VZJ sign up for citetrack
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

Published online 13 September 2005
Published in Vadose Zone J 4:915-923 (2005)
DOI: 10.2136/vzj2004.0162
© 2005 Soil Science Society of America
677 S. Segoe Rd., Madison, WI 53711 USA
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 (4)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Dyck, M. F.
Right arrow Articles by de Jong, E.
Right arrow Search for Related Content
PubMed
Right arrow Articles by Dyck, M. F.
Right arrow Articles by de Jong, E.
GeoRef
Right arrow GeoRef Citation
Agricola
Right arrow Articles by Dyck, M. F.
Right arrow Articles by de Jong, E.
Related Collections
Right arrow Soil Physics
Right arrow Vadose Zone Processes and Chemical Transport

ORIGINAL RESEARCH

Spatial Variability of Long-Term Chloride Transport under Semiarid Conditions

Pedon Scale

M. F. Dycka,*, R. G. Kachanoskib and E. de Jongc

a Dep. of Renewable Resources, 751 General Services Building, Univ. of Alberta, Edmonton, AB T6G 2H1, Canada
b V. P. Research Office, 3rd Floor University Hall, Univ. of Alberta, Edmonton, AB T6G 2J9, Canada
c Dep. of Soil Science, 51 Campus Drive, Univ. of Saskatchewan, Saskatoon, SK S7N 5A8, Canada
* Corresponding author (mdyck{at}ualberta.ca)

Received 16 November 2004.



   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 CONCLUSIONS
 REFERENCES
 
Knowledge about the magnitude and variability of water and solute fluxes in semiarid environments is required to assess the environmental risk associated with contaminants in the vadose zone. The purpose of this study was to examine the pedon-scale spatial variability in the transport of a chloride tracer after 34 yr at a site 50 km southwest of Saskatoon, SK. Soil cores were taken from two, pedon-scale transects (transects encompassing many pedons), and solute transport was quantified for each depth breakthrough curve using the method of moments. Modal transport depth (depth of peak concentration) did not vary significantly over the length of the transects (average = 1.34 m), indicating a relatively uniform soil water balance at the pedon scale. Therefore, trends in central moments (mean travel depth (E[z]), and variance about E[z]) were attributed to spatial variations in soil layering. Transport variance was negatively correlated (P < 0.05) to the thickness of a fine-textured, varved layer. The high water content (transport volume) of the varved layer decreased the velocity of the leading edge of the solute depth breakthrough curve which, over time, decreased V[z]. The two-dimensional distribution of the chloride shows that the estimated horizontal velocity of the chloride pulse (approximately 25 mm yr–1) is more than twice the estimated downward, vertical velocity (approximately 11 mm yr–1). The large horizontal flow component is attributed to anisotropy in soil hydraulic properties. The significant influence of layers on convective and dispersive solute fluxes in a low flow, semiarid environment under natural conditions, complements previous observations in high flow field experiments where two- and three-dimensional flow occurred along layer boundaries, and suggests anisotropy should be incorporated into transport models.

Abbreviations: BTC, breakthrough curve • CLT, convective lognormal transfer function • CS, convective stochastic • DBTC, depth breakthrough curve • E[z], mean travel depth • V[z], transport variance


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 CONCLUSIONS
 REFERENCES
 
ASSESSMENT OF THE ENVIRONMENTAL RISK associated with the transport of contaminants in the vadose zone, requires detailed knowledge of transport processes in field soils. Currently, knowledge about the magnitude and variability of solute fluxes in field soils in semiarid environments is limited. The field transport experiments undertaken to date have been performed under high water flux, and often steady-state boundary conditions which may not be indicative of semiarid environments (e.g., Butters et al., 1989; Ellsworth et al., 1991; Ward et al., 1995). In semiarid environments, boundary conditions at the soil surface are transient and deep drainage, which drives deep convective fluxes, is often small and difficult to quantify. Transport process problems are further complicated when soil heterogeneity (i.e., soil layering), is taken into account. For example, the physics of flow at layer boundaries is not well understood, but layer interfaces appear to be focal points for two- and three-dimensional flow.

