SSSAJ Journal of Natural Resources and Life Sciences Education
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF) Free
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via ISI Web of Science (3)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Strock, J. S.
Right arrow Articles by Gumpertz, M. L.
Right arrow Search for Related Content
PubMed
Right arrow Articles by Strock, J. S.
Right arrow Articles by Gumpertz, M. L.
GeoRef
Right arrow GeoRef Citation
Agricola
Right arrow Articles by Strock, J. S.
Right arrow Articles by Gumpertz, M. L.
Soil Science Society of America Journal 65:1607-1617 (2001)
© 2001 Soil Science Society of America

DIVISION S-1 - SOIL PHYSICS

Spatial Variability of Water and Bromide Transport Through Variably Saturated Soil Blocks

J. S. Strock*,a, D. K. Casselb and M. L. Gumpertzc

a Southwest Research and Outreach Center, Univ. of Minnesota, Lamberton, MN 56152
b Dep. of Soil Science, North Carolina State Univ., Raleigh, NC 27695
c Statistics Dep., North Carolina State Univ., Raleigh, NC 27695

* Corresponding author (jstrock{at}soils.umn.edu)


   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES
 
Water and solute transport pathways through soil are very complex. Soil properties affecting solute transport vary spatially and temporally within a soil profile and across landscape positions. The objective of this laboratory study was to evaluate water and bromide (Br) transport through 38- by 38- by 60-cm-deep undisturbed blocks of Cecil soil (Clayey, kaolinitic, thermic Typic Kanhapludults) collected from three contrasting landscape positions (interfluve, linear slope, and foot slope) in the Piedmont region of North Carolina. Two replicate soil blocks from each position were placed on a grid lysimeter-plate effluent collection system which facilitated collection of the effluent from 81 discrete cells under -2.5 kPa pressure. Each block was equilibrated for 5 d with a once daily application of 3.5 L of 0.005 M CaSO4 solution by a water drop applicator at a rate of 14 mm h-1. Four-hundred mL of KBr solution (4000 g Br m-2) was uniformly sprayed onto the soil surface. Thereafter, 3.5 L of 0.005 M CaSO4 solution was applied daily for the duration of each experiment (19–33 d). Effluent volume and Br concentration in the effluent were measured daily for each of the 81 4 by 4-cm cells. Cumulative water outflow and Br distribution plots, spatial distribution of cumulative effluent percent, frequency plots, and Br breakthrough curves (BTCs) showed that differences in preferential flow of water and Br occurred for soil blocks from different landscape positions. Differences in preferential flow of water and Br were attributed to soil horizon thickness, soil texture and structure, macroporosity, and slope gradient. Preferential flow of water and Br under variably saturated conditions was found to be highly variable within a given soil profile and that differences in the distribution and magnitude of preferential flow occurred across topographic positions.

Abbreviations: BTC, breakthrough curves • REV, representative elementary volume • MLRA, Major Land Resource Area • TDR, time domain reflectometry • SWP, soil water pressure


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES
 
PUBLIC CONCERN about the impacts of agricultural chemical leaching on water quality has led to many chemical transport studies on soils throughout the USA. These studies have been conducted both in situ and in the laboratory (Quisenberry and Phillips, 1976; Andreini and Steenhuis, 1990; Jardine et al., 1990; Shipitalo et al., 1990; Ogden et al., 1992; Bowman et al., 1994; Wildenschild et al., 1994; Phillips et al., 1995; Ren et al., 1996; Hatfield et al., 1997; Kranz et al., 1998; Olson and Cassel, 1999; Bejat et al., 2000; de Rooij and Stagnitti, 2000; and Shaw et al., 2000). Solute transport depends on multiple factors including: soil properties, climate, weather, slope aspect and gradient, land use management, vegetation type, and landscape position. Minor changes in only one factor can have a large impact on the transport process. Furthermore, spatial and temporal variability in these factors contribute to the complexity associated with characterizing solute transport in the soil environment.

Landscape position and soil steepness or slope create a complex pattern of water and solute transport and soil profile development (Hall and Olson, 1991). Soil properties affecting solute transport vary spatially and temporally within a soil profile, and across different topographic locations (Anderson and Bouma 1977a,b; Bathke and Cassel, 1991). In order to provide a meaningful description of the flow process the level of observation or representative elementary volume (REV) must be smaller than the entire flow domain (i.e., watershed or field), but must also be large enough that it includes sufficient flow pathways capable of transporting water and solute preferentially (Bear, 1972).

