| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
a Dep. of Geographical Sciences, Univ. of Plymouth, Drake Circus, Plymouth UK PL4 8AA
b Dep. of Geology, Univ. of Georgia, Athens GA 30602
c Institute of Grassland and Environmental Research, North Wyke, Okehampton, Devon UK
d Scottish Crop Research Institute, Invergowrie, Dundee, UK
* Corresponding author (a.williams{at}plymouth.ac.uk)
 |
ABSTRACT |
|---|
 TOPNOTES ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
|
INTRODUCTION |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
Preferential pathways can occur in structureless media where the mechanism appears to be due to fluid instability. More generally, such flow develops because of the inherent structure of soils and is associated with macropores created by soil fauna, decayed root channels as well as shrinking clay minerals (Williams et al., 2000). The exact nature of the pathways depends on the soil medium in terms of its hydraulic conductivity, continuity of pores and water repellency, if any (Ritsema and Dekker, 1995). While it is straightforward to quantify the macropore area of planes within the soil (Deeks et al., 1999), this does not elucidate the nature of preferential flow because not all macropores participate. As well as being dependent on soil physical properties, preferential flow is influenced by soil water content, rainfall intensities, and infiltration rates.
Unlike macropore structure there is no universally agreed definition of preferential flow. The phenomenon implies that there is a large flux or high velocity of flow through a limited number of pathways bypassing regions of immobile water. The thresholds of velocities are undefined although, as they occur rapidly, the flux rate is high and the relative concentration of preferential flow to input is high. One of the standard techniques, therefore, used to characterize flow and transport processes in the soil is the breakthrough curve. The concentration of a solute is measured through time at one or more locations in the soil to determine the flow pattern. The method relies on being able to instrument a large number of sites and to monitor them frequently to determine the overall pattern.
Solute transport experiments investigating flow and chemical transport processes have often been conducted in the laboratory rather than the field because it is easier to control conditions and replicate measurements. An added advantage is that the phenomena can be monitored in great detail and an intensive monitoring program can be established. A number of laboratory column experiments have examined the importance of preferential flow through structured and unstructured soil (Zurmühl et al., 1991; Saxena et al., 1994; Quisenberry et al., 1994; Chendorain and Ghodrati, 1999). However, one of the major limitations of using cores is that the sample volume may be too small to represent functional macroporosity (Ogden et al., 1992). Other problems include soil samples being too small to encompass much of the inherent field variability, there is the possibility of substantial edge effects influencing water movement results, and most flow is limited to the vertical dimension.
Valuable information about solute transport and the influence of soil heterogeneity has been gained from field experiments. Biggar and Nielsen (1976), for example, in a classic field experiment examined the leaching patterns within 20 subplots of a 150-ha field and demonstrated that values of the hydraulic conductivity and soil water flux were log-normally distributed. They noted that 20 samples would allow the true mean to be calculated within an order of magnitude and 100 samples would allow the mean to be estimated within ± 50% of its value. In spite of this study the instrumentation of many field site experiments is limited which compromises the spatial resolution of the results. Similarly, the frequency of sampling is often of the order of once per day or week (Butters et al., 1989; Roth et al., 1991; Saxena et al., 1994), although hourly rates have been achieved for limited periods (Biggar and Nielsen, 1976).
Isolated large in situ soil blocks have a number of advantages when considering the importance of preferential flow because detailed observations about transport mechanisms can be made at a suitable spatial and temporal resolution (Addiscott et al., 1978; Cameron and Wild, 1982; Bowman et al., 1994; Poletika and Jury, 1994). In particular, recent advances in monitoring technology, including tensiometer-transducer systems and the in-situ spectrophotometer, have ensured that measurements of soil water status and chemical transport can be made using a dense sampling pattern at a sufficiently high resolution to encompass much of the structural heterogeneity (Holden et al., 1995; Ritsema and Dekker, 1995; Ju and Kung, 1997; Rasmussen et al., 2000).