Numerous field transport experiments have investigated solute transport in layered soils. Ward et al. (1995) compared unsaturated solute transport on a 20-m transect in the field and in undisturbed columns taken parallel to the field transect every 40 cm (n = 50). They found that 80% of the field-scale transport variance occurred at scales of 0.4 m for the soil columns while scales of 1.8 to 3.0 m accounted for the majority of the transport variance in the field transect. The different transport variance scales of the columns and field were attributed to lateral flow along horizon boundaries in the field. Layering effects on solute transport at the same site were further elucidated by van Wesenbeeck and Kachanoski (1991)( 1994). Van Wesenbeeck and Kachanoski (1991) found that the majority of transport variance occurred at scales corresponding to B horizon thickness variations. Furthermore, it was shown that lateral flow along the B–C horizon boundary accounted for significant three-dimensional redistribution of tracer mass (van Wesenbeeck and Kachanoski, 1994). Hamlen and Kachanoski (1992) found that the travel time of a tracer through the A horizon was partially correlated to transport through the B horizon, but significant lateral flow at the A–B horizon boundary may have invalidated the one-dimensional assumptions of convective stochastic (CS) process models such as the convective lognormal transfer function (CLT) model (Jury and Roth, 1990).

Field unsaturated solute transport experiments in California (Butters et al., 1989; Ellsworth et al., 1991) also showed the effects of layering on solute transport. Both experiments were performed at the same site. Transport variance increased to depths of approximately 3.0 m, decreased between 3.0 and 4.5 m and then increased again. The decrease in transport variance was attributed to slower flow through a fine-textured silty layer between 3.0 and 4.5 m. There was also evidence of lateral flow along the upper boundary of the silty layer as the three-dimensional mean travel depth calculated with the method of moments was offset from the center of the original cubic plume in the direction of the slope of the fine-textured layer (Ellsworth et al., 1991).

Another example of a transient field tracer experiment is that of Hammel et al. (1999). In this case a tracer was applied to the surface of a 20-m transect in Germany with variable A horizon thickness and texture. Significant redistribution of the tracer occurred along the A–B horizon boundary. The distribution of the tracer was predicted with a numerical solution of the Richards' equation based on measured {theta} ({psi}) and K({psi}) functions assuming horizontally homogeneous soil horizons and horizontally heterogeneous soil horizons (correlation length of 0.2 and 0.5 m). The tracer distributions modeled with the horizontally heterogeneous soil horizon assumptions provided the best prediction of the field-scale breakthrough curve (BTC) and variance. The comparability of the measured and predicted tracer distributions suggests that the inherently different hydraulic properties of pedogenic horizons as well as variability within horizons explain a large portion of the field-scale transport variance.

More recently, Javaux and Vanclooster (2004a)(2004b) monitored long-term (15 yr) chloride transport through a heterogeneous vadose zone beneath a lake in Belgium. They used inverse procedures with a layered, convective dispersive model to get an optimized estimate of dispersivity. The optimized dispersivity estimates were unrealistically high (Javaux and Vanclooster, 2004b). The authors attributed the high dispersivity estimates to fingered flow or flow convergence below sand–clay layer interfaces. Due to practical constraints, the number of in situ measurements (using solution cup samplers) was limited making detailed observations of the processes difficult.

The transport experiments discussed above were performed under high water flux (cm h–1 to cm d–1), convection dominated boundary conditions. Deep convective fluxes in semiarid environments are lower by several orders of magnitude (mm yr–1 or cm yr–1). There are few examples of significant lateral flow in semiarid environments. McCord and Stephens (1987) and McCord et al. (1991) performed a tracer experiment under transient conditions in New Mexico in a cross-bedded dune sand. They also observed significant lateral flow and transport downslope, but no obvious layers were present in the soil. It is unclear whether significant lateral flow at horizon boundaries occurs under low flow conditions found in semiarid environments.

Lateral flow at soil horizon boundaries has often been explained by the fact that the layered soils are anisotropic at the profile scale (Zaslavsky and Rogowski, 1969; Zaslavsky and Sinai, 1981; Stephens and Heermann, 1988). Specifically, layered soils are anisotropic with respect to hydraulic conductivity such that water flux in the horizontal direction may be greater than the vertical direction even through hydraulic gradients are greatest in the vertical direction. Microscale pore structure is another cause of anisotropy in soils (for a recent review of anisotropy, see Raats et al., 2004; Zhang et al., 2003).

