HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Abstract Figures Only Full Text (PDF) Free Alert me when this article is cited Alert me if a correction is posted Services Similar articles in this journal Similar articles in ISI Web of Science Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via Google Scholar Google Scholar Articles by Hamlen, C. J. Articles by Kachanoski, R. G. Search for Related Content PubMed Articles by Hamlen, C. J. Articles by Kachanoski, R. G. GeoRef GeoRef Citation Agricola Articles by Hamlen, C. J. Articles by Kachanoski, R. G. Related Collections Laboratory Column Studies Time Domain Reflectometry, TDR Vadose Zone Processes and Chemical Transport

DIVISION S-1—SOIL PHYSICS

Influence of Initial and Boundary Conditions on Solute Transport through Undisturbed Soil Columns

^a Eastern Cereal and Oilseed Research Centre, Agriculture and Agri-Food Canada, 960 Carling Ave., Ottawa, ON, Canada, K1A 0C6
^b 3-7 University Hall, Univ. of Alberta, Edmonton, AB, Canada, T6G 2J9

^* Corresponding author (gary.kachanoski{at}ualberta.ca).

	ABSTRACT

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION CONCLUSIONS REFERENCES

 
The effect of initial soil water content () and steady and transient water flow boundary conditions (J_w) on the transport of identical solute pulses was examined in four undisturbed soil columns. Solute breakthrough curves (BTCs) and were monitored nondestructively using time domain reflectometry (TDR). Pulse concentrations were predicted from TDR impedances () using either a steady J_w calibration or a transient J_w calibration. The steady J_w calibration, ß_s, was estimated from the steady J_w BTC. The transient calibration, ß(), was estimated from the changes in ^–1() between three different calibrating solutions of Cl^– (1.106, 1.503, and 1.900 g L^–1) in the pore space. All ß() curves had a very similar decreasing curvilinear shape. At one location the ß() curve was high and residual water was not completely replaced by the calibrating solutions. The steady J_w calibration values were generally slightly lower than the equivalent ß() value, but collectively they followed a similar decreasing curvilinear trend with steady _s. Solute applied to initially dry soil tended to lead the transient J_w wetting front and breakthrough began with the start of cumulative drainage. These BTCs also had the greatest amount of BTC tailing. Solute applied to initially wet soil had less BTC tailing. The least amount of BTC spreading occurred during the steady J_w condition. The two transient J_w experiments had very similar distributions of solute with depth, despite the contrasting initial conditions. The pores actively transporting solute appear to be determined by the J_w boundary condition.

Abbreviations: BTC, breakthrough curve • EC, electrical conductivity • J_w, water flux • MPA, mass per unit area • pdf, probability density function • RPM, revolutions per minute • subscript [s], steady • TDR, time domain reflectometry • , water content • , soil water potential

	INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION CONCLUSIONS REFERENCES

 
THE MOVEMENT OF water and aqueous chemicals through soils is a complex process. The distribution of pore sizes in the soil and the boundary conditions of the process have a significant influence on transport of water and aqueous chemicals. In a forest soil 73% of saturated water flux occurred through macropores occupying only 0.04% of the soil volume (Watson and Luxmoore, 1986). In a fine sand, 48% of saturated water flux occurred in pores that became operational when water was supplied at a potential of –4 to 0 cm (Clothier and White, 1981). Flow through these highly conductive pores depends on the initial and boundary conditions of the transport process.

Some studies have specifically considered the effect of experimental boundary conditions or initial soil conditions on the transport process: A field study investigating the occurrence of preferential flow in a loamy sand used 10 cm of water applied as either a single ponding event or as 5 d of sprinkler irrigation to transport a pulse of surface applied dye (Ghodrati and Jury, 1990). The dye leached to a shallower depth under ponded water compared with under sprinkler irrigation, despite the expected rapid transport through larger pores when water was ponded. Another study, also investigating the occurrence of preferential flow on a selection of soils, applied dye to soils that were initially at a "wet" and a "dry" soil water content (Flury et al., 1994). The dye either moved deeper in the wet sites or showed no difference attributed to initial soil water content. The combination of pore domain solute initially enters and the pore domain transporting leaching water will define the solute transport behavior.

The influence of initial and boundary conditions is relevant for transport experiments that out of necessity use transient high flow boundary conditions to simulate steady low flow boundary conditions (i.e., van Wesenbeeck and Kachanoski, 1994). It is assumed that the infiltrating water will more or less replace existing water within the soil, and the corresponding steady boundary conditions can be approximated from an average of the transient conditions. Similarly, it is suggested a single measured cumulative drainage probability density function (pdf), the probabilistic model appropriate for transient water flow conditions (Jury and Roth, 1990), can be used to predict transport for other water flow conditions by adjusting the pdf to the new cumulative drainage regime and taking into account any differences in the initial soil water content, . This transformation is permissible when the wetted pore space geometry remains relatively constant, but acceptable variations in the wetted pore space have not been assessed.