The main rationale for the experiment was to investigate the spatial variability of preferential flow in a large soil block. This paper describes four steady-state applications of water with chloride and nitrate tracers to evaluate the importance of preferential flow in a well-structured soil. A fifth experiment was conducted in which leaching was delayed 10 d to determine the effect of postponing the spray application on a tracer stored in the soil. A specific objective was to simulate how solutes move when rain follows after a dry period. Comparisons were made between the initial and this final set of preferential flow results to provide information on flow path function and temporal variability.
|
MATERIALS AND METHODS |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
A 1.5 m-wide trench was excavated using a backhoe to a depth of 1.2 m leaving an undisturbed volume of soil 5.4 by 3.4 by 1.2 m. The vertical surfaces of the soil block were covered with a paste of nonexpanding clay as a primary seal. The sides of the trench were then supported using exterior plywood and posts.
For the purposes of instrumentation, the soil block was divided into three sections: at each end of the block a zone of 2 by 3 by 1 m for sampler installation, and a central zone 1 by 3 by 1 m for destructive sampling. All probes installed in the soil block were inserted horizontally to prevent the creation of unnatural vertical flow paths. Vertical installation would have created excessive surface damage. Several types of probes were installed including tensiometers to measure matric potential and suction cup samplers to extract soil solution for tracer analysis. Six probes of each type were installed in nine layers 0.1 m apart and 0.1 m below the soil surface.
Simulation of Rainfall and Tracer Application
A fundamental prerequisite of the steady-state experiments was the uniform application of a tracer. Natural rainfall was excluded from the site by placing a large polythene tunnel over the block and trench, and artificial rainfall was supplied using a spray rig. To simulate rainfall, water had to be applied at a realistic intensity for the temperate-maritime location (i.e., 0.5–10 mm h-1) and the application had to be homogenous over the surface of the soil block. Given a soil porosity of 0.55, using an application rate of 10 mm h-1, one pore-volume required water to be applied for 2.0 d, and at 5, 2, and 1 mm h-1 required 4.1, 10.3 and 20.6 d, respectively.
The desired rainfall range was achieved by using low-flow pressure regulators and a solenoid valve to pulse water flow. The 15 nozzles functioned with a minimum internal pressure of 0.04 MPa (0.35 bar) (controlled by the pressure regulators), which for instance, when combined with a 3-s pulse every 20 s resulted in simulated rainfall of 2 mm h-1. Results of rainfall collection experiments using storage collectors confirmed that spatial variations in application were minimal.
For the first four experiments potassium chloride (Cl 250 mg L-1) and potassium nitrate-N (NO3–N 150 mg L-1) were applied at steady-state application rates of 10, 5, 2, and 1 mm h-1. The surface water fluxes were established 10 d before tracer application. At the most intense application rate, it took about 2 h to apply the tracer from two 200-L containers and 20 h at the lowest application rate. Each experiment was completed within 5 d and then the soil was leached for a further five days. In the fifth experiment, nitrate-N was added in solution at the rate of 50 kg ha-1 (NO3–N 375 mg L-1) at a rate of 1 mm h-1 for 8.5 h followed by a period of 10 d without any water application. Finally, the soil was flushed with a flux of 5 mm h-1 for several days. Nine soil samples were collected from the central destructive sampling zone both before and after the leaching process, and the nitrate, nitrite, and ammonia concentrations analyzed.
Measurement of Soil Hydrological Parameters
Soil matric potential was monitored using tensiometers equipped with small pressure transducers (‘Micro, 0-15 PSI, differential’, RS Components, Corby UK). The transducers were individually calibrated according to the method of Dowd and Williams (1989) within a range of 0- to 3-m head of suction. The millivolt responses from the 54 transducers were multiplexed at 15-min intervals into a Campbell 21X datalogger (Campbell Scientific Ltd, Leics, UK).