Recently, the long-term ( >30 yr) average movement of a chloride tracer through soil in a semiarid environment was reported for a 10-m-long transect (Dyck et al., 2003). The spatial variability of mean travel depth along the 10-m transect after 34 yr of transport was relatively low (CV = 4%), but not insignificant. The purpose of this paper is to examine, in detail, the pedon-scale spatial variability in the transport of a 34-yr-old chloride tracer applied to a layered, field soil near Saskatoon, SK. Specifically, the objectives of this study are: (i) determine the effects of the local scale distribution of soil layers on the local scale variability of solute transport under semiarid conditions; and (ii) to quantify the long-term (34 yr) two-dimensional transport of the chloride tracer.


   MATERIALS AND METHODS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS
 DISCUSSION
 CONCLUSIONS
 REFERENCES
 
Site Description
The field site (51°52' N, 107°18' W) is located approximately 50-km west of Saskatoon, SK and has been described in detail by Dyck et al. (2003) and by Ballantyne (1974)(1980). The site was first established to determine the effects high rates of potash (KCl) on crop growth. In 1966, granular KCl was applied to the soil surface of five, 6 by 90-m plots (Fig. 1) . The two eastern most chloride plots received additional chloride applications in 1971 (Fig. 1). The soils are classified as the Elstow association: Dark Brown Chernozems (Typic Ustolls) developed on loamy glacio-lacustrine parent material (Ellis et al., 1968). The lacustrine sediments are underlain by glacial till which is drained by the Tessier aquifer (Meneley, 1975). The water table occurs at approximately 15-m below the surface within a sand layer (Meneley, 1975). The site has been under a crop-fallow rotation dominated by wheat (Triticum aestivum L.) with some barley (Hordeum vulgare L.) since the site was established (Gordon Carr, landowner, personal communication, 2000). The long-term average (1967–2000) precipitation at the site is 321 mm yr–1 with approximately 40% coming from September to May when evapotranspirational demand is low. Annual precipitation varied considerably from 1966 to 2000 ranging from 225 mm yr–1 to 481 mm yr–1, with a coefficient of variability of 64% (Dyck et al., 2003).



View larger version (63K):
[in this window]
[in a new window]
  		Fig. 1. Topographic map of field site with location of original potash (KCl) test plots. Intensive sampling transect (oriented north-south) and perpendicular sampling transect (oriented east-west) are represented by red lines. The background core is represented by a cross.

 
Soil Sampling and Sample Processing
In the summer of 2000, two pedon-scale sampling transects (transects encompassing many pedons) were established on the 3.4 kg KCl m–2 plot (Fig. 1). The first transect was oriented north–south, 2.0 and 8.0 m from the west and south boundaries of the 3.4 kg m–2 plot (Fig. 1). Fifty-one cores (5.3-cm diam., 5.3–6.6 m deep) were taken at 0.2-m intervals for a total length of 10 m. This transect shall subsequently be referred to as the intensive transect. A second transect was oriented east–west, 12.2 m from the south boundary of the 3.4 kg KCl m–2 plot; intersecting the 4.2-m sampling point of the intensive transect (Fig. 1). Ten cores were taken at 2.0-m intervals, for a total length of 18 m. This transect shall subsequently be referred to as the perpendicular transect. In addition to the two sampling transects, one core was taken approximately 20-m away from the chloride treatments to assess background concentrations (Fig. 1). The intensive transect was established with a small sampling interval to address layering effects on vertical transport at the pedon scale. Therefore, many cores were required over a short distance. The perpendicular transect was established to get an estimate of any significant lateral transport outside of the plot boundaries and therefore did not require intensive sampling.