The objective of this study was to examine the effect of initial and boundary conditions on the transport of Cl^– through undisturbed soil columns. Three unsaturated pulse transport experiments were conducted: Two experiments had transient water flow boundary conditions and one had a steady water flow boundary condition, but in all cases the cumulative infiltration over each 3-d period was constant. In one transient experiment, the Cl^– pulse was applied when the initial was at a minimum. In the other transient experiment the Cl^– pulse was applied when was at a maximum. In the steady water flow experiment, remained between the minimum and maximum transient values.

Solute transport subject to these initial and boundary conditions was measured nondestructively at the same locations in the same undisturbed soil columns using TDR. Thus, an additional objective was to calibrate TDR for pore water Cl^– concentration during transient conditions. An assessment of this calibration was made with respect to the accepted steady calibration used for pulse transport experiments.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION CONCLUSIONS REFERENCES

 
The experiments were conducted on four columns of undisturbed Fox Sand (Typic Hapludalf). The columns (15-cm i.d. and 160 cm long) were collected at 40-cm intervals from a field at the Agriculture Canada Research Station in Delhi, ON. The soil at this site is characterized by spatially periodic tonguing of the B horizon into the C horizon, and a discontinuous Bt horizon (van Wesenbeeck and Kachanoski, 1994).

The columns were fitted onto a vacuum assembly (1.5-m air-entry pressure head) for unsaturated transport experiments (Ward et al., 1994), and pressure at the lower boundary was maintained at approximately –20 cm of water. The columns were instrumented at 10-cm depth intervals with two-wire TDR probes (2.0-mm o.d. and 148.0-mm long rods, spaced 20 mm apart) and 0.1 MPa (1 bar) ceramic cup (5.99 mm o.d. and 80 mm long) tensiometers (Soilmoisture Equipment Corp., Santa Barbara, CA) equipped with manometers (7-mm i.d. acrylic tubing). The tensiometers were installed 25 mm to the right of the TDR probes, beyond the estimated 21.0-mm diameter of influence of the TDR probes (Ward et al., 1995).

Soil dielectric constants (or relative permittivities), , were estimated from the speed of the electromagnetic signal propagating along the TDR probes, and soil water contents, (cm³ cm^–3), were calculated according to Topp et al. (1980).

[1]

Time domain reflectometry impedance values, (ohms), were collected at a fixed distance after multiple reflections had ceased. The values are assumed to represent direct current (d.c.) resistances (i.e., Nadler et al., 1991; Wraith et al., 1993) and are linearly related to pore water electrolyte concentration. Soil water potentials, (cm), were the heights of water in the manometer tubes relative to the tensiometer cups.

The surface flux densities of water, J_w (cm d^–1), chosen for the transient and steady water flow conditions gave equivalent cumulative infiltration over each 3-d period. The transient condition used a repeating cycle of J_w = 9.0 cm d^–1 for 1.0 d followed by J_w = 0 for 2.0 d. Figure 1 shows and measured at the 10-cm depth of Column 3 during the transient water flow condition (the first two cycles of the transient transport experiments and all transient calibration cycles, presented later in this section). No ponding at the soil surface was observed during J_w = 9.0 cm d^–1 and internal drainage was minimal at the end of 2 d of J_w = 0. The steady water flow condition used a continuous J_w = 3.0 cm d^–1. In all cases, the fluxing water contained a low concentration of CaCl₂ (0.475 g L^–1 of Cl^–).

View larger version (27K): [in this window] [in a new window]   Fig. 1. Soil (a) water potential and soil (b) water content measured at the 10-cm depth of Column 3 during the transient Jw cycles of the transport experiments and the calibration procedure. The data spans a period of approximately 6 mo. Initial conditions of the transient experiments are indicated (dry and wet).

The curves in Fig. 1 show greater variability. Soil dielectric constants were obtained from manual measurements of travel time taken from the oscilloscope screen. The oscilloscope cursor was positioned at the beginning and end of the waveguides, as indicated by impedance mismatches. The variability is due to more than one operator collecting these readings during each wetting-draining cycle. In circumstances where J_w is precisely set and (t) remains constant, an average (t) can be estimated for the wetting and draining cycle. The averaging of (t) for this work will be presented later.