Afyuni et al. (1994) and Olson and Cassel (1999) reported on field experiments evaluating the effect of landscape position on solute transport. In one experiment, bromide ion (Br) transport on a Georgeville silt loam soil in the North Carolina Piedmont was observed at three landscape positions: interfluve, linear slope, and foot slope (Afyuni et al., 1994). The authors detected, after 140 mm of natural rainfall, Br concentrations 1 m deep as high as 10 mg/kg at the interfluve and linear slope positions and as high as 20 mg/kg at the foot slope. Variability in solute transport was attributed to observed differences in soil physical properties, mainly lower clay content and higher saturated hydraulic conductivity at the foot slope compared with the interfluve and linear slopes. More recently, Olson and Cassel (1999) reported for a Hiwassee clay loam soil, that linear and shoulder slopes were less susceptible to Br leaching than foot slopes because of higher clay content in the subsoil at the latter landscape positions. Had these experiments been intended to measure and describe preferential flow it is questionable, using the reported experimental procedures, whether the researchers would have been successful. The experimental methods and scale of observation in these experiments was inadequate to provide a meaningful description of the preferential flow process. In addition, soil sampling procedures used for determining depth of solute transport were destructive and the results were often highly variable.

In order to describe the spatial variation of solute transport, experiments have been conducted in the laboratory on intact, undisturbed soil blocks (Andreini and Steenhuis, 1990; Bejat et al., 2000; Bowman et al. 1994; Phillips et al., 1995; de Rooij et al., 2000; Shaw et al., 2000; Shipitalo et al., 1990; Vervoort et al., 1999). The use of intact soil blocks allow measurement of solute transport under a variety of controlled boundary conditions and permit nondestructive sequential measurements over time. Recently, a system was designed to conduct solute transport experiments on variably saturated, intact, undisturbed soil blocks (Strock and Cassel, 2001). The control offered by this system allows for detailed study of the mechanisms that affect solute transport and preferential flow, mainly flow path continuity, hydrodynamic dispersion, and ionic diffusion along with the effect of topographic and soil properties such as slope gradient, soil horizon thickness, soil texture, soil structure, hydraulic conductivity and porosity on preferential flow.

In most cases, information on solute transport using intact, undisturbed soil blocks has been limited to soil blocks collected from level terrain and blocks with one representative soil horizon. Sufficient information is not available to understand and adequately describe preferential flow processes in relation to the complex pattern of water flow along a landscape continuum. Most attempts at describing water and solute transport at the landscape scale have either looked at too little detail or completely overlooked the process of preferential flow. An understanding of variability of solute transport and preferential flow across a landscape continuum is beneficial for identifying vulnerable agricultural areas for reducing nonpoint source pollution and for developing a rationale for site-specific alternative best management practices. The objective of this laboratory study was to evaluate water and Br transport through large, intact, undisturbed soil blocks collected from three contrasting landscape positions from the Piedmont region of North Carolina.


   MATERIALS AND METHODS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES
 
Two intact soil monoliths (38 by 38 by 60-cm-deep), with well developed horizonation, were collected between September and November 1997 from a site near Raleigh, NC (N 35° 45', W 78° 42') at three landscape positions: interfluve, linear slope, and foot slope. The study site was dominated by nearly level (0–2%) interfluve and foot slope areas with an associated gently sloping (2–6%), convex-concave, linear slope. This Piedmont soil occurred in Major Land Resource Area (MLRA) 136 and consisted primarily of well-drained soils mapped as Cecil sandy loam (Clayey, kaolinitic, thermic Typic Kanhapludults). The Piedmont is an erosional landscape and soils found in concave footslope positions generally have a cover of material transported from nearby higher elevations. The interfluve and foot slope positions were in orchardgrass (Dactylis glomerata L.) and the linear slope was in red clover (Trifolium pratense L.).

