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 HighWire Citing Articles via ISI Web of Science (11) Citing Articles via Google Scholar Google Scholar Articles by Vanderborght, J. Articles by Flühler, H. Search for Related Content PubMed Articles by Vanderborght, J. Articles by Flühler, H. GeoRef GeoRef Citation Agricola Articles by Vanderborght, J. Articles by Flühler, H. Related Collections Fate Vadose Zone Processes and Chemical Transport Soil Models

DIVISION S-1—SOIL PHYSICS

Identification of Transport Processes in Soil Cores Using Fluorescent Tracers

Jan Vanderborght^*,a, Paul Gähwiller^,b and Hannes Flühler^b

^a Laboratory of Soil and Water, Katholieke Universiteit Leuven, Vital Decosterstraat 102, B-3000 Leuven, Belgium
^b Soil Physics, Inst. of Terrestrial Ecology, Swiss Federal Institute of Technology, ETHZ, Grabenstrasse 11a, CH-8952 Schlieren, Switzerland

* Corresponding author currently at Inst. of Chemistry and Dynamics of the Geosphere, ICG-IV Agrosphere, Research Center Jülich, D-52425 Jülich, Germany (j.vanderborght{at}fz-juelich.de)

	ABSTRACT

TOP NOTES ABSTRACT INTRODUCTION METHODS RESULTS SUMMARY AND CONCLUSIONS REFERENCES

 
To identify soil properties that control transport of adsorbing solutes in natural soil, we carried out leaching experiments in undisturbed soil cores taken from three soil layers of a Stagni-Humic Cambisol. Breakthrough curves (BTCs) of Cl^- and two adsorbing fluorescent dye tracers, brilliant sulfaflavine (BF; 1H-Benz(de)isoquinoline-5-sulfonic acid, 2,3-dihydro-6-amino-1,3-dioxo-2-(p-tolyl)-, monosodium salt) and sulforhodamine B (SB; xanthylium, 3,6-bis(diethylamino)-9-(2,4-disulfophenyl)-, inner salt, sodium salt), were measured. Three cores were scanned with x-rays to determine the three-dimensional (3-D) structure of large pores. After the leaching experiment, soil cores were horizontally sliced and dye concentration distributions on cross sections were derived from fluorescence signal images. Transport was investigated using BTCs and concentration maps, adsorption isotherms, and predictions by three different transport models: convection dispersion model (CDM), stream tube model (STM) and physical nonequilibrium model (PNEM). The dense network of large pores in the two upper soil layers induced a uniform lateral spreading of dyes and the CDM described the transport fairly well. In cores from the deeper layer, the large pore network was considerably less dense and dye patterns followed closely the few large pores without lateral mixing indicating preferential flow and explaining the fast dye breakthrough. Predictions by the STM revealed that the fast SB breakthrough could not be explained solely by preferential flow. Fitting the PNEM to breakthrough data and the low total dye concentration in the preferential flow region suggested a small sorption capacity of the preferential flow region for SB. Therefore, preferential leaching of dyes resulted from small-scale variations in physical and chemical soil properties.

Abbreviations: BF, brilliant sulfaflavine • BTC, breakthrough curve • CDM, convection dispersion model • CT, computer tomography • EMPA, Eidgenössische Materialprüfungs-und Forschungsanstalt • PNEM, physical nonequilibrium model • SB, sulforhodamine B • STM, stream tube model • 3-D, three dimensional

	INTRODUCTION

TOP NOTES ABSTRACT INTRODUCTION METHODS RESULTS SUMMARY AND CONCLUSIONS REFERENCES

 
THE MOVEMENT OF SURFACE APPLIED AGROCHEMICALS through soils has been an important research topic in soil science during the last few decades because these agrochemicals may contaminate subsurface water. Two processes need to be understood in order to model leaching of solutes through soils. The first is the so-called lateral mixing process (Flühler et al., 1996) that describes how water applied at the soil surface is distributed in the pore space and mixed with the initial soil solution. The second is the adsorption process or the interaction between solutes and the solid soil phase.

The lateral mixing process can be inferred from leaching of water tracers that do not sorb to the soil particles. It depends on the structure of pore space at the microscopic scale and of the soil profile at the pedon scale. In unsaturated soils, pores or larger soil structures that contribute to the leaching process are activated depending on the state, that is, water content, and on boundary conditions, that is, the water flux at the soil surface (Vanderborght et al., 2000). Therefore, a complete understanding of the lateral mixing process, which is indispensable for predictions of leaching, requires linking soil structure, water content, water flux, and the lateral mixing process.