The initial conditions for the two transient experiments are indicated in Fig. 1. In one experiment the pulse of Cl^– was applied when was a minimum, at the end of 2 d of J_w = 0 or just before J_w = 9.0 cm d^–1. This experiment is referred to as the dry transient experiment. In the other transient experiment the pulse of Cl^– was applied when was a maximum, at the end of 1 d of J_w = 9.0 cm d^–1. This second experiment is referred to as the wet transient experiment. At this same probe location, the steady J_w = 3.0 cm d^–1 experiment had a constant _s = –10.5 cm and a constant _s = 0.22 cm³ cm^–3 (steady and are indicated by subscript ‘s’). This () point corresponded to the draining portion of the transient data shown in Fig. 1 rather than the wetting portion. The _s(_s) points generally were associated with the draining portion of the transient () curves, and it is felt that the tensiometers did not always have time to completely equilibrate to the rapidly changing hydraulic conditions at the wetting front. Very consistent timing of and measurements incorporates this error into the monitoring protocol of the calibrations and experiments, and it is not given any special consideration.

Figure 2 shows the minimum and maximum transient values and the _s value from eight measurement depths (10–80 cm) in the four columns. Figure 2 also shows the approximate locations of the A-B horizon boundary and the B-C horizon boundary. Column 1 had the most uniform distributions of with depth, with very little difference in values between the A and B horizons. The other columns show more variable distributions of , especially in the B horizon. Column 3 had a very high _s and maximum transient at the 40-cm depth in the B horizon, followed by a sharp decrease in the values at the 50-cm depth in the Ck horizon. A very thin, discontinuous Bt horizon above the Ck has been observed in the Fox Sand very close to where these columns were collected (van Wesenbeeck and Kachanoski, 1994), and a Bt horizon appears to be present at the 40-cm depth of Column 3. Columns 2 and 4 also show an increase in the maximum transient value in the B horizon (at the 50- and 40-cm depth, respectively), but the minimum transient values of these locations do not remain high and clay accumulation may not be as high as in Column 3.

View larger version (34K): [in this window] [in a new window]   Fig. 2. Soil water contents measured in the four columns. Minimum values (triangles) were measured after 2 d of Jw = 0, maximum values (circles) were measured after 1 d of Jw = 9.0 cm d–1, and midpoint values (diamonds) were measured during continuous Jw = 3.0 cm d–1. Note that horizontal lines correspond to approximate A-B (upper) and B-C (lower) horizon boundaries.

Two different calibrations of ^–1 for Cl^– concentration were obtained for each measurement depth in the columns (each TDR probe), a steady and a transient calibration. The steady calibration followed the procedure described by Ward et al. (1994): The pulse input BTC of ^–1(t), corrected for the initial ^–1 value, was numerically convoluted to estimate the step input BTC, ^–1_step(t). The steady calibration coefficient, ß_s ( g L^–1), was estimated from the pulse concentration, C_o = 130 g L^–1, and the t value of step input BTC,

[2]

Because the pulse input was not a true Dirac delta function, convolution involved superposition in time of the ^–1(t) BTC lagged by t. The t value of the step input BTC was inversely proportional to t, so it was critical to use a value of t that was consistent with the boundary condition of the steady water flow experiment. An effective t was estimated such that the surface flux density of solute (solution volume, V, over soil surface area, A) was equal to J_w,

[3]

The effective t was approximately 0.019 d rather than the true input time of 0.0007 d (1 min). Values were calculated for each column because there were small but consistent differences in the flow of each water line.

The transient calibration had to account for the effect of both and pore water Cl^– concentration on ^–1 values (i.e., Dalton et al., 1984) and involved a different calibration procedure: The approach was to completely replace all fluid in the pore space with a solution of known Cl^– concentration and measure and over the transient J_w cycle. This was repeated for three calibrating solutions of Cl^–: 1.106, 1.503, and 1.900 g L^–1. The transient calibration coefficient, ß(), was estimated from

[4]

It is essential to have complete pore water replacement by each calibrating solution (Mallants et al., 1996; Nissen et al., 2000), and at least 90.0 cm of each solution was flushed through the columns before collecting the transient calibration data. This amount was equivalent to between 5.3 and 6.9 pore volumes, based on the maximum transient of each measurement depth in the columns.

Resident concentration BTCs were estimated from the ^–1 BTCs, corrected for the initial ^–1 values

[5a]

[5b]

Equation [5a] is used for the transient J_w experiments and Eq. [5b] is used for the steady J_w experiment. The choice of ß() was based on the transient value of . Although the curves show variability due to manual measurements of travel time from the oscilloscope screen, it was desirable to eliminate the effect of hysteresis and use as the independent variable rather than . To minimize errors due to variability in , an average wetting cycle (t) and an average draining cycle (t) were calculated using all (t) measurements. As a test of this averaging procedure, unit normal deviates of the residuals (Draper and Smith, 1981) were calculated for each measurement location. Figure 3 shows the values from the 10-cm depth of Column 3 roughly fall within two standard deviations (±2), 95% of the time. The unit normal deviates of the residuals for all measurement locations were similarly distributed (not presented).