Two replicate soil blocks from each position were excavated and transferred to the laboratory using a method described by Strock and Cassel (2001). The authors described the development and evaluation of the hardware and experimental procedures designed to overcome the limitations of recent protocols for conducting laboratory solute transport experiments. Briefly, a rigid steel frame with inside dimensions the same as the desired soil block was hydraulically pressed into the soil. Soil material surrounding the frame-enclosed monolith was carefully excavated and the isolated soil pedestal was encased in an oversized, open-ended plywood frame lined with polyethylene plastic (Phillips et al., 1995). An expanding closed-cell polyurethane foam was injected into the void between the soil pedestal and the polyethylene lined plywood frames. The encased soil pedestal was separated from the soil material below, taking care not to break the soil pedestal, removed from the field, and transported to the laboratory.

Each soil block was placed on a grid lysimeter-plate constructed to separate effluent collected from the block into 81 contiguous 4 by 4-cm areas in a 9 by 9 grid (Strock and Cassel, 2001). During block equilibration and experimentation, a soil water pressure of -2.5 kPa was continuously maintained at the base of the soil block with a small vacuum pump to 81 3.9 by 3.9 by 0.5-cm-thick fritted glass tiles. Each of the 81 cells drained into a separate 60 mL capacity collection bottle contained in an acrylic vacuum collection chamber.

Water and solute transport through the soil blocks was a transient flow process. Each block was equilibrated for five days by applying 3.5 L of 0.005 M CaSO4 solution, at a rate of approximately 14 mm h-1, once daily at 24-h intervals to the entire surface of the soil block with a bi-directional water drop applicator (Strock and Cassel, 2001). Weak ionic strength CaSO4 solution was used to prevent deflocculation. A quasi-equilibrium flow process was assumed to be achieved when soil matric potential and volumetric water content measurements returned to the same values for three consecutive 24-h intervals. After the equilibration period, a onetime pulse application of 400 mL KBr, equivalent to 4000 g Br m-2, was applied uniformly to the surface of a soil block with a single agricultural spray nozzle (Delavan-Delta, Inc., Lexington, TN, model D2.5). Thereafter, 3.5 L of 0.005 M CaSO4 was applied to the entire surface of the block at 24 h intervals for the duration of the experiment ranging from 19 to 33 d depending on landscape position. Volume and Br concentration of the outflow were measured once daily for each of the 81 cells. Effluent Br concentrations were analyzed colorimetrically on a Latchat QuickChem automatic ion analyzer (Latchat Instruments, Milwaukee, WI).

Soil water matric potential and volumetric soil water content were measured once daily at 0800 h, prior to daily CaSO4 solution application, at 0.05, 0.25, 0.35, and 0.55 m below the surface of each soil block from the interfluve and linear slope, and at 0.05, 0.33, 0.43, and 0.55 m for blocks from the footslope (Strock and Cassel, 2001). Differences in instrument depths were due to variable horizon thickness. Miniature tensiometers (1 cm diam., 2.5 cm-long, 10 kPa bubbling pressure) were inserted horizontally through the foam surrounding the soil block at the four depths and sealed in place with expanding polyurethane foam. Soil water content was measured with time domain reflectometry (TDR) (Topp, 1993); pairs of parallel 2.5-mm-diameter by 11-cm-long stainless steel rods were installed horizontally in the soil block in a manner similar to that for the tensiometers. Small soil water monitoring devices were chosen to minimize possible soil disturbance during installation and interference with water and Br transport. No observable evidence was detected to indicate that the small soil water monitoring devices impeded water or Br transport.

Soil samples (7.6-cm-diam. by 7.6-cm) for textural analysis and saturated hydraulic conductivity were not collected from the soil blocks but rather collected separately at the field site. Soil samples were taken from soil blocks after completion of solute transport experiments (Strock and Cassel, 2001). Each soil block contained two well developed soil horizons. The A horizon of all soil blocks was divided by depth into two equal halves. Nine equally spaced undisturbed (5.0-cm-diam. by 5.0-cm) soil cores were collected from each half of the A horizon. The Bt horizons for the interfluve and linear slope was divided into two equal halves and 18 undisturbed soil samples were collected in the same manner as for the A horizon. The B horizon from the foot slope was not subdivided because this horizon was rather thin. A total of 9 equally spaced soil cores were collected from this horizon. A total of 36 undisturbed soil cores were collected from each interfluve and linear slope block and 27 cores were collected from each footslope block. Soil samples collected from individual soil blocks were used to measure soil physical properties. Each soil core taken from the soil blocks was placed in a pressure outflow system (Danielson and Sutherland, 1986), slowly saturated from the bottom, and desorbed step-wise through a series of decreasing soil water pressures (0, -0.25, -1.0, -2.5, -5.0, -10.0, -20, and -30 kPa).