Dye tracers have been used to investigate qualitatively the lateral mixing process in soil cores or field plots by visualizing the spatial structure of the leaching process. Dye tracer studies illustrate the effect of the flow rate and antecedent (before tracer application) water content on leaching and lateral mixing (Ghodrati and Jury, 1990; Flury et al., 1994; Forrer et al., 1999; Perillo et al., 1999). Also the role of soil structural elements such as macropores (Booltink and Bouma, 1991; Wildenschild et al., 1994), size of soil peds (Vervoort et al., 1999), weathered fractures in a clayey till (Jørgensen et al., 1998), and soil layers (Kung, 1990; Koch and Flühler, 1994; Perillo et al., 1999) on lateral mixing has been illustrated using dye tracers. In soil core and field-plot studies, flow paths visualized by dye tracers were used to interpret time series of tracer concentrations observed in the effluent from the soil cores (Smettem and Trudgill, 1983; Seyfried and Rao, 1987) or in water from field drains (van Ommen et al., 1989; Stamm et al., 1998; Kung et al., 2000).

The choice of an appropriate transport model that conceptualizes transport in a soil can be linked to the soil structure (Feyen et al., 1998). Model parameters can be inferred directly from the soil structure or empirical relationships between model parameters, and soil structural properties can be determined. Aggregate or grain sizes have been linked to the first-order exchange rate parameter that defines the rate of solute exchange between different flow regions in which solutes are advected downwards at different velocities (Saxena et al., 1994; Griffioen et al., 1998; Larsson et al., 1999; Vervoort et al., 1999). Villholth et al. (1998) and Villholth and Jensen (1998) used the dye stained area to estimate the fraction of the pore space in which rapid leaching occurs. Empirical relations between parameters that characterize the width of the pore-size distribution and the dispersion coefficient, a measure of the vertical spreading of an injected tracer pulse, have been reported by Vervoort et al. (1999) and Bejat et al. (2000). Also, the relation between the hydraulic conductivity and the saturation degree can be used to infer transport parameters (Steenhuis et al., 1990; Durner and Flühler, 1996).

Adsorption of tracers to soil particles is often determined in batch adsorption experiments, in which a tracer solution is equilibrated with a given amount of dry soil. However, in batch experiments, the interaction between the liquid and solid phases is optimized using ground soil material suspended in a mixture with a large liquid to solid phase ratio (generally much larger than the liquid-to-solid phase ratio in natural soils). A number of studies illustrated that the mobility of pesticides in natural soils may be much larger than the mobility expected on the basis of adsorption isotherms (Jury et al., 1986; Kladivko et al., 1991; Ghodrati and Jury, 1992; Flury et al., 1995; Jørgensen et al., 1998; Meyer-Windel et al., 1999; Kung et al., 2000). The rapid leaching of adsorbing tracers was linked with fast advective transport through only a small part of the total pore volume (preferential flow and transport). Rapid leaching was found in many, but not all, cases to depend strongly on the initial state of the soil and the flux at the soil surface with more rapid leaching in wet and very dry soils and for high surface fluxes. When preferential flow was triggered, the rapid leaching was not strongly influenced by the adsorption characteristics of the pesticides. These observations point at a less optimal contact between solutes and sorption sites in naturally structured soils than in batch experiments.

Since pesticide analyses are expensive, sorbing dye tracers can be used as a surrogate to investigate the combined effect of lateral mixing and sorption on leaching. Dye tracers can be easily measured in the effluent from a soil core (Smettem and Trudgill, 1983) or from a field drain (Kung et al., 2000). Recently, methods have been developed to derive dye concentrations on excavated profiles (Aeby et al., 1997, 2001; Forrer et al., 2000; Vanderborght et al., 2002). This allows a quantitative interpretation of dye patterns observed on soil surfaces. Using fluorescent dye tracers, concentration patterns of dyes with different sorption characteristics or of dyes that were applied at different initial states of the soil profile, can be simultaneously measured (Aeby et al., 2001), which offers opportunities to identify transport processes and parameters. Time series of dye concentrations or BTCs and dye concentration patterns can be used to test transport models (Forrer et al., 1999).