Tracer Extraction and Analysis
Soil water was extracted using ceramic tube samplers 0.25 by 0.04 m, located as shown in Fig. 1, and connected to sample traps at the soil surface. A vacuum of a 1.5-m suction was applied using two small electric pumps, each routed through a manifold distributing the vacuum between 27 small traps. An integrated suction sampling and chloride determination system was developed using solenoid routing valves and flow-injection analysis equipment (Holden et al., 1995). Nine samplers were connected to a flow-injection system, with six systems servicing the soil block permitting analysis of 54 independent samples. The six flow-injection systems were operated in parallel so in any 3-min interval, six samples were being analyzed. Samples for nitrate analysis were obtained by manually removing each sample vial and the water concentration determined using test-strip methodology (Holden et al., 1994). For each soil sample, 50 mL of 1 mol KCl was added to 25 g of soil and the total oxidized N (nitrate plus nitrite) and ammonium-nitrate contents were analyzed using standard auto-analyzer techniques.
|
|
|
|
|
RESULTS |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
|
|
|
The highest peak concentrations were associated with the most intense application rates. Considerable mixing of existing water and incoming tracer took place in the block and mixing was almost as effective at the surface as at depth.
Advective-flow velocities were calculated for each sampler based on the time taken for 50% of the nitrate tracer to transfer (Table 1). The mean velocities calculated are slightly lower than values quoted in Williams et al. (2000) derived from chloride tracers' time to peak concentration (TPC). (The results were identical for the most intense rainfall intensity and diverged most at the lowest application rate.) Maximum velocity was 4.2 m d-1 measured at the 0.7-m depth during the 10 mm h-1 application rate. The greatest mean velocities of about 1.3 m d-1 were related to the most intense spray application rates (Runs 1–3, 10–2 mm h-1, Table 1). The mean velocities were lower at 0.7 m d-1 (Run 4) and 0.4 m d-1 (Run 5) when the application rates were lower (Run 4, 1 mm h-1) or there was a period of drying before leaching (Run 5),
There is limited evidence that faster routes were not distributed uniformly throughout the block but were important in the southeast quadrant, where five out of the fastest 10 routes were found. The same areas of the block exhibited fastest times to peak and had the greatest concentrations during each experiment
Temporal and Spatial Variability
Typical breakthrough curves found for chloride and nitrate are presented in Fig. 4 and 5. The most common types were a rapid response with high peak concentration and a rapid loss of mass, and a slower response with a lower peak concentration and a slow decline in mass. These types are characteristic of soils with dual porosity or in which water flows at a range of velocities. A third type, characteristic of two macropore routes, was observed infrequently in which a rapid initial response to peak concentration was followed by a decline and subsequent rise to secondary peak. The nature of the breakthrough was not confined to any depth and similar spatial variation in types of breakthrough and peak concentrations were found at all depths. A similar result was reported by Butters et al. (1989), who noted a high degree of variability of downward solute movement. They reported that a rapid breakthrough was observed at some sites but was delayed at others.
Figure 4 presents an example of the complexity of the chloride responses at the 50-cm depth during Run 2 and shows that the behavior was spatially variable in the horizontal plane. Of the six samplers at this depth, two of the samplers recorded rapid rises in concentration with peak C/C0 concentrations of 0.4. The tracers were transported rapidly through the soil and there was a tendency to bypass the resident water. Short times to a high relative peak concentration (C/C0 ) suggest that larger cracks and fissures were conducting the solutes rapidly to depth and therefore mixing was limited. At those sites that exhibited a slower response and reached C/C0 concentrations of 0.1 to 0.3, the existing water held in the small pores mixed with the more slowly infiltrating water. Relatively few samplers collected water at the 0.7-m depth or lower—in general only three out of six collected any water (Fig. 6). This reduction in the number of active routes can be explained by the difficulties installing samplers in the denser subsoil.
Comparison of Run 2 breakthrough curves for chloride and nitrate (Fig. 4 and 5), given that the sampling intervals were not identical, shows that they behaved in a similar way. Most samplers showed peak concentrations within the top 0.4 m occurring around 4.5 h for chloride and 5 h for nitrate. At the 0.5-m depth, both tracers peaked after 6 h at two sites and peaked after 10 h at a third. However, peak concentrations differed somewhat and C/C0 values were about 0.4 for chloride and 0.2 for nitrate. There is also some evidence to suggest that nitrate returned to background levels faster. As with chloride, losses of nitrate were mainly due to leaching. This agrees with the findings of Cameron and Wild (1982) who reported that there was no difference in rates of movement between chloride and nitrate.