Cumulative distribution plotting techniques were used to evaluate whether water and/or Br transport occurred preferentially. Cumulative distribution plotting techniques have recently been used to provide evidence of preferential flow pathways through undisturbed soil blocks (Bowman et al., 1994; Quisenberry et al., 1994; and de Rooij, 2000). Cumulative distribution plotting methods require measuring cumulative percentage effluent volume and/or cumulative percentage of solute mass collected for each of the 81 individual cells. The procedure for describing data by cumulative distribution plots begins, for example, by ranking individual cell values for total mL of effluent obtained during an experiment in descending order. Next, the highest two ranked values were added together to obtain a cumulative effluent volume value. Finally, each subsequently ranked value was consecutively added to the previous cumulative value, and plotted as a function of percent cumulative basal area of the soil monolith.

Moran's I statistic was used to test cumulative mL of effluent for spatial correlation (Gumpertz, 1997; Upton and Fingleton, 1985). Moran's I is a measure of spatial autocorrelation and has the form

[1]
where n is the number of locations, 81 in our case, Wij is a measure of the physical proximity of the locations i and j, with Xi and Xj being the observed cumulative mL of effluent at locations i and j (Gumpertz, 1997). The term Wij is the element in the ith row and the jth column of the summary matrix W.

Two methods used to express physical proximity for data organized in rectangular lattices use the "rook's" and the "bishop's" definitions (Upton and Fingleton, 1985). According to the rook's definition of proximity, locations are neighbors if they share a common boundary edge, excluding corners. The bishop's definition of proximity is that locations are considered to be neighbors if they share common corners. Table 1 shows W matrices with a hypothetical arrangement of 9 neighbor locations for a 3 x 3 lattice. Neighbors to location X, the center element in the array, for the rook's and bishop's definition are shown. For a given pair of locations, Wij = 1 if the locations are neighbors, while Wij = 0 otherwise.


View this table:
[in this window]
[in a new window]
  		Table 1. Definition of neighbors for rectangular 3 x 3 lattices.

 

   RESULTS AND DISCUSSION
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
CONCLUSIONS
 REFERENCES
 
Water Outflow
Percent cumulative effluent volume and percent cumulative Br in the effluent versus percent cumulative basal area are shown in Fig. 1 for the two replicate soil blocks at each landscape position. If discharge through all cells were homogenous, the resulting plot would be linear, as indicated by the straight 1:1 lines. Fifty percent of the total cumulative effluent volume was measured in 30 and 17% of the basal area for the interfluve, in 14 and 15% of the basal area for the linear slope, and in 27 and 25% of the basal area for the foot slope (Fig. 1a). Correspondingly, 50% of the cumulative Br mass was collected in 22, 21,12, 15, 22, and 22% of the basal area (Fig. 1b) for the same replicate soil blocks from the interfluve, linear slope, and foot slope, respectively.



View larger version (23K):
[in this window]
[in a new window]
  		Fig. 1. (a) Cumulative effluent volume (%) vs. cumulative basal area (%), and (b) cumulative bromide mass (%) vs. cumulative basal area (%) for undisturbed soil blocks from a Piedmont landscape.

 
The slope of all cumulative effluent volume plots in Fig. 1a decreases as percent cumulative basal area increases. The cumulative effluent volume plots for these soil blocks show a pattern of symmetric variation across landscape positions. Curves with steep initial slopes represent flow pathways with high flow rates. Pathways with slower flow rates are indicated by curves with gradually increasing initial slopes.

The duration of each Br transport experiment and volume of CaSO4 solution applied and collected in the effluent are shown in Table 2. Variability in experiment duration between linear and foot slope positions was considerable, 11 d, but variability between blocks from the same landscape position was negligible for these two positions. Blocks from the interfluve position exhibited considerable variability in duration of experiment. At the time of Br application to soil blocks the background level of Br in effluent collected from all blocks was negligible. Experiments were terminated when Br concentration of the effluent approached the initial background levels.