The objectives of this paper are (i) to investigate the effect of soil structure and adsorption on the movement of two differently mobile fluorescent dye tracers, BF and SB, through undisturbed soil cores, and (ii) to illustrate the added value of dye concentrations maps in identifying causes for rapid dye breakthrough, selecting appropriate transport models, and determining transport model parameters.

Breakthrough curves of the dye tracers and concentration patterns in horizontal cross sections of the soil cores were determined and used to calibrate and validate different transport models. Sorption characteristics of the two dyes, BF and SB, are investigated using batch adsorption experiments. The soil structure in the soil cores is determined using x-ray computer tomography (CT) so that a direct link between observed dye concentration patterns and soil structure can be made. With x-ray CT, the 3-D density structure within the soil cores is determined in a nondestructive way (Hopmans et al., 1994; Heijs et al., 1995).

	METHODS

TOP NOTES ABSTRACT INTRODUCTION METHODS RESULTS SUMMARY AND CONCLUSIONS REFERENCES

 
Tracer Experiment
A leaching experiment was carried out in 9 undisturbed soil cores (107-mm i.d., 110-mm length) that were taken at three depths (0–0.1 m, 0.2–0.3 m, and 0.5–0.6 m) in a loamy Stagni-Humic Cambisol under mixed Norway spruce [Picea abies (L.) H. Karst.] and beech (Fagus sylvatica L.) forest (Unterehrendingen, Switzerland). Physical and chemical properties of this soil are listed in Table 1. The soil samples were taken by pushing stainless steel cylinders into the soil. In the lab, the samples were pushed out of the steel cylinders and encased in resin (Araldite CW 2418-1 and hardener HR 5162, Novartis, Switzerland). In some of the first set of samples the resin intruded from the side walls through larger pores into the core (see Table 2). To avoid resin intrusion, we coated the side walls of the remaining samples with a fine loam paste. Following resin hardening, the samples were saturated with water and put on fritted glass plates (10- to 16-mm pore size, -50-kPa air entry value, 8-mm thickness) in sample holders. The samples were placed in a plexiglass chamber. In the chamber, a mist of fine water droplets was injected by four nozzles that were horizontally installed at the side walls (Kasteel, 1997). By switching the nozzles periodically on and off, a rather small quasi steady state water flux was established at the surface of the soil cores. The applied flow rate to each soil core was measured using a catch container that was put on top of the soil core. It ranged from 33 to 51 mm d^-1 depending on the location in the chamber. In some soil cores, the flow rate applied onto the surface exceeded the infiltration rate and ponded the surface. A suction of -0.3 m was applied at the bottom of the soil cores. The outflow was collected in 0.5-h intervals using autosamplers and fluxes through the cores were calculated from weighting the collected outflow samples. The outflow rate remained fairly constant in all soil cores except in Soil Core 2 of the 0.5- to 0.6-m layer where it varied considerably during the leaching experiment. In some cores, the actual water flux was considerably smaller (by a factor of 10) than in the others. Slicing the cores at the end of the leaching experiment showed that resin intruded and clogged larger pores in these cores. Once a steady flow rate was achieved, 1 mL of a tracer cocktail with 2 g L^-1 BF (Sigma Chemical Co., St. Louis, MO), 0.2 g L^-1 SB (Fluka Chemie AG, Buchs, Switzerland) (dye concentrations are expressed as mass formulated dye powder per unit volume water) and 10 g L^-1 CaCl₂ · 2H₂O was applied every 30 min to approximate a step input of tracer solution with concentration C₀. The cocktail was manually sprinkled on soil surfaces with a syringe. The input concentration, C₀, was calculated from the amount and concentration of the applied tracer cocktail and the flux of solute free water at the soil surface that diluted the applied tracer concentration between two subsequent applications. The applied C₀ ranged from 0.2 to 0.32 g L^-1 for BF, 0.02 to 0.032 g L^-1 for SB, and 1.0 to 1.6 g L^-1 for CaCl₂ · 2H₂O. The infiltration of the tracer solution was stopped 72 h after the start of the tracer application for a first set of soil cores and after 96 h for a second set. Flow rates in the soil cores, cumulated drainage from the cores from the start of the tracer application until leaching was stopped, dilution factors used to calculate C₀, and the duration of the leaching experiment are given in Table 2. After stopping the infiltration, the cores were drained by gravity for 24 h. The cores were horizontally sliced at 100 mm, 95 mm, and then by 10-mm increments to 15 mm above the bottom surface. The cores from the second set were wrapped in plastic to prevent drying and stored for 5 d before slicing. To avoid smearing of the dye over the soil surface, the resin was first sawn at marked depths using a band saw, and then the soil in the detached resin ring was cut from the soil sample using a sharp knife. From each horizontal slice a fluorescence image was taken using a slow-scan cooled charge-coupled device (CCD) camera to determine the concentration patterns of the two dyes at the cross sections. The four slices at a certain height above the bottom of the soil cores in one set were imaged at the same time. The fluorescence image was corrected for variations in illumination and light absorption by the soil surface. The dye concentration expressed as total mass of dye per unit volume of bulk soil, C_t, was derived from the corrected fluorescence signal using a linear calibration relation. Detection limits of dye concentrations on soil surfaces were derived from the spatial variability of the background fluorescence of nonstained soil. The detection limit of total BF concentrations amounted 50 mg L^-1 and 150 mg L^-1, respectively, for the top (0- to 0.1-m and 0.2- to 0.3-m layers) and the sub soil (0.5- to 0.6-m layer). For total SB concentrations, the detection limit was 3 mg L^-1. Detailed information about the image acquisition set-up, image analysis, and calibration are given by Aeby et al. (2001) and Vanderborght et al. (2002). Dye and chloride concentrations in the collected outflow, C^f, were measured using a fluorescence spectrometer (Perkin Elmer LS50B, Norwalk, CT) and a chloride electrode, respectively.