View larger version (22K): [in this window] [in a new window]   Fig. 3. Unit normal deviates of the averaged (t) residuals from the dry and wet transport experiments at the 10-cm depth of Column 3. Roughly 95% of the values fall between the limits (–2,2).

The cumulative drainage BTCs have been summarized in terms of two I_z(t) values: the value associated with the leading edge of the BTC and the value associated with the peak of the BTC. The cumulative drainage BTCs have also been summarized in terms of the spreading of the pulse. To accommodate a variety of BTC shapes, the spread of each BTC is given by the variance of the cumulative drainage pdf, (I_z) (cm³ cm^–2) (Jury and Roth, 1990),

[6]

where C(I_z) is the resident concentration measured at depth z. The cumulative drainage pdfs are parameterized by the expected value, E[I_z] (cm³ cm^–2), and variance, VAR[I_z] ([cm³ cm^–2]²),

[7]

[8]

Jury and Roth (1990) suggest (I_z) is invariant under different transient flows as long as the wetted pore space geometry does not significantly change. A comparison of the VAR[I_z] values from the three transport experiments is used to test the invariance of (I_z).

As a check on the success of using only TDR measurements to estimate pore water concentrations during transient conditions, the mass per unit area (MPA) of the applied pulse (0.00735 g cm^–2 Cl^–) is predicted from the depth distribution of C and during each of the pulse transport experiments,

[9]

Equation [9] was numerically solved using C(z,t) values and averaged (z,t) values, interpolated to 2.5-cm intervals.

	RESULTS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION CONCLUSIONS REFERENCES

 
Calibration of ^–1 for Pore Water Concentration
Figure 4 shows and ^–1 measured at the 10-cm depth of Column 3 during the transient calibration. The (and ) curves were previously shown in Fig. 1 by the dashed lines. The ^–1 curves show a distinct separation due to the different pore water Cl^– concentrations, with 1.900 g L^–1 giving the highest ^–1 over the wetting-draining cycle. Each ^–1 curve follows the shape of the transient curves. Slight changes in ^–1 during the maximum period (between approximately 0.25 and 1.0 d) do occur. These are associated with very slight changes in (and Cl^– content) due to wear of the flow-rated pump tubes and/or adjustments to the pump RPM to maintain the constant J_w = 9.0 cm d^–1. They are real changes in ^–1 values rather than variations due to manually positioning the TDR cursor on the oscilloscope screen at impedance mismatches, as was done for measurements.

View larger version (22K): [in this window] [in a new window]   Fig. 4. (a) Soil water content and (b) TDR impedance values measured at the 10-cm depth of Column 3 during the transient Jw calibration. Three calibrating solutions of Cl– were used, 1.1 g L–1 (diamonds), 1.5 g L–1 (squares), and 1.9 g L–1 (triangles).

[10]

where j identifies the calibrating solution (j = 1,2,3). Equation [10] gives a continuous response in ^–1 to a continuous change in from the discrete values collected during the wetting-draining cycle. It was adapted from a model relating bulk soil electrical conductivity (EC) to and pore water EC (Eq. [5], Rhoades et al., 1976) and is only considered to be an empirical description. The regression procedure for each TDR probe used the data from all three calibrating solutions, blocked in such a way that a single intercept (a) and unique coefficients on (b_j and c_j) were obtained. The F test statistic from this model (not presented) was significant for the 32 measurement locations. A linear relationship between ^–1 and was also estimated, but for 28 of the 32 measurement locations a partial F test indicated Eq. [10] was statistically the better model at the 95% confidence level.

View larger version (22K): [in this window] [in a new window]   Fig. 5. (a) The relationship between –1 and for the three calibrating solutions of Cl– and (b) the transient calibration coefficients calculated using the best fits of Eq. [10] from the 10-cm depth of Column 3.

Figure 6 shows the ß() curves from all eight measurement depths in Column 3. There is surprising consistency in the decreasing curvilinear shape of the ß() curves that was also seen among all 32 measurement locations (not presented). The curves in Fig. 6 are identified by horizon rather than depth to show the negligible dependence of ß() on soil horizonation for these undisturbed Fox Sand columns. The ß() curve measured at the 80-cm depth in Column 3 is the only ß() curve that does not follow the general trend. It seemingly indicates pore water replacement by each calibrating solution was incomplete at the 80-cm depth of this column (progressively increasing uniform concentrations were used during the transient calibration). The thin Bt horizon near the 40-cm depth (suggested in Fig. 2) likely disrupted the vertical movement of water and solute in such a way that more pore volumes were needed to completely displace water and solute initially in the pore space of the Ck horizon. Also, the high minimum and maximum transient at the 40-cm depth means this column was flushed with the lowest number of pore volumes of the calibrating solution before collecting the transient calibration data.