View this table:
[in this window]
[in a new window]
  		Table 2. Duration of Br transport experiments and volume of CaSO4 applied and collected from replicate soil blocks from three landscape positions.

 
Soil blocks from the interfluve and foot slope positions were taken from level terrain and the surfaces of these blocks were also level. In contrast, replicate soil blocks taken from the linear slope possessed the same surface gradient as present in the field, 6%. Soil properties within and among soil blocks from contrasting landscape positions were vertically and horizontally heterogeneous (Table 3). Compared with the Bt horizon, the A horizon of the interfluve, for example, had bulk density values about 0.2 to 0.4 Mg m-3 greater, total porosity values of 0.09 to 0.15 m3 m-3 smaller, and macroporosity values 0.05 m3 m-3 smaller to 0.04 m3 m-3 greater. Vertical soil property variability for the linear slope was similar (Table 3). The degree of soil property variability within a given soil layer of a single core is indicated by the standard error which was greatest for bulk density, ranging from <0.01 to 0.04 in block L2. Soil properties from the foot slope varied by depth within blocks but exhibited similar soil property values between the two blocks.


View this table:
[in this window]
[in a new window]
  		Table 3. Selected soil property data for three contrasting landscape positions.

 
The size distribution of coarse fragments in the A horizon for each landscape position are shown in Table 4. Of the coarse fragments found in the A horizon at the linear slope position 39% by volume were >9.5 mm. The majority of the coarse fragments were located in a 5- to 10-cm-thick layer at the interface between the 30-cm thick surface and 30-cm thick subsurface horizons. This layer was very dense and difficult to dig through and varied in thickness across the linear slope.


View this table:
[in this window]
[in a new window]
  		Table 4. Volume percentage coarse fragments in A horizon of a Cecil sandy loam soil from three landscape positions.

 
Differences in the spatial distribution of cells with rapid flow rates among landscape positions is partially related to the size and number of preferential pathways, or macropores. Macropores, as defined in this study, were pores which drained at soil water pressures between 0 and -4.9 kPa (diameters >60 µm). This range of pore sizes was chosen because it is close to the theoretical range of pore sizes drained during the 24 h drying cycle. The mean soil water pressure near the surface of all soil blocks after 24 h (Fig. 2) was -4.7 kPa. In general, macroporosity decreased with depth for I2, L1 and L2 (Table 3). Macropores occupied 0.15 to 0.24 m3 m-3 in the A horizons for I2, L1, and L2, but only 0.09 to 0.14 m3 m-3 in the Bt horizons for the same profiles. The relatively high macroporosity in the A horizon for L1 and L2 could be attributed to the abundance of coarse fragments in this horizon, root channels, or due to fracturing during sampling. Fracturing during collection of the undisturbed soils cores was not observed. Prominent macropores from root channels were not detected. Soil aggregates with equivalent mean diameter >=5 mm can have considerable macroporosity in the form of inter-aggregate voids (Hillel, 1982).



View larger version (24K):
[in this window]
[in a new window]
  		Fig. 2. Mean volumetric water content and soil water pressure, from bromide transport experiments lasting 19–33 d, at four depths and after 22 hr of drainage.

 
In general, macroporosity increased with depth at I1, F1 and F2 (Table 3). Macroporosity ranged from 0.11 to 0.13 m3 m-3 in the A horizons for I1, F1, and F2, and from 0.16 to 0.17 m3 m-3 in the Bt horizon of I1 and the B horizons of F1 and F2. The somewhat lower macroporosity in the surface layers at these locations, particularly at F1 and F2, may be attributed to depositional processes and the configuration of a wide range of particle sizes including fine and very fine sands transported from higher elevations on the landscape.

