with a spacing of 20 m. The drainage water was discharged through collectors placed at a depth of 1.5 m with a spacing of about 90 m. In 1980, major rehabilitation works were performed, including the flushing of the old field drains and the construction of new collectors (polyethylene pipes) with a gravel filter package at a depth of 1.5 m below soil surface.
|
The main soil type in the test area is classified as Mollic Gleysol. Table 1 lists some pertinent information about this soil, which is rich in carbonate, organic matter, and silt. The average thickness of the humus-rich topsoil was 0.25 m with a spatial variability ranging from 0.20 to 0.40 m. Similar to the topsoil, the silty subsoil (0.25–0.40 m) also had a high porosity and a low bulk density. Hence, the silty subsoil and the topsoil layers were treated as one layer in the model. At the 0.40-m depth there was a sharp distinction in the bulk density and the soil porosity.
|
|
The tracer was applied on 23 Aug. 1995, on a dry and wind free day. The groundwater table was at a depth of 1.23 m below surface, and drainage discharge from the monitored collector was small (0.12 L s-1) and decreasing. Water tensions were 50.9 kPa at 5 cm below soil surface, 22.2 kPa at 15 cm, 22.2 kPa at 30 cm, and 3.8 kPa at 60 cm depth.
The homogeneity of the tracer application was checked using six water absorbent sheets, placed on the soil surface. The first two sheets were placed in the first strip of the application, while the other four sheets were placed each in the following strips. Each sheet had a surface area of 0.125 m2. Immediately after the application the sheets were put into a plastic bag and weighed. The weight increase of the sheets after spraying had a coefficient of variation of 14%.
Sampling and Chemical Analysis of Drainage Water
Drainage outflow was sampled at the observation hole (Fig. 1) using an ISCO sampler (Model 2900, Isco, Inc., Lincoln, NE) with 24 bottles, each with a sample volume of 500 mL. The ISCO sampler was connected to a datalogger, which automatically started sampling as soon as the discharge of the drainage reached a threshold value of 0.22 L s-1. After the onset of sampling during a drainage event, one subsample was collected every 15 min with two subsamples being combined into one sample. This sampling scheme was maintained as long as the discharge kept rising. When the discharge started to decrease, the sampling interval was increased to 120 min with four subsamples of 30 min each followed by a 240-min interval with eight subsamples of 30 min each.
Of each sample, 10 mL was transferred into small polyvinyl chloride bottles and frozen for storage. After thawing, the sample was filtered through a 0.25 µm cellulose acetate filter and analyzed for Br- using a Dionex anion chromatograph (Dionex Corp., Sunnyvale, CA). Analysis of 20 replicate samples gave a coefficient of variation of 10%.
Separation of Preferential and Matrix Transport
According to Everts and Kanwar (1990), the drain discharge curve can be regarded as a combination of matrix flow, occurring continuously, and preferential flow occurring only during discharge events. The total discharge is given by
 | [1] |
 | [2] |
 | [3] |
 | [4] |
Only qt and Ct could be directly measured. Cm was determined, according to Everts and Kanwar (1990), by assuming that base-flow concentration during each event can be determined by linear interpolation of the discharge concentrations at the beginning and at the end of the event.
For the estimation of Cp two approaches were used. In the first approach qm was assumed to be negligible at the peak of a discharge event and thus the concentration of the preferential flow, Cp, was equal to the peak concentration of the drain outflow. This represents an upper boundary estimate of the preferential flow as it implies qp = qt at the moment of maximum flow. The second approach, following Everts and Kanwar (1990), assumed that preferential flow concentration was equal to the concentration of the infiltration solution and that the applied mass M of tracer was gradually dissolved into the infiltrated water (cumulative net infiltration). In this case, the concentration of the preferential flux Cp (mg L-1) decreases with every additional rainfall according to the following expression:
 | [5] |
 | [6] |
Bucket Model
Bouma (1990) reviewed various mechanisms for triggering preferential flow. Apart from soil factors such as antecedent moisture content, surface roughness, water repellency of the soil matrix, and soil layers with differing soil hydraulic properties, meteorological factors such as peak intensity and cumulative amount of rainfall were also considered very important. Assuming soil factors other than initial moisture content to be constant, two simple bucket models were used to analyze the triggering of preferential flow.