View larger version (22K): [in this window] [in a new window]   Fig. 6. Transient Jw calibration coefficients from the eight measurements depths in Column 3 presented as a (a) function of transient , and (b) all steady Jw calibration coefficients from the four columns presented as a function of steady .

Table 1 lists ß_s and the value of the transient coefficient at _s, ß(_s), from all measurement locations. Only 10 of the ß(_s) values are within 5% of ß_s. ß(_s) is higher at 26 of the 32 measurement locations. Of the remaining six measurement locations, four have unusually high ß_s values (identified as outliers in Fig. 6b). In Column 1 where values were most uniform with depth and idealized piston-like flow of water and solute was expected, most ß(_s) were higher than ß_s.

View this table: [in this window] [in a new window]   Table 1. Steady Jw and transient Jw estimates of the calibration coefficient, ß, associated with the steady soil water content, s.

Several ß_s values were higher than ß(_s) values (Table 1), and most occurred in Column 2 (outliers identified in Fig. 6). Hydrodynamic differences between the calibration methods in this column may explain these unusual trends. At _s the same pores were water-filled during the two calibration procedures. During the transient calibration, each calibrating solution was assumed to be uniformly distributed across the wetted pore space (excluding the 80-cm depth seen in Fig. 6). During the steady calibration, however, it is unlikely the transporting pulse was as uniformly distributed across the wetted pore space since preferential flow paths are considered the norm rather than the exception (i.e., van Wesenbeeck and Kachanoski, 1994). Any errors measuring representative amounts of the preferential flow paths will contribute to errors in ß_s values.

Figure 7 shows the steady J_w BTCs from Column 2. Pore water Cl^– concentrations were calculated using ß_s. The BTC at the 30-cm depth is lagged in a manner unexpected for the distribution of _s near the surface (Fig. 2). Also, there is essentially identical BTC tailing at the 10- and 20-cm depths after 2 d. These deviations from expected piston-like flow of water and solute are attributed to preferential flow paths that were not detected in representative amounts by the TDR probes, and Nissen et al. (2000) report similar discrepancies with TDR monitored BTCs. Erroneously low ^–1(t) BTCs would result in higher estimates of ß_s when using C_o = 130 g L^–1 with Eq. [2].

View larger version (20K): [in this window] [in a new window]   Fig. 7. Breakthrough curves measured in Column 2 during the steady Jw = 3.0 cm d–1 transport experiment. Concentrations (g L–1) are predicted with ßs.

Pulse Breakthrough Curves
Concentration BTCs plotted as a function of time for the wet and dry transient J_w experiments are not shown. In general, there were minimal changes in concentration during the draining periods and concentration breakthrough was detected only after the start of each transient experiment's first wetting cycle (first J_w = 9.0 cm d^–1). Figure 8 shows the steady and transient J_w BTCs from the 30- and 40-cm depths of Column 3, C(I₃₀) and C(I₄₀). These curves show excellent BTC detail and early breakthrough characteristics are easily captured. These BTCs also show the disruptive effect of the Bt horizon near the 40-cm depth.

View larger version (26K): [in this window] [in a new window]   Fig. 8. Cumulative drainage breakthrough curves measured at the 30- and 40-cm depths in Column 3. Concentrations (g L–1) for the steady Jw = 3.0 cm d–1 transport experiment (diamonds) were predicted with ßs and concentrations for the dry (triangles) and wet (circles) transient Jw experiments were predicted with ß().

View larger version (30K): [in this window] [in a new window]   Fig. 9. Cumulative drainage (cm3 cm–2) associated with the leading edge of the breakthrough curves measured in all columns during the steady Jw = 3.0 cm d–1 transport experiment (diamonds), the dry transient Jw experiment (triangles), and the wet transient Jw experiment (circles).

View larger version (30K): [in this window] [in a new window]   Fig. 10. Cumulative drainage (cm3 cm–2) associated with the peak concentration of the breakthrough curves measured in all columns during the steady Jw = 3.0 cm d–1 transport experiment (diamonds), the dry transient Jw experiment (triangles), and the wet transient Jw experiment (circles).

View larger version (31K): [in this window] [in a new window]   Fig. 11. Variances [(cm3 cm–2)2] of the cumulative drainage probability density functions (pdfs) from all columns, representing the spread of the breakthrough curves measured during the steady Jw = 3.0 cm d–1 transport experiment (diamonds), the dry transient Jw experiment (triangles), and the wet transient Jw experiment (circles).