[1]

where S is the sorbed dye mass per mass of dry soil that is in equilibrium with a solution with concentration C, S_max is the sorbed dye mass per mass of dry soil at maximum sorption, and C_0.5 is the dissolved equilibrium concentration at 50% of S_max. The Langmuir model was fitted to the experimental isotherm using a nonlinear least-squares optimization procedure.

X-Ray Computer Tomography Scans of Three Soil Cores
Before the leaching experiment, x-ray CT images were acquired of three soil cores (Core 1 of the 0.2- to 0.3-m layer, and Cores 1 and 2 of the 0.5- to 0.6-m layer) that were drained by gravity to characterize the 3-D soil density structure in the cores. The attenuation of x-rays that are transmitted through soil cores in different directions perpendicular to the soil core axis are used to reconstruct a two-dimensional horizontal cross section of x-ray attenuation coefficients. A map of x–ray attenuation coefficients can be interpreted as a map of the soil densities. The 3-D density structure is reconstructed from several horizontal cross-sections.

An industrial x-ray CT scanner (Model CITA 101B+, Scientific Measurement Systems, Austin, TX) at the Eidgenössische Materialprüfungs-und Forschungsanstalt (EMPA; Dübendorf, Switzerland) was used to scan the soil cores. For each core, 100 horizontal cross sections were made with a vertical separation distance of 1 mm and a horizontal resolution of 0.5 mm (voxel size 0.5 x 0.5 x 1 mm). Scans were made at 150 kV tube voltage, 6 mA current, and a scan time of 2.5 min per slice. Reconstructed images of the core cross sections were converted to 8-bit images. Since the soil cores were drained by gravity, the pore space was filled by both water and air. Also, a water content gradient existed over the soil cores, which were nearly water saturated at the bottom. Therefore, an exact interpretation of the pixel values in terms of soil bulk density or porosity was not possible. However, a threshold value could be defined to binarize the images and identify larger pores or regions that were considerably less dense than the soil matrix. Since the average pixel values did not change considerably with depth, the same threshold value was used to binarize all cross sections of the three cores. Comparison of pixel values in the soil matrix with the threshold value revealed that also larger pores filled with loose material were identified in the binarized images as less dense regions. To reduce the noise, each binarized image was separately opened using a 2 by 2 pixels square as structuring object. The opening procedure is formally defined as the union of all sets containing the structuring object in the original image (Serra, 1982). Incoherent objects and object parts of the image in which the structuring object does not fit are removed. Since the contrast between high- and low-density regions was low and the histogram of pixel values was not clearly bimodal, the threshold value and the structuring object were chosen on the basis of a visual comparison between the original and processed images.