The first model called surface trigger is depicted in Fig. 3. This model assumes that the preferential flow is generated at the soil surface. The necessary condition is that rainfall intensity exceeds the matrix infiltration capacity of the topsoil. Preferential flow then arises as soon as the capacity for surface storage is exceeded, and the surplus is channeled into macrovoids. For simplicity, a constant rate of matrix infiltration was assumed ignoring initial water sorption due to matrix potential gradients at a dry soil surface. The steady-state infiltration capacity of the topsoil was determined to be 18.7 mm h-1 by dye tracer infiltration experiments conducted by Gada and Felber (1996). The surface storage capacity was estimated to range between 0 and 10 mm.
|
The dye infiltration test demonstrated that in the two soil layers down to 0.4 m, matrix flow occurred and that these two topsoil layers acted as an infiltration storage layer, while in the denser subsoil, transport occurred only in preferential flow paths such as root channels and worm holes. Therefore, as stated above, the soil layers down to the 0.4-m depth were lumped into one functional layer referred to as the topsoil layer, and the soil layers beneath were taken as the subsoil layer. The soil hydraulic parameters of the two functional soil layers in the subsoil-trigger model were determined by fitting the calculated water contents of the topsoil and the subsoil on the measured water contents at the 0.3- and 0.6-m depths (Table 2). Figure 4 shows a comparison of the calculated and the measured water contents. The mean of the differences between the measured and the calculated water contents was 0.03 for the topsoil layer and 0.02 for the subsoil layer.
|
|
Dependence of Tracer Leaching on Discharge Events
Figures 5, 6, and 7 show variations of the Br- concentration in the drainage effluent for different time periods. After the application of Br- in August 1995, the first discharge event in September of 1995 (Fig. 5) showed Br- concentration increasing to a peak and dropping again as the flow rate decreased toward the base-flow conditions. With increasing peak concentrations, the changes in drainage flow rate and tracer concentration paralleled each other more closely, until peak concentration decreased below 3 mg Br L-1 in late 1996 as shown in Fig. 7. The relationship between discharge and concentration changed with time not only for peak concentrations, but also for base-flow concentration. From the first discharge event in September 1995 until the last observed events in July 1996, the base-flow concentration continuously increased. From July 1996 until February 1997, base-flow concentration decreased while at the same time the relationship between concentration and discharge deteriorated (Fig. 7). In the last period from February 1997 to August 1997, there again was a higher base-flow concentration and an improvement in the relationship between concentration and discharge.
|
|
|
Since base flow is a result of slow matrix flow, the base-flow concentration is therefore expected to have a smooth breakthrough curve. This, however, was not seen in the measured base-flow concentration reported in the last column of Table 3, as there was a distinct drop of the concentration after 419 mm of net infiltration. This drop in the concentration was due to solute uptake by the crops. Approximately 51% of the applied mass of the tracer was removed by the growing sugar beet. In the subsequent decomposition of the plant residues after the harvest in the winter of 1996, almost 50% of the initially applied tracer was again returned to the soil surface. This explains why the concentration increased in February of 1997. In contrast to the sugar beet, wheat uptake in 1977 was small, accounting for only 7% of the applied mass of Br-. Nevertheless, an increase in the concentration of Br- could still be seen in the drainage water as a result of wheat residue decomposition at the soil surface and transport through preferential paths.
|
Table 4 provides the data describing the profile of measured water contents, soil water potential, and water storage in the soil on this day. In total, the water storage above the level of the field drains between 0 and 1 m depth was 480 mm. This figure is defined as equilibrium above-drain pore-water volume (EAPV). Table 3 shows that the highest peak-flow concentration of 16 mg Br L-1 was observed after a cumulative net infiltration of 210 mm, or 44% of EAPV, had occurred. The second maximum peak-flow concentration of 8 mg Br L-1 occurred at 78% of EAPV, coinciding with the base-flow maximum tracer concentration of 2.4 mg Br L-1. Contrary to the base-flow concentration, the variation of the peak-flow concentration showed considerable fluctuations between subsequent events depending on the peak-flow rate.
|
|
|
|
During the first event on 13 Sept. 1995, where the ratio of net infiltration to EAPV was equal to 0.16 (Table 3), only a small amount of Br- (0.002 g Br m-2) was leached (Table 5, 25 Aug. 1995–31 Oct. 1995). The ratio of preferential/matrix flow contributing to this export was about 1:1. In the events that followed from 17 Nov. 1995 (ratio = 0.25) to 19 Feb. 1996 (ratio = 0.61), this ratio increased to approximately 5:1. In the event on 3 May 1996 (ratio = 0.69), which had a peak-flow rate of only 0.24 L s-1, the ratio decreased to about 1:3, indicating that the importance of matrix transport increased with decreasing discharge flow rates.