Similar to Fig. 9, Fig. 10 shows the dry transient experiment has the smallest I_z values associated with the peak concentration of solute, and the steady and wet transient J_w experiments have larger values of I_z. One exception occurs at the 40-cm depth of Column 3. Both transient C(I₄₀) curves seen in Fig. 8 appear to be missing the expected main peak of the BTC and the information given in Fig. 10 is more representative of the tailing portion of the transient BTCs. This is another example showing that the paired TDR waveguides did not always monitor representative amounts of the transporting pulse.

Figure 11 shows the VAR[I_z] values. At all measurement depths, the dry transient experiment has the greatest BTC spreading and the steady J_w experiment has the smallest. Thus, although portions of the dry transient pulse can immediately be transported to depth (Fig. 9), these dry BTCs have the greatest amount of tailing. The disruptive effect of the Bt horizon at the 40-cm depth of Column 3 can be seen in the increase in steady J_w VAR[I] values below this depth. Additional comments on the column hydraulic responses will be presented in the Discussion section below.

Pulse distributions with depth are shown in Fig. 12 . Early concentration values were chosen for Column 3: after 6.6 cm of applied water for the steady J_w experiment (2.2 d), and after approximately 4.5 cm of applied water for the transient experiments (0.5 d for the dry experiment and 2.5 d for the wet experiment). Later concentration values were chosen for Column 1: after 7.8 cm of applied water for the steady J_w experiment (2.6 d), and after approximately 9.0 cm of applied water for the transient experiments (1.0 d for the dry experiment. and 3.0 d for the wet experiment). In Column 3, the TDR wiring at the 60-cm depth was damaged during the wet transient experiment and the resident concentration in Fig. 12 is erroneously low.

View larger version (30K): [in this window] [in a new window]   Fig. 12. Distributions of resident concentration with depth in all columns during the steady Jw = 3.0 cm d–1 transport experiment (diamonds), the dry transient Jw experiment (triangles), and the wet transient Jw experiment (circles).

Table 2 lists MPA calculated using the data shown in Fig. 12. Also given in Table 2 are percentages relative to the applied MPA of the solute pulses, 0.00735 g cm^–2 Cl^–. The best predictions of MPA were from the steady J_w experiment, while the dry and wet transient experiments overestimated MPA. The worst predictions were in Column 2 where both steady and transient predictions were 154 to 160% higher than applied MPA. Excluding Column 2, the dry transient predictions of MPA were closer than the wet transient predictions.

View this table: [in this window] [in a new window]   Table 2. Mass per unit area of Cl– from the steady Jw transport experiment and the transient Jw (dry and wet) experiments. Values in parentheses are percentages of applied mass per unit area.

	DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION CONCLUSIONS REFERENCES

 
When a pulse of solute was applied to the dry soil of Columns 2, 3, and 4, the solution application rate briefly exceeded the infiltration capacity of the wetted pore space and some of the pulse entered the larger, air-filled pores that would subsequently carry the leading edge of the wetting front. Thus, solute breakthrough near the surface of these columns began as soon as cumulative drainage started. As the wetting front carrying the solute pulse moved deeper it appeared to displace residual water in a manner more consistent with the plane of separation between residual water and infiltrating water (personal communication, D.E. Elrick, 1993), and the pulse began lagging behind cumulative drainage. In Column 3 this did not occur until below the 40-cm depth, in the Ck horizon. It appears this column has a few preferential flow paths that function relatively independently of much of the pore space when J_w = 9.0 cm d^–1. In Column 1, solute did not lead the wetting front during the dry transient experiment, even to the 10-cm depth. As suggested by the lower, more uniform distribution of in Column 1 (Fig. 2), there appears to be a lack of preferential flow paths that function independently and the wetting front displaces residual beginning at or near the surface.

During the wet transient J_w experiment, solute breakthrough did not begin as soon as cumulative drainage started, even at the 10-cm depth. There were only very small increases in ^–1 at the wet experiment's first wetting front (see Fig. 8). Due to the high solute flux density (81.5 cm d^–1), it was unlikely the pulse effectively displaced water in the drainable pore space. During the subsequent 2 d of J_w = 0, the redistribution process also did not concentrate the pulse into pores that carried the leading edge of the wetting front at J_w = 9.0 cm d^–1. Residual water was likely left as coatings on these and other pores.

The relative spreading, or VAR[I_z] values (Fig. 11), between the dry and wet transient experiments is consistent with the idea that the dry transient pulse filled pores near the surface that had emptied during the 2 d of J_w = 0. The conductivity of some of these smaller sized pores was so low and/or the concentration of solute in those pores was so high, they operated like a slow-release source of Cl^– for the dry transient pulse. Diffusion and/or convection of solute out of these pores required more cycles of transient J_w, giving the dry transient BTCs the highest VAR[I_z] values. The redistribution process during the first 2 d of the wet transient experiment did not concentrate the Cl^– pulse into these smaller pores and fewer cycles of transient J_w were needed to displace the pulse.