Transport Models
We used three different transport models to describe transport. In each model, instantaneous equilibrium between dissolved and adsorbed concentrations is assumed (equilibrium sorption). First, we considered the classical equilibrium CDM for steady-state flow in a vertically homogeneous soil core:

[2]

where C (mass L^-3) is the solute concentration in the soil solution, q (L time^-1) the water flux, (L) the dispersivity, _eff the effective water volume in which transport takes place, _b (mass L^-3) the bulk density, K_deff(C) = dS_eff/dC (L³ mass^-1) is the effective distribution coefficient, t is the time, and x is the depth. The CDM assumes: (i) uniform advective transport in the soil cores, (ii) perfect lateral solute mixing, (iii) equilibrium adsorption, and (iv) a uniform distribution of sorption sites in the soil cores. For the 0.2- to 0.3-m layer, we used the adsorption isotherms that were derived for the 0- to 0.1-m layer. Since fewer sorption sites are accessible per unit soil mass when transport is confined to the effective volume _eff, we defined the effective adsorption isotherm S_eff(C), consistent with the assumption that adsorption sites are uniformly distributed across the entire pore space, as:

[3]

where _s is the saturated water content or porosity (Table 1).

The CDM was solved numerically using the following initial and boundary conditions:

[4a]

[4b]

and

[4c]

where L is the length of the numerical soil profile or soil core. Equation [4c] implies there is no dispersive transport across the bottom boundary, and this condition influences transport within the soil column. When dispersion is predominantly caused by microscopic variability of advection velocities, transport within the column should remain unaffected by a downstream boundary, except for its effect on the water flow itself, and the solution of the CDM in a semi-infinite soil column (L ) can be used to predict solute concentrations in finite soil columns. Since the use of Eq. [4c] is questionable and in fact inconsistent with Eq. [4b] (van Genuchten and Parker, 1984), the length of the numerical soil column was chosen to be much larger (L = 100 cm) than the actual length of the soil core (L_c = 11 cm). Flux concentrations, C^f, which correspond to concentration measurements in the outflow from the soil cores, were calculated from resident concentrations in the numerical soil column at depth L_c:

[5]

An adapted version of the HYDRUS software (imnek et al., 1998) was used to solve the CDM (Eq. [2]) for the given boundary conditions (Eq. [4]) and calculate flux concentrations in the outflow (Eq. [5]) of the soil cores.

Two parameters of the CDM, _eff and the dispersivity , were derived from fitting the analytical solution of the CDM in a semiinfinite soil column to the Cl^- breakthrough, assuming that chloride does not adsorb. Using the fitted _eff and and the S(C) from the batch adsorption experiments, transport of the two dyes was predicted.

The second considered model was the STM. A STM conceptualizes the soil as a set of stream tubes in which advective transport takes place without local dispersion and without interaction with other stream tubes, that is, the surface applied concentration C₀ is not diluted in the stream tube. The distribution of travel times, , from the input to the output surface in the stream tubes, pdf(), is derived from the breakthrough of an inert tracer, C^f(L, t) (Jury and Roth, 1990):

[6]

Assuming that sorption sites are uniformly distributed in the soil core and that the volumetric water content is the same in all stream tubes, the effective travel time, _eff, of the nonlinearly sorbing tracer in a given stream tube for a given input concentration at the soil surface, C₀ is:

[7]

where R(C₀) is the retardation coefficient. For a step input of reactive tracer at the soil surface with concentration C₀, the breakthrough of the reactive tracer is predicted by:

[8]

From Eq. [6] and [8], it follows that for a step input, a STM simply scales the time axis of an inert tracer BTC to predict the BTC of nonlinearly adsorbing tracer.

The third considered model is the PNEM (van Genuchten and Wierenga, 1976). In the PNEM, the pore volume is partitioned into a mobile and a bypassed immobile region. Convective dispersive transport takes place in the mobile pore region and mass transfer between mobile and immobile regions is modeled as a first-order kinetic exchange. The PNEM is given by:

[9a]

[9b]

where the subscripts m and im stand for the mobile and immobile pore regions, (T^-1) is the first-order mass transfer rate coefficient between mobile and immobile pore regions, and f the fraction of sorption sites in the mobile pore region. The HYDRUS software was used to solve the PNEM for the same boundary conditions as the CDM (Eq. [4] with C replaced by C_m) and in a much longer numerical soil column than the actual one. The concentrations in the outflow were calculated from the simulated resident concentration in the mobile pore region at the depth L_c (length of the core) using Eq. [5] with C replaced by C_m.