The morphology and dimensions of the major soil horizons and sedimentary layers were described and recorded for each core. After description, each core was sliced into 0.10-m increments (approximately 3600 total samples) that were bagged and weighed in the field. The samples were subsequently dried and weighed for calculation of bulk density and field water content (gravimetric and volumetric). Chloride and other soluble salts were extracted by adding 30 g of water to a 15-g subsample of dried, ground soil and shaking for 1 h. Chloride concentrations in the extractions were determined colorimetrically on an autoanalyzer with a technique similar to the one described by Tel and Hesseltine (1990). Extract concentrations were used to calculate resident soil Cl concentrations (kg Cl kg–1 soil; kg Cl m–3 soil). For further details of field sampling and laboratory procedures see Dyck et al. (2003).

Background Cl concentrations were measured on the core taken from outside of the chloride plots (Fig. 1) and subtracted from each sample from the intensive and perpendicular transects depending on which soil horizon or sedimentary layer the sample was taken from. The background Cl concentrations were at least an order of magnitude lower than all samples from the chloride plots. Resident Cl concentrations (kg Cl m–3 soil) as a function of depth, Cj(z) , were calculated by multiplying the soil Cl concentration (kg Cl kg–1 dry soil) by bulk density.

Quantification of Solute Transport
Vertical solute transport on the intensive and perpendicular transects was quantified using the method of moments (Jury and Roth, 1990). The 0th, first, and second central moments for each spatial location, j, are the mass recovery, Mj (kg m–2), mean travel depth, Ej[z] (m), and variance about the mean travel depth, Vj[z] (m2):

[1]

[2]

[3]
where Z is the maximum depth of sampling. To be consistent, Z was limited to 5.3 m as this was the sampling depth of the shallowest depth breakthrough curve (DBTC).

For the perpendicular transect, two-dimensional mass recovery, M2D (kg m–1), and mean lateral displacement, E[x] (m), were calculated by:

[4]

[5]
where x is the location (m) along the perpendicular transect, M(x) is the lateral mass recovery function (approximated with linear interpolation), and x = A and x = B are the horizontal coordinates (m) of the start and end of the perpendicular transect.

Modal transport depth was estimated from the depth of the peak concentration on the 51 samples from the intensive transect. Using MathCad (Mathsoft, Cambridge, MA), the modal transport depth was calculated by fitting a cubic spline to each DBTC from the intensive transect. The depth at which the derivative of the cubic spline was zero (calculated using the root function in MathCad) on the interval where the peak occurred (assessed visually) was assumed to be the modal transport depth.


   RESULTS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
DISCUSSION
 CONCLUSIONS
 REFERENCES
 
Variations in Soil Layering
The depth distribution of the soil horizons and sedimentary layers for the intensive and perpendicular transects are presented in Fig. 2 and 3 . Surface topography is also included in Fig. 2 and 3, and the vertical axes represent absolute elevations according to the topographic map in Fig. 1. The perpendicular transect crosses the intensive transect at the 4.2-m location on the northing axis (Fig. 1). The lines representing the layer boundaries on the perpendicular transect (Fig. 3) appear to be smoother because of greater distance between sampling points. The parent material is lacustrine in origin, and well sorted. The texture of the A and B horizons is silty loam. The texture of the C horizon is fine-sandy loam. The varved layers consist of alternating beds of clay (about 1- to 3-cm thick) and silt (5- to 15-cm thick). The silt layer is fine-medium silt displaying some vertical fractures and the sand layer is fine-medium sand. The boundary between the layers labeled varving 1 and varving 2 is represented by a dashed line in Fig. 2. The difference in morphology and texture between these two layers is small. The layer labeled varving 2 is slightly coarser, and lighter in color. Subsequent references to the thickness of the varved layer include both the varving 1 and varving 2 layers. For further detail of layer descriptions see Dyck et al. (2003).



View larger version (24K):
[in this window]
[in a new window]
  		Fig. 2. Variations in depth to layer interfaces along the intensive transect. The dotted vertical line indicates the intersection point of the perpendicular transect.

 


View larger version (25K):
[in this window]
[in a new window]
  		Fig. 3. Variations in depth to layer interfaces along the perpendicular transect. The dashed vertical lines represent plot boundaries shown in Fig. 1. The dotted vertical line represents the intersection point of the intensive transect.