The steady J_w experiment displayed the smallest VAR[I_z] values (Fig. 11). Again, the high solute flux density of the steady J_w pulse would have been ineffective at displacing water in the pore spaces (wetted at J_w = 3.0 cm d^–1) and the pulse would have spilled into the larger unsaturated pores. The immediate resumption of J_w = 3.0 cm d^–1 would have minimized any redistribution into the smaller pores, thereby minimizing the amount of BTC spreading. Also, diffusion of the pulse out of the large, initially unsaturated pores into the mobile domain would have likely occurred in the most rapidly conducting pore networks and prevented excessive tailing of the BTCs.

The VAR[I_z] is a parameter of each experiment's cumulative drainage probability density function, (I_z), and the values from the transport experiments can be used to test for the invariance of the (I_z) under different initial and boundary conditions. The VAR[I_z] of each column shown in Fig. 11 are not similar between experiments. Thus, (I_z) is not an invariant model of solute transport for these initial and boundary conditions. The pore domain in which solute initially resided, before convective flow begins, had a very significant effect on the resulting (I_z). The experimental boundary conditions, steady versus transient J_w, also significantly affected the resulting (I_z). Solute transport through the sandy soil used in this study was not simply a function of I_z.

Characteristics of transport observed in Column 3 relative to Column 1 provide an example of how the spatial variability of soil properties at the pedon scale affects solute transport. The single peak for the steady and transient C(z) distributions in Column 1 indicate this column does not have many small pores with low conductivities or many large, continuous pores with high conductivities. A much greater proportion of the pore space appears to contribute to transport than for Column 3, and plane of separation describes the displacement of residual in Column 1. In Column 3, there appears to be a greater proportion of small pores with low conductivities. The smaller pores near the surface behave like a slow release source of Cl^–. The larger pores, which operate when J_w = 9.0 cm d^–1, quickly transport portions of the pulse to depth and result in bimodal transient C(z) distributions [the wet transient C(60) value is erroneously low because the TDR wiring was damaged during this experiment]. The largest pores do not operate when J_w = 3.0 cm d^–1 and the steady C(z) distribution is very drawn out rather than bimodal. This reaffirms the suggestion that the boundary condition (i.e., J_w) defines the actively transporting network of pores.

	CONCLUSIONS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION CONCLUSIONS REFERENCES

 
The effect of initial conditions () and boundary conditions (J_w) on the transport of Cl^– was examined in this study. Initial conditions influenced which surface pores a pulse of Cl^– could initially enter. When was initially low, the surface-applied pulse entered a greater proportion of the smaller sized pores. When was initially high, the pulse bypassed a greater proportion of the smaller, low conductivity pores, which remained water-filled. Solute initially residing in small pores acted like a slow release source of chemical, and the resulting solute BTCs had much greater spread (high VAR[I_z]) compared with when these pores were bypassed. An unexpected finding was the very rapid transport that occurred when the pulse was applied to dry soil then immediately wetted at J_w = 9.0 cm d^–1. This translates into a greater risk for rapid transport of chemicals through soil under this and similar conditions.

The J_w boundary condition defined the pore domains that would fill with water and actively transport water and solute. Similar pore domains, subject to similar monitoring protocol by TDR, appeared to transport the two transient pulses (J_w = 9.0 cm d^–1) despite differences caused by the initial conditions. The pore domain transporting the steady J_w pulse (3.0 cm d^–1) was significantly different. Thus, a transient, high J_w condition will not generate solute transport that is characteristic of the equivalent steady J_w flow condition. Transient J_w conditions will give greater BTC spreading (in terms of cumulative drainage) compared with steady J_w conditions.

Mass per unit areas predicted for the steady J_w transport experiment were within 90% of the applied MPA in three of the four columns. In Column 2, the steady J_w MPA prediction was high and reflected the relatively high ß_s values from this column. Detecting representative amounts of the transporting pulse were considered to be the main source of ß_s error. In general, the ß_s values appear to be appropriate for the steady J_w boundary condition.

Mass per unit area predictions for the transient J_w pulses were higher than applied MPA. Some of the ß() curves may have been erroneously high. Incomplete pore water replacement by the calibrating solutions was suspected at one depth, and similar, but smaller, errors in ß() at other depths could contribute to the high MPA predictions. The similarity in the ß() curves validates the transient calibration procedure, but for this work the transient calibration would have benefited from using larger solution volumes to flush residual water out of the pore space.