Detailed examination of tracer behavior at different depths, and for the various spray rates showed a very complex picture. However, for individual runs, peak concentrations of both tracers were not closely related to depth (Fig. 7). For chloride, the peak C/C0 ranged from 0.7 at the surface to 0.3 at depth. In contrast nitrate showed less of a trend with values of 0.4 at the surface and 0.3 at depth.
|
|
Results of the ‘Delayed Leaching’ Experiment
Examination of the nitrate-N breakthrough curves presented in Fig. 9 and 10 shows that despite the interval of 10 d without any application of water, the breakthrough curves were similar to those monitored earlier, although the time to reach peak concentration was longer in all cases. The type of pathways, which were active, therefore, was similar to those recorded without the delay. Peak C/Co concentrations were about 0.3 as before and it seemed as if there had been no delay between nitrate application and leaching. Nitrate variability at the 0.5-m depth in Run 5 was as great as in Run 2. The pattern is not so clear at this depth with only one rapid response and the remainder slow. The main preferential flow pathway at Site 18 operated much as before although the peak was delayed by about 6 h. Similarly Site 32 was almost identical in Runs 2 and 5 except that the peak concentration was higher and the tail was more extended in the later experiment.
|
|
Storage of nitrate in the soil was determined by removing nine cores from the central section and subsampling at 0.1-m increments to 0.5 m following the application of nitrate first immediately before and second after leaching. Although 50 kg ha-1 was added to the block initially, the total nitrate-N content for eight of the 0.5-m cores ranged between 17 and 22 kg ha-1. One site however contained around 80 kg ha-1 indicating that the spray nozzle above it was probably leaking. The majority of the nitrate was stored in the top 0.1 m after application. When the soil was sampled 7 d after leaching began, the average nitrate-N content at seven sites was 3 and 12 kg ha-1 at the other two. The soil at 0.3- to 0.4-m depth tended to have higher concentrations than the soil above.
Ammonium nitrate concentrations in the cores were more uniform and ranged between 2.8 and 5.5 kg ha-1. The greatest concentration was found at the surface before leaching. Concentrations increased following leaching and averaged 5.3 kg ha-1 and ranged from 3.8 to 9.5 kg ha-1. The surface layer, 0 to 0.1 m, contained the highest nitrate-N concentrations.
|
DISCUSSION |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
Flow through the soil block was found to be extremely heterogeneous although observations of the profile suggested that its structure and texture were relatively uniform laterally. Considerable spatial variation in preferential flow was monitored in the soil block both between and within experiments. Preferential flow was mainly monitored in the southeastern quadrant of the block. This area exhibited the fastest times to peak and had the largest peak concentrations during every experiment. Such results are in accord with Quisenberry et al. (1994) who reported that macropores in grassland are not isolated and randomly distributed but are clustered in continuous zones. Other researchers such as Steenhuis et al. (1990) and Ritsema and Dekker (1994) have described the influence of soil structure in detail while the theoretical operation depending on biological and other pedological characteristics is discussed by Ju and Kung (1997).
The dominant zones of preferential flow appeared to be long lived. Comparison of advection velocities for pathways operating after 1 yr found that the preferential flow system was very stable. Several of the faster pathways operated 1 yr later although the rank order changed somewhat. Quisenberry et al. (1994) also clearly demonstrated that macropores tended to be very stable and hence preferential pathways endured for a long time. Furthermore, evidence of the stable and recurring pattern of preferential flow paths has been described by Ritsema et al. (1998) for nonstructured water repellent sandy soils.