 
The dimensions of several layers for the intensive transect were correlated (Table 1), and reflect trends in Fig. 2. For example, the depth from the soil surface to the top of the varved layer and varved layer thickness were positively correlated (r = 0.35; P < 0.05). Depth from the soil surface to the top of the silt layer was strongly correlated to the thickness of the varved layer (r = 0.96; P < 0.01). The thickness of the silt layer was also correlated to the depth from the soil surface to the top of the sand layer (r = 0.58; P < 0.05).


View this table:
[in this window]
[in a new window]
  		Table 1. Correlation matrix for mass recovery (Mj), mean travel depth (Ej[z]), variance (Vj[z]) and layer dimensions.

 
Spatial Distribution of Volumetric Water Contents
The spatial distribution of volumetric water contents with respect to the layer boundaries is presented in Fig. 4 for the intensive transect, and in Fig. 5 for the perpendicular transect. High water contents are associated with the fine-textured varved layer. Dyck et al. (2003), suggested that the water contents below the root zone (z > 1.2 m) are likely in a quasi-steady state. Higher water contents in the first 0.0- to 0.5-m depths represent an increase in soil water storage during the fallow season before sampling.



View larger version (52K):
[in this window]
[in a new window]
  		Fig. 4. Spatial distribution of water contents with respect to layering along the intensive transect. Major layer interfaces (black lines) are superimposed.

 


View larger version (48K):
[in this window]
[in a new window]
  		Fig. 5. Spatial distribution of water contents with respect to layering along the perpendicular transect. Major layer interfaces (black lines) are superimposed.

 
Spatial Distribution of Chloride and Mass Recovery
The spatial distribution of resident soil chloride concentrations and depth moments along the intensive transect is presented in Fig. 6 . Surface elevations were not included in these plots to better compare the depth moments; that is, the magnitude of Ej[z] represents the distance below the soil surface, not the absolute elevation. Although samples were taken as deep as 6.6 m, soil chloride concentrations in Fig. 6 are shown only to depths of 4.0 m because 96% of the recovered chloride occurs in the first 4 m. The intensive transect average mass recovery as given by Dyck et al. (2003) was 97% (0.15 kg m–2) but varied between 76 to 122% for individual depth breakthrough curves (Fig. 6). Mean travel depth of individual DBTCs ranged from 1.53 to 1.83 m. Locations between 6.2 and 10.0 m had lower mean travel depths than the first 6.2 m of the transect. The transect average E[z] was 1.68 m (Fig. 6). Average modal transport depth (i.e., peak concentration depth) was 1.34 m and ranged from 1.14 to 1.56 m for individual DBTCs. The transect average V[z] was 0.71 m2 and ranged from 0.48 to 0.87 m2 for individual DBTCs.



View larger version (54K):
[in this window]
[in a new window]
  		Fig. 6. Spatial distribution of: (a) resident soil chloride concentrations with respect to layering; (b) mass recovery (Mj); (c) mean travel (Ej[z]) and modal transport depth; and (d) variance (Vj[z]) of the intensive transect. Mean travel depth is represented by triangles, and modal transport depth is represented by circles. The dashed reference lines represent the transect average value of that depth moment.

 
Resident soil chloride concentrations for the perpendicular transect are presented in Fig. 7 . The two-dimensional mass recovery (M2D calculated by Eq. [4]) was 1.55 kg m–1 and when compared to the Cl application rate of 0.92 kg m–1 results in a mass recovery of 168%. Mass recovery for individual DBTCs ranged from 12 to 115% (Fig. 7b). Mean travel depth (Ej[z]) and Vj[z] were not calculated for positions where KCl was not applied to the surface. Mean travel depth (Ej[z]) and Vj[z] for the perpendicular transect (at locations where they were calculated) were in the same range the DBTCs from the intensive transect.



View larger version (46K):
[in this window]
[in a new window]
  		Fig. 7. Spatial distribution of: (a) resident soil chloride concentrations with respect to layering; (b) mass recovery (Mj); (c) mean travel depth (Ej[z]); and (d) variance (Vj[z]) of the perpendicular transect. The dashed vertical lines represent plot boundaries shown in Fig. 1. The dotted lines on the mass recovery plot (b) represent estimated background mass of chloride and the calculated mass distribution following potash (KCl) applications in 1966 and 1971.