Mean soil water pressure (SWP) and volumetric water content ({theta}) values at four depths for each soil block are shown in Fig. 2 for the duration of each simulation period given in Table 2. Mean measured SWP for all soil blocks 5 cm above the base of each block was -2.5 kPa with a coefficient of variation of 2.8%. Differences in {theta} and especially SWP among interfluve, linear slope, and foot slope positions were attributed to variations in thickness of sandy loam horizons (0.3 to 0.6 m), occurrence of an argillic horizon, and type and class of soil structure (Table 3). Soil water pressure and volumetric water content for each soil block were monitored at periodic intervals for one 24 h period to measure changes in soil water status during block wetting and draining. Flow within the sandy loam surface horizon at all landscape positions approached nearly saturated conditions during the 2 h wetting period as indicated by soil water pressure values near 0 kPa (Fig. 3) . Soil water pressure generally increased more slowly for blocks from the interfluve and most rapidly for the foot slope. Differences in duration of nearly saturated conditions among blocks was in part due to the thickness of sandy loam soil at the foot slope (0.6 m) and the interfluve and linear slope (0.3 m). In addition, a subsurface argillic horizon was present at 0.3 m depth for the interfluve and linear slope positions.



View larger version (31K):
[in this window]
[in a new window]
  		Fig. 3. Mean soil water pressure versus time and at four depths for replicate soil blocks taken from interfluve, linear slope, and foot slope landscape positions.

 
The complexity of the flow process can be visualized in Fig. 4 where the percent of total block effluent passing through each cell is plotted in one of seven classes: 0.00–0.50, 0.51–1.00, 1.01–1.50, 1.51–2.00, 2.01–2.50, 2.51–3.00, and >3.00%. Flow was considered to be preferential in a given cell if the cumulative effluent percent was >=1.51%. Homogeneous flow of a once daily application of 3.5 L of CaSO4 solution to each soil block would result in the daily collection of 1.1% or 0.04 L of water and solute per cell per day. The observation that multiple cells discharged considerably greater amounts of effluent than this on a daily basis provides evidence of preferential flow. Quisenberry et al. (1994) suggested that the total volume of flow through a pore intersecting the soil surface would be small under non-ponded conditions if the source of water in the pore is limited to water entering at the soil surface. The total volume of flow through the same pore would be increased if a significant amount of water enters the main pore through a network of interconnected pores beneath the soil surface (Quisenberry et al., 1994). According to Phillips et al. (1989), once flow is established in a pathway capable of conducting large volumes of water, it may have the ability to siphon appreciable amounts of water from nearby soil pores.



View larger version (88K):
[in this window]
[in a new window]
  		Fig. 4. Spatial distribution of cumulative effluent percent by cell from interfluve 1 (a), interfluve 2 (b), linear slope 1 (c), linear slope 2 (d), foot slope 1 (e), and foot slope 2 (f).

 
Figure 4 shows the spatial variability of cumulative effluent percent between blocks from the same landscape position and the variability of cumulative effluent percent among landscape positions. Visual observation of the cumulative effluent percent for these plots lead to the supposition that an appreciable amount of wall flow occurred, especially for the interfluve and linear slope positions. However, as previously indicated, the gradient of the subsurface horizon at the linear slope position paralleled the surface slope gradient of 6% and we cannot state with certainty that some wall flow did not occur. In addition, a layer of coarse fragments existed at the interface between the A and Bt horizons at this landscape position. Finally, no surface flow of simulated rainfall down gradient was ever observed for either linear slope block. Diversion of water horizontally within the soil blocks at the interface between horizons as a consequence of slope gradient and the dense consolidated layer of coarse material between horizons (Table 3) would promote lateral transport and thus the observed spatial distribution of cumulative effluent percent along the down slope side in these two soil blocks and the apparent wall flow (Fig. 4). The frequency distributions in Fig. 5 show that the foot slope position had more preferentially flowing cells (cumulative outflow >=1.51%) than the interfluve and linear slope positions (46, 38, and 35%, respectively). On the basis of daily observations, at no time during experimentation did ponding or surface runoff occur for any of the six soil blocks. Ponding or runoff could lead to wall flow if it occurred near the interface between the soil and the protective wall surrounding the block. Consequently we conclude that the wall flow that did occur for interfluve and linear slope positions originated primarily as a result of the argillic subsurface horizon and flow through pores that intersected the soil/wall interface at some depth below the soil surface. In addition, the lack of apparent wall flow within soil blocks from the foot slope position was due to the thickness of the sandy loam soil material found at this position and the lack of an argillic subsurface horizon.