The nondestructive TDR measurements provided extremely detailed information on solute transport in these undisturbed soil columns. It was possible to assess the effect of initial and boundary conditions on solute transport as well as detect hydrodynamic differences between the soil columns using TDR. The apparent sensitivity of TDR to residual water during the transient calibration indicates it is an ideal tool to investigate complex transport processes such as the spatial distribution of preferential flow paths in soil. Its main limitation for transient applications is the very time consuming calibration procedure that requires complete replacement of residual water in the pore space.

	ACKNOWLEDGMENTS

 
The authors gratefully acknowledge the support from the Natural Sciences and Engineering Research Council (NSERC) of Canada and the technical support of Peter von Bertoldi.

Received for publication September 30, 2002.

	REFERENCES

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION CONCLUSIONS REFERENCES

 

• Clothier, B.E., and I. White. 1981. Measurement of sorptivity and soil water diffusivity in a field. Soil Sci. Soc. Am. J. 45:241–245.[Abstract/Free Full Text]
• Dalton, F.N., W.N. Herkelrath, D.S. Rawlins, and J.D. Rhoades. 1984. Time-domain reflectometry: Simultaneous measurement of soil water content and electrical conductivity with a single probe. Science 224:989–990.
• Draper, N.R., and H. Smith. 1981. Applied regression analysis. 2nd ed. John Wiley & Sons, New York.
• Flury, M., H. Fluhler, W.A. Jury, and J. Leuenberger. 1994. Susceptibility of soils to preferential flow of water: A field study. Water Resour. Res. 30:1945–1954.
• Ghodrati, M., and W.A. Jury. 1990. A field study using dyes to characterize preferential flow of water. Soil Sci. Soc. Am. J. 54:1558–1563.[Abstract/Free Full Text]
• Jury, W.A., and K. Roth. 1990. Transfer functions and solute movement through soils. Birkhauser Verlag Basel, Germany.
• Mallants, D., M. Vanclooster, N. Toride, J. Vanderborght, M.Th. van Genuchten, and J. Feyen. 1996. Comparison of three methods to calibrate TDR for monitoring solute movement in undisturbed soil. Soil Sci. Soc. Am. J. 60:747–754.
• Nadler, A., S. Dasberg, and I. Lapid. 1991. Time domain reflectometry measurements of water content and electrical conductivity of layered soil columns. Soil Sci. Soc. Am. J. 55:938–943.
• Nissen, H.H., P. Moldrup, and R.G. Kachanoski. 2000. Time domain reflectometry measurements of solute transport across a soil layer boundary. Soil Sci. Soc. Am. J. 64:62–74.[Abstract/Free Full Text]
• Rhoades, J.D., P.A.C. Raats, and R.J. Prather. 1976. Effects of liquid-phase electrical conductivity, water content, and surface conductivity on bulk soil electrical conductivity. Soil Sci. Soc. Am. J. 40:651–655.
• Si, B.C. 2002. Spatial and statistical similarities of local soil water fluxes. Soil Sci. Soc. Am. J. 66:753–759.
• Topp, G.C., J.L. Davis, and A.P. Annan. 1980. Electromagnetic determination of soil water content: Measurements in coaxial transmission lines. Water Resour. Res. 16:574–582.
• van Wesenbeeck, I.J., and R.G. Kachanoski. 1994. Effect of variable horizon thickness on solute transport. Soil Sci. Soc. Am. J. 58:1307–1316.[Abstract/Free Full Text]
• Ward, A.L., R.G. Kachanoski, and D.E. Elrick. 1994. Laboratory measurements of solute transport using time domain reflectometry. Soil Sci. Soc. Am. J. 58:1031–1039.[Abstract/Free Full Text]
• Ward, A.L., R.G. Kachanoski, A.P. von Bertoldi, and D.E. Elrick. 1995. Field and undisturbed-column measurements for predicting transport in unsaturated layered soil. Soil Sci. Soc. Am. J. 59:52–59.[Abstract/Free Full Text]
• Watson, K.W., and R.J. Luxmoore. 1986. Estimating macroporosity in a forest watershed by use of a tension infiltrometer. Soil Sci. Soc. Am. J. 50:578–582.[Abstract/Free Full Text]
• Wraith, J.M., S.D. Comfort, B.L. Woodbury, and W.P. Inskeep. 1993. A simplified waveform analysis approach for monitoring solute transport using Time-Domain Reflectometry. Soil Sci. Soc. Am. J. 57:637–642.

This Article Abstract Figures Only Full Text (PDF) Free Alert me when this article is cited Alert me if a correction is posted Services Similar articles in this journal Similar articles in ISI Web of Science Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via Google Scholar Google Scholar Articles by Hamlen, C. J. Articles by Kachanoski, R. G.