 
Central Moments and Layer Dimensions
There were significant (P < 0.05) linear correlations between Ej[z] and Vj[z], and modal transport depth and Vj[z] (Table 1). Mean travel depth is positively correlated to Vj[z] (r = 0.73; P < 0.01), but modal transport depth (Fig. 6c) is negatively correlated to Vj[z] (r = –0.37; P < 0.01), and not significantly correlated to Ej[z]. Mean travel depth had a coefficient of variation of 4% and modal transport depth has a coefficient of variation of 5%.

Several of the correlations between central moments and depth to and thickness of layers were also significant. Mean travel depth was negatively correlated to both the depth to the top of and thickness of the varved layer (r = –0.38; P < 0.01,–0.64; P < 0.01, respectively). Modal transport depth was positively correlated to the depth to the top of the varved layer (r = 0.50; P < 0.01). Transport variance also showed a significant negative correlation with the depth to the top of, and thickness of the varved layer (r = –0.50; P < 0.01, r = –0.44; P < 0.01). Negative correlations between depth to the top of the silt layer and Ej[z], and Vj[z] (r = –0.66; P < 0.01, –0.52; P < 0.01, respectively) are also observed (Table 1).


   DISCUSSION
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
CONCLUSIONS
 REFERENCES
 
The relatively low spatial variability of the modal transport depth indicates that convective fluxes along the intensive transect are relatively uniform. This conclusion is reasonable because: (i) significant variability in precipitation over a length of 10 m is unlikely; (ii) the spatial variability in run-off/run-on is unlikely given the level topography; (iii) the A, B, and C horizons are relatively homogeneous across the length of the transect (Dyck et al., 2003); and (iv) despite the trend in Ej[z] (Fig. 6c), the coefficient of variation in Ej[z] is only 4% (Dyck et al., 2003). Therefore, the trends in Ej[z], Vj[z], and the covariance between Ej[z] and Vj[z] are not likely a result of spatial variations in surface water input, but rather caused by variations in deep soil layer dimensions (particularly the varved layer) and their inherent water contents (i.e., transport volume). For example, the negative correlation between the varved layer dimensions and moments can be explained by the abrupt change in water transport volume at the top of the varved layer. The varved layer has a significantly higher soil water content, and, presumably, transport volume. Thus, when the leading edge of the chloride DBTC encounters the varved layer (with a greater transport volume) the velocity of the chloride would decrease, explaining the positive correlation between modal transport depth and depth to the top of the varved layer (Fig. 8) . In addition, a thick varved layer (i.e., thick layer of high transport volume) would subsequently further decrease the net downward displacement of the leading edge of the chloride DBTC over a given time (in this case 34 yr) relative to a thin varved layer. Figure 6a confirms, visually, that there has been less net downward movment of the leading edge of the chloride DBTC in locations where the varved layer is thicker, explaining the negative correlation between varved layer thickness and Vj[z] (Fig. 9) . From Eq. [2], higher chloride concentrations at greater depths (i.e., large z) will result in a greater Ej[z], further suggesting that the positive correlation between Ej[z] and Vj[z] (Fig. 10) is more a result of the spatial variability of the varved layer dimensions (and therefore transport volume), and not spatial variability in deep drainage. The convergence of Ej[z] and modal transport depth in locations where the varved layer is thicker (Fig. 6) also suggests that the influence of the varved layer on Vj[z] is responsible for variations in Ej[z], not variations in soil water balance.



View larger version (14K):
[in this window]
[in a new window]
  		Fig. 8. Relationship between modal transport depth and depth to the varved layer.

 


View larger version (14K):
[in this window]
[in a new window]
  		Fig. 9. Relationship between varved layer thickness and variance (Vj[z]) for the intensive transect.

 


View larger version (13K):
[in this window]
[in a new window]
  		Fig. 10. Relationship between mean travel depth (Ej[z]) and variance (Vj[z]) for the intensive transect.