View larger version (26K):
[in this window]
[in a new window]
  		Fig. 5. Frequency distribution of cumulative outflow percent from interfluve 1 (a), interfluve 2 (b), linear slope 1 (c), linear slope 2 (d), foot slope 1 (e), and foot slope 2 (f).

 
The Moran's I statistic was used to test the hypothesis that no spatial correlation existed between adjacent cells. When there is a maximum positive spatial autocorrelation between adjacent cells Moran's I approaches a value of 1 (Upton and Fingleton, 1985). When no spatial correlation exists, the expected value of Moran's I will be -0.01 [-1/(n-1), where n = 81 is the number of locations]. Moran's I statistic of 0.3 or greater in Table 5 presents evidence of spatial correlation between cells sharing common edges (rook's definition of neighbors) for all landscape positions. No evidence of diagonal (bishop's definition of neighbors) spatial correlation among blocks existed for the interfluve or linear slope, but did exist for the foot slope position (Table 5). These results indicate that, using rook's definition of neighbors, groups of cells exhibiting similar cumulative effluent percentages tend to be clustered in a non-random fashion. Localized interconnected networks of flow pathways could lead to the observed clustered regions of preferential water and solute transport.


View this table:
[in this window]
[in a new window]
  		Table 5. Moran's I for cumulative effluent percent based on rook's and bishop's definition of proximity.

 
The above results, that is, the cumulative distribution plots, the spatial distribution of cumulative effluent volume percentage, and the frequency of cumulative effluent volume percentage >=1.51 collectively provide evidence that preferential flow paths existed for the soil blocks at each landscape position. The data also indicate that flow of water was highly variable within and between soil blocks from contrasting landscape positions.

Bromide Transport
A Br breakthrough curve (BTC) for each soil block (Fig. 6) was constructed from cumulative total daily effluent volume and cumulative total daily Br concentration data. Data from all soil blocks indicate initial rapid breakthrough of Br. Close inspection of the BTC's from the interfluve (Fig. 6a) and the foot slope (Fig. 6c) reveal the presence of a second maximum. The existence of a second peak on BTCs suggest nonuniform flow of Br through these soil blocks. The observation of double peaks in solute transport is well documented, but there is little agreement regarding the causes.



View larger version (19K):
[in this window]
[in a new window]
  		Fig. 6. Breakthrough curves for replicate soil blocks taken from the interfluve (a), linear slope (b), and foot slope (c).

 
Several investigators have attributed preferential flow and bimodal BTCs to flow heterogeneities in soils (Skopp et al., 1981; Hamlen and Kachanoski, 1992). Soil heterogeneity, such as differences in pore-size distribution and abrupt changes in soil texture, contribute to but might not completely explain preferential flow of water and solute flow through soil. In laboratory experiments, wall effects may contribute to preferential flow and occurrence of double peaks. Short input solute pulse duration, as in this study, have also been attributed to producing bimodal BTCs due to rapid solute flow velocities occurring in preferential pathways (Ma and Selim, 1994). Ma and Selim (1994) hypothesized that preferential flow through structured soils also might result in solute BTCs with a bimodal distribution. Roth et al. (1991) suggested that, during intermittent rainfall events, when the rainfall rate is less than the infiltration rate of the soil, a portion of the soil profile becomes saturated and water begins to move through, creating a preferential flow event. Initially, this preferential flow event flushes solutes from the soil matrix near the preferentially conducting pathways. After infiltration ceases, flow in the preferential pathways diminishes or stops and solute concentrations in the immediate region of the preferential pathways increases again due to diffusive processes within the soil matrix. This sequence of events is repeated with every preferential flow event. The resulting BTCs for a soil undergoing a series of successive infiltration events, as in the current study, would exhibit a bimodal distribution of solute in the effluent.