The total load in the drainage outflow between August 1995 and August 1997 amounted to 1.87 g Br m-2, which corresponded to 18.7% of the applied dose. Most of the export occurred during the winter following the application, when 0.88 g Br m-2, corresponding to 47% of the total observed export, occurred between 17 Dec. 1995 and 12 Feb. 1996. Since most of the export during this event originated from preferential flow, the contribution of the preferential transport amounted to 1.29 g Br m-2, corresponding to 73% of the total Br- exported by drainage for this period.
Triggering of Preferential Transport
While the two preferential flow-triggering models, surface trigger and subsoil trigger, employ the same mechanistic concept, they differ considerably in the relative importance of the critical factors affecting flow rate and water storage. In the surface-trigger model, storage capacity of the surface is quite small, but infiltration capacity is rather large. Thus, rainfall intensity plays the dominant role and antecedent moisture content is irrelevant. In the subsoil-trigger model, water storage capacity is quite large, but the limiting subsoil infiltration rate is relatively small in comparison to the first model. Thus, preferential flow depends very much on the initial volume of the storage, that is, the history of antecedent events.
The times at which the two models predicted the onset of preferential flow in comparison with the observed drain discharge and tracer concentrations are shown in Fig. 10. Assuming a surface storage capacity of zero, and an average infiltration capacity of 10 mm h-1, preferential flow was predicted for six events from 24 Aug. 1995 to 13 Aug. 1997. Three predictions (7 Jan. 1996, 9 June 1996, 18 July 1997) clearly coincided with drainage discharge peaks where preferential flow contributed to tracer export according to previous analysis. In one case (24 July 1996), precipitation was not sufficient to trigger a discharge peak, but a concentration peak was observed. This was probably also the case on 23 June 1996 when the increase of discharge was not sufficient to justify sampling and thus no concentrations were measured. On 17 June 1997, the surface model predicted a preferential flow due to precipitation of 15 mm within an hour, but drains did not respond, indicating that precipitation was probably intense enough to trigger preferential flow but the total infiltration was not sufficient to raise the groundwater table enough to trigger a drain discharge.
|
CONCLUSIONS
Analyses of the drainage discharge vs. concentration of the drainage water indicated that preferential transport was consistently triggered during the 2 yr of monitoring. Aside from the correspondence between the peak flow and peak concentration, which is strong evidence of preferential flow, we also noticed that maximum peak concentration (16 mg Br L-1 in December 1995) occurred after the discharge of 45% of the drainable pore water volume. While the maximum Br- concentration of the base flow occurred after twice as much infiltration, that is, at 80% of the drainable pore water volume. Dye infiltration tests from a previous study had also yielded the same conclusion.
A simple two-domain modeling approach of Everts and Kanwar (1990) was used to partition the total Br- export to drains into preferential and matrix (or base) flow. This separation showed that preferential transport was the dominant transport mechanism, accounting for 73% of the Br- exported within nearly 24 mo after the application.
Using a simple bucket model to analyze which factors may have been decisive in triggering preferential flow events reaching down to the drains, it was concluded that a rather large storage volume had to be filled before preferential flow was generated by overflow conditions. The size of this storage suggests that channeling of water into preferential flow pathways generally occurred at the boundary between the topsoil layers and the dense subsoil layers. Due to the low permeability of the subsoil, low infiltration rates were sufficient in generating preferential flow, provided that the storage capacity of the soil above was exceeded. Only in four cases did the analysis suggest that preferential flow leading to a discharge peak in the drains was generated directly at the soil surface and not within the soil profile.
REFERENCES
This article has been cited by other articles:
|
|
 |
|
  H. H. Gerke, J. Dusek, T. Vogel, and J. M. Kohne Two-Dimensional Dual-Permeability Analyses of a Bromide Tracer Experiment on a Tile-Drained Field Vadose Zone J., August 23, 2007; 6(3): 651 - 667. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
A. Boivin, J. Simunek, M. Schiavon, and M. Th. van Genuchten Comparison of Pesticide Transport Processes in Three Tile-Drained Field Soils Using HYDRUS-2D Vadose Zone J., June 21, 2006; 5(3): 838 - 849. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 A. Gaur, R. Horton, D. B. Jaynes, and J. L. Baker Measured and Predicted Solute Transport in a Tile Drained Field Soil Sci. Soc. Am. J., April 19, 2006; 70(3): 872 - 881. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| The SCI Journals | Agronomy Journal | Crop Science | |||
| Journal of Natural Resources and Life Sciences Education |
Soil Science Society of America Journal | ||||
| Journal of Plant Registrations | Journal of Environmental Quality |
The Plant Genome | |||