 
Even though background chloride concentrations were subtracted from all samples, there appears to be a considerable amount of chloride outside of the original KCl plot boundaries (Fig. 7). Possible explanations for high mass recoveries outside of KCl plot boundaries are: (i) there is considerable spatial variability in background chloride levels and the core that was taken to assess background concentrations is not representative of this area of the landscape (Fig. 1) and (ii) lateral transport of chloride from adjacent chloride treatments could result in chloride concentrations higher than background. The background chloride levels underneath the intensive and perpendicular transects were not assessed before KCl application making it difficult to infer the exact background concentrations from the present data. Subtracting the mass outside of the KCl plots (Fig. 7), M2D was reduced from 168 to 97%. The average mass recovery of 98% reported for the intensive transect in this paper and in a previous paper (Dyck et al., 2003), assumed that the background core accurately reflected the background chloride concentrations. The data here suggest it is more likely that the background core underestimated the background chloride concentrations, and the average mass recovery of the intensive transect is likely closer to 70% because of lateral two-dimensional transport (discussed below).

There is evidence to show that the chloride mass outside of the KCl plots is representative of background chloride levels. Hayashi et al. (1998) reported chloride concentrations in precipitation of 0.041 mg L–1 for a catchment approximately 100-km east of this site. The long-term average precipitation at this site is 321 mm yr–1 (Dyck et al., 2003). Assuming the 0.041 mg L–1 chloride concentration holds for this site, then annual meteoric chloride deposition is approximately 13 mg m–2 yr–1. Assuming this deposition rate has been relatively constant since parent material deposition (10–15 ka bp), then the background soil chloride concentrations are estimated to be 0.13–0.20 kg m–2, very close to the chloride mass outside the KCl plots (Fig. 7b). Furthermore, it is unlikely that weathering of soil minerals has contributed to background chloride levels. The glacial sediments are dominated by silicate and carbonate minerals which are low in chloride. Hayashi et al. (1998) extracted naturally occurring soil chloride in similar sediments with deionized water by shaking vigorously for 4 h. The total mass of soil chloride agreed well with the chloride flux entering the soil through precipitation over time.

Assuming the chloride mass outside of the KCl plots is representative of the background Cl mass, the dotted line in Fig. 7 shows the calculated initial chloride mass distribution after the two KCl applications (1966 and 1971). It is apparent that there has been considerable lateral mass flow downslope from west to east (Fig. 7). The lateral displacement of the chloride (E[x], using Eq. [5]) is 0.76 m. The fact that two chloride applications occurred in the past (Fig. 7) complicates assessment of an average lateral velocity. However, historical data of the vertical chloride transport at this site (Dyck et al., 2003) suggests that the transport of the chloride through the root zone was quick (about 4 yr) and that root extraction of water has merged the two chloride pulses making them indistinguishable. Once the chloride has moved below the root zone, quasi steady-state transport appears to take over. Therefore, the most conservative estimate of net, lateral, chloride transport velocity is 25 mm yr–1 (i.e., 0.76-m mean travel distance over 30 yr), more than twice as large as the net, downward vertical transport velocity estimate of 11 mm yr–1 (Dyck et al., 2003).

The land surface along the perpendicular transect slopes from west to east (Fig. 3), but the total relief is only 0.4 m, making surface redistribution of water an unlikely reason for the high magnitude horizontal flux. The change in total relief along the varved–silt layer interface where most of the lateral flow appears to have occurred, however, is 1.7 m. The alternating clay and silt beds of the varved layer dip in the same direction as the varved–silt layer interface. Therefore, the two-dimensional flow (and likely three-dimensional flow) can be explained by the anisotropic nature of the varved layer. As mentioned in the introduction, anisotropy is caused by microscale pore structure and soil layering (Raats et al., 2004; Zhang et al., 2003). Within the varved layer at this site, pore structure likely contributes to anisotropy in terms of discontinuous pores at the boundaries between clay and silt layers. The alternating silt and clay beds within the varved layer contribute to anisotropy at the profile scale because they have unique hydraulic conductivity curves. As water and solute move vertically below the root zone, textural discontinuities within the varved layer and the varved-silt layer interface are encountered. Because the pore space within the varved layer is more continuous parallel to the dip of the clay and silt beds within the varved layer than perpendicular to the dip, (i.e., greater hydraulic conductivity parallel to the dip), water flow and transport is two dimensional with a strong horizontal component (and likely three dimensional).