Preferential flow was established in the well-structured soil at a range of rainfall intensities even though the soil was not saturated. The rapid flux of chemicals through these pathways by-passed the biologically active zone and potentially reduced the time for the degradation of potentially harmful pollutants. The results showed that the breakthrough pattern of response was similar in both the standard and delayed rainfall application experiments. For Run 5, analysis of the shape of the curves and times to peak suggested that leaching was suspended during period without rain. When rainwater application commenced, the nitrate enriched tracer water remained in the rapid pathways and started to move again as it was leached out by incoming spray water. Any change in behavior could be linked to the slight interaction of water held in the larger and finer pores. Using the argument advanced by Rasmussen et al. (2000), who indicated that during rainfall events kinematic wave energy causes the redistribution of water and therefore solutes within peds, it appears that when the rainfall application ceased there was no driving force to push the nitrate into the smaller pores. On rewetting, while some nitrate was pumped into the fine pores, most was transported further down the profile by macropore flow. A similar situation is also discussed by Youngs and Leeds-Harrison (1990) in their paper on transport processes in aggregated soils for the case in which the macropores are empty and the aggregates can regarded as being almost isolated. During the time that there is little water movement between aggregates, there is little convective movement of solutes and redistribution of solutes within the aggregates is minimal. However, during a subsequent high rainfall period, there is a rapid loss of nitrate because of the larger surface area of the packed soil aggregates. The implication of the experiment for nitrate leaching from agricultural land without crop cover is that there is relatively limited movement within the soil during intervening dry periods, and that the chemical is stored where it can be accessed rapidly during subsequent storms.
|
CONCLUSIONS |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
The pattern of response was similar in both the standard and delayed rainfall application experiments. Analysis of the shape of the breakthrough curves and time to peak in Run 5 suggested that nitrate movement was minimal during the intervening period. Our results also suggest that it is difficult to calculate nitrate mass flux in N field experiments because of the high variability in response even in a relatively uniform soil where a few preferential flow paths could dominate. The problem is exacerbated because, based on the soil analysis after the delayed spray application, some nitrate remained in the soil after leaching. Given the minimal concentration in the effluent when leaching ceased, it is likely that this nitrate would remain in the soil until mineralization or denitrification occurred.
These results have important implications for chemical transport in soils. If pesticides are sprayed on crops in spring while the soil is wet, the results suggest that even if the next rainfall event is delayed as long as the soil remains wet, then the pesticides will be leached when it rains.
Parts of the block were found to have faster response zones than others and these preferential flow zones persisted for at least 1 yr. Therefore, application of a one-dimensional modeling approach based on the Richard's equation plus the advection-dispersion equation (ADE) will be limited. One option is to use a range of possible soil parameters along the lines of a transfer function (Jury and Flühler, 1992) to predict a range of possible outcomes and compare them with the measured distribution. Most ADE models rely on advective velocities, which are strongly dependent on soil parameters. Two domain models may be more successful but require knowledge about the partitioning between matrix and preferential flow. A two dimensional modeling approach, which stresses the importance of spatial variability of properties and the interaction between areas in the profile, may offer the best hope of predicting solute transport in a structured soil. A further problem is that the pathways were very dynamic under relatively steady conditions and this will compound determining which pathways are active as rainfall intensity varies continuously through a storm.
|
ACKNOWLEDGMENTS |
|---|
|
NOTES |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
Received for publication August 31, 2001.
|
REFERENCES |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  K.-J.S. Kung, E. J. Kladivko, C. S. Helling, T. J. Gish, T. S. Steenhuis, and D. B. Jaynes Quantifying the Pore Size Spectrum of Macropore-Type Preferential Pathways under Transient Flow Vadose Zone J., August 24, 2006; 5(3): 978 - 989. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 K.-J. S. Kung, M. Hanke, C. S. Helling, E. J. Kladivko, T. J. Gish, T. S. Steenhuis, and D. B. Jaynes Quantifying Pore-Size Spectrum of Macropore-Type Preferential Pathways Soil Sci. Soc. Am. J., June 28, 2005; 69(4): 1196 - 1208. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 G. M. Preston and R. A. McBride Assessing the Use of Poplar Tree Systems as a Landfill Evapotranspiration Barrier with the SHAW Model Waste Management Research, August 1, 2004; 22(4): 291 - 305. [Abstract] [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 |
Vadose Zone Journal | ||||
| Journal of Plant Registrations | Journal of Environmental Quality |
The Plant Genome | |||