Recovery of the applied Br in effluents at the termination of the study were 94 and 90% for the interfluve, 92 and 83% for the linear slope, and 83 and 76% for the footslope. Since all experiments were conducted until Br recovery in the daily effluent became negligible, we are unable to account for the relatively poor Br recovery at the foot slope position. One possibility is that Br might have been adsorbed by soil organic matter which was greater at the foot slope position compared to the interfluve and linear slope (data not shown). The similarity in the cumulative effluent volume percent and cumulative Br mass cumulative distribution plots (Fig. 1) suggests that the mass of Br transported through a given cell was related to the percent effluent being discharged through that cell. The relationship between percent Br mass collected versus percent effluent discharged for each cell for two replicate soil blocks from each landscape position is shown in Fig. 7 . Good agreement exists between percent total Br mass and cumulative effluent volume percent discharged suggesting that those cells conducting the most effluent correspondingly transported the most Br for all landscape positions. Correlation coefficient (r2) values from regression analysis were 0.82, 0.87, and 0.89 for interfluve, linear slope, and footslope blocks, respectively.



View larger version (20K):
[in this window]
[in a new window]
  		Fig. 7. Regression of percent cumulative effluent versus percent total bromide for (a) interfluve 1 and 2, (b) linear slope 1 and 2, and (c) footslope 1 and 2.

 

   CONCLUSIONS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
REFERENCES
 
Preferential flow in variably saturated soil blocks undergoing diurnal wetting and draining cycles was highly variable for soil blocks collected from each of three landscape positions. Furthermore, the cumulative effluent volume and the fraction of total Br mass collected were highly variable among the 81 cells for each replicate soil block. Differences in preferential flow among landscape positions are attributed to a combination of soil profile characteristics, mainly, thickness of the sandy surface horizon, presence of an argillic subsurface horizon, type and class of soil structure, macroporosity, and slope gradient. The foot slope, which had the lowest clay content, weakest soil structure, and greater subsurface macroporosity, exhibited the most rapid solute transport and thus is considered to be the most susceptible to leaching losses. The combination of a decrease in subsurface macroporosity, an increase in clay content, a coarse-textured dense layer at the interface between the surface and subsurface layers, and a natural slope gradient gave rise to a horizontal component of solute transport at the linear slope position. Evidence of horizontal solute transport was distinguished by regions of relative high cumulative effluent volume percent in the direction of the down slope gradient and low flow regions in the upslope regions.

Analysis of preferential flow at these landscape positions for the imposed soil conditions indicated that the spatial distribution of solute transport was not random but occurred in distinct clusters. Spatial analysis by Moran's I indicated horizontal (rook's definition of proximity) correlations between adjacent cells for all landscape positions. The cumulative effluent volume and Br concentration in the effluent of individual cells were positively correlated. The spatial distribution of cumulative effluent volume percent indicated clustering among cells of similar effluent volume. Cells which conducted the most effluent also transported the greatest masses of Br.

The effects of soil morphology, landscape position, and slope gradient on preferential flow were evident under the experimental conditions imposed during this study. Wall flow, although not conclusively identified as the cause of flow patterns near the perimeters of blocks from the interfluve and linear slope positions, cannot be discounted. Our study substantiates that soil morphological attributes that change with landscape position affect water and solute transport pathways. For this landscape the foot slope position would be more likely to serve as a nonpoint source of mobile nutrients such as nitrate to surface and ground waters.

Received for publication May 22, 2000.


   REFERENCES
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES
 




This article has been cited by other articles:


Home page
 Vadose Zone JHome page
G. H. de Rooij, O. A. Cirpka, F. Stagnitti, S. H. Vuurens, and J. Boll
Quantifying Minimum Monolith Size and Solute Dilution from Multi-Compartment Percolation Sampler Data
Vadose Zone J., October 3, 2006; 5(4): 1086 - 1092.
[Abstract] [Full Text] [PDF]

Home page
 Vadose Zone JHome page
U. Buczko and H. H. Gerke
Modeling Two-Dimensional Water Flow and Bromide Transport in a Heterogeneous Lignitic Mine Soil
Vadose Zone J., December 16, 2005; 5(1): 14 - 26.
[Abstract] [Full Text] [PDF]

This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF) Free
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via ISI Web of Science (3)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Strock, J. S.
Right arrow Articles by Gumpertz, M. L.
Right arrow Search for Related Content
PubMed
Right arrow Articles by Strock, J. S.
Right arrow Articles by Gumpertz, M. L.
GeoRef
Right arrow GeoRef Citation
Agricola
Right arrow Articles by Strock, J. S.
Right arrow Articles by Gumpertz, M. L.

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