The two-dimensional flow also helps to explain differences in spatial chloride distribution between the perpendicular and intensive transects. For the intensive transect (Fig. 6), the leading edge of the chloride DBTC is shallower in locations where the varved layer is thicker (5.0–10.0-m locations), but for the perpendicular transect, the leading edge of the chloride DBTC is deeper in locations where the varved layer is thicker (–3.0- to 9.0-m locations). Since two-dimensional flow seems to be most influenced by the dip of the varved–silt layer interface (Fig. 7), one possible explanation for the difference between the two transects is that the west to east dip of the varved–silt layer interface is spatially variable. To explore this possibility, a conceptual model of the three-dimensional topography of the varved–silt layer interface was developed and is presented in Fig. 11 . Although it is likely that the top of the varved layer has some influence on transport, it appears from Fig. 2 and 3 that the topography of the top of this layer is much less variable in the locations where chloride was applied, and the topography of the varved–silt layer interface will be the focus of discussion. As Fig. 11 shows, the deeper chloride movement between the 0.0- and 5.0-m spatial locations on the intensive transect (Fig. 6), and 3.0- to 9.0-m locations on the perpendicular transect (Fig. 7) may be explained by the steeply dipping varved–silt layer interface in these locations; that is, the vertical component of flow in the varved layer is greater when the varved layer dips downward from west to east. It follows that a possible explanation for the shallower movement of the chloride between the 5.0- and 10.0-m spatial locations on the intensive transect is that the varved–silt layer boundary dips less at these locations. This would result in the clay and silt beds of the varved layer to be essentially horizontal, impeding vertical flow, and also reducing gravitational gradients driving the two-dimensional flow along layer interfaces.



View larger version (48K):
[in this window]
[in a new window]
  		Fig. 11. Proposed conceptual model of the three-dimensional surface of the varved–silt layer interface.

 

   CONCLUSIONS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 CONCLUSIONS
REFERENCES
 
This study indicates a significant influence of layers on convective and dispersive solute fluxes in a low flow, semiarid environment under natural conditions. In this case, the anisotropic varved layer, significantly contributed to two-dimensional and likely three-dimensional transport. This study complements other observations that layers contribute to anisotropy of the soil profile causing lateral flow at layer boundaries and affirms the relevance of high flux, experiments for studying transport processes in the field (e.g., Butters et al., 1989; van Wesenbeeck and Kachanoski, 1994; Ellsworth et al., 1991). The very low fluxes encountered in semiarid environments require intensively sampled, long-term experiments to actually measure the influence of layers on solute transport. It is also clear that more research about the physics of water flow and solute transport at layer boundaries needs to be performed.


   ACKNOWLEDGMENTS
 
Funding for the project was provided by the Natural Sciences and Engineering Research Council of Canada and the Saskatchewan Wheat Pool. The technical contributions of Sid Farkas, Angela Taylor, Robin Weseen, Kim Weinbender, Jaime Hogan, and Tara Fazakas are greatly appreciated. Special thanks go to Gordon and Kathleen Carr for the use of their land. The original work of the late Archie Ballantyne made this project possible.


   REFERENCES
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 CONCLUSIONS
 REFERENCES
 




This article has been cited by other articles:


Home page
 Vadose Zone JHome page
A. M. Tarquis, N. R. A. Bird, A. P. Whitmore, M. C. Cartagena, and Y. Pachepsky
Multiscale Entropy-based Analysis of Soil Transect Data
Vadose Zone J., May 27, 2008; 7(2): 563 - 569.
[Abstract] [Full Text] [PDF]

Home page
 Vadose Zone JHome page
S. A. Woods, R. G. Kachanoski, and M. F. Dyck
Long-Term Solute Transport under Semi-Arid Conditions: Pedon to Field Scale
Vadose Zone J., March 8, 2006; 5(1): 365 - 376.
[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 (4)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Dyck, M. F.
Right arrow Articles by de Jong, E.
Right arrow Search for Related Content
PubMed
Right arrow Articles by Dyck, M. F.
Right arrow Articles by de Jong, E.
GeoRef
Right arrow GeoRef Citation
Agricola
Right arrow Articles by Dyck, M. F.
Right arrow Articles by de Jong, E.
Related Collections
Right arrow Soil Physics
Right arrow Vadose Zone Processes and Chemical Transport

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