The parameters of the PNEM were derived in two steps. In the first step, the S(C) isotherm and the mobile water content fraction _m/ were derived directly. For linear sorption (constant K_d) _m/ and f cannot be estimated independently from breakthrough data since _m and f are lumped in one dimensionless parameter of the dimensionless form of the transport equation (van Genuchten and Wierenga, 1976; Toride et al., 1995). Therefore, _m/ was estimated directly from the minimal area, A₅₀, in which 50% of the total dye mass in a horizontal cross section can be found. To calculate A₅₀, pixels were ordered according to concentration and the number of pixels that was needed to capture 50% of the dye mass was counted and divided by the total number of pixels in the cross section. When dye concentration is uniform over the cross section, A₅₀ equals 0.5. When preferential flow occurs through a fraction B of the cross section with constant concentration in this fraction B and zero in the remaining fraction, then A₅₀ equals B/2. Shortly, relative to the diffusion time into the immobile zone, after the start of solute application, most of the dye mass is still in the mobile pore region. Therefore, we postulate that _m/ 2 A₅₀. By excluding the lower concentrations from the calculation of A₅₀, we excluded the area of the immobile zone in which dyes already diffused during the leaching experiment and the period between the end of the leaching experiment and the imaging. Although this procedure is to some extent arbitrary, _m/ is derived independently from the BTC and from a parameter that is closely related to the mobile pore region. In contrast to the CDM fits, we did not estimate the fraction of the core volume that was accessible to solute, _eff, but fixed it to the saturated water content _s since the soil cores were nearly water saturated. In the second step, the parameters _m, f, and were obtained by fitting the solution of the PNEM to the BTCs using the HYDRUS 1-D software.

	RESULTS

TOP NOTES ABSTRACT INTRODUCTION METHODS RESULTS SUMMARY AND CONCLUSIONS REFERENCES

 
Dye Adsorption
Figure 1 shows the adsorption isotherms S(C) of BF and SB in the two soil layers. The fitted Langmuir model and its parameters are plotted in Fig. 1. Also shown on the plots is the linear approximation of S(C_eq) for the range of considered concentrations, with the slope representing the linear distribution coefficient, K_d (mL g^-1).

View larger version (23K): [in this window] [in a new window]   Fig. 1. Adsorption isotherms of brilliant sulfaflavine (BF) and sulforhodamine B (SB) of soil from the top (0- to 0.1-m layer) and subsoil (0.5- to 0.6-m layer). C, solute concentration in the soil solution; C0.5, dissolved equilibrium concentration at 50% of Smax; Kd, linear distribution coefficient. S, sorbed dye mass per mass of dry soil; and Smax, sorbed dye mass per mass of dry soil at maximum sorption.

Breakthrough Curves and Total Concentration Depth Profiles
Breakthrough curves of Cl^-, SB, and BF are plotted in Fig. 2 . Breakthrough curves are only shown for those soil cores in which the water flux was sufficiently large to collect enough effluent for analyses during the 30-min sampling period. Relative outflow concentrations, cf = C^f/C₀ are plotted vs. the cumulative amount of water that drained from the soil cores. Figure 3 shows depth profiles of average total dye concentrations that we obtained from the fluorescence images of the soil core cross sections. Since flow rates and duration of the leaching experiments varied for the different cores, these profiles represent different stages of the dye leaching. The total dye concentration, C_t, (mass dye per unit volume bulk soil) was normalized against the total dye concentration when the dye concentration in the soil solution were identical to the input concentration C₀:

[10]

View larger version (26K): [in this window] [in a new window]   Fig. 2. Breakthrough curves of Cl-, brilliant sulfaflavine (BF), and sulforhodamine B (SB) in different soil core (# refers to soil core number) layers. Normalized concentrations, c = C/C0, in the outflow from soil cores are plotted vs. cumulative drainage. C is the concentration in the effluent; C0 is the input concentration. Lines are breakthrough curves predicted by the convection dispersion model that is calibrated to the Cl- breakthrough.

View larger version (34K): [in this window] [in a new window]   Fig. 3. Depth profiles of normalized total concentrations, ct, of brilliant sulfaflavine (BF) and sulforhodamine B (SB) in soil cores (# refers to soil core number). Total concentrations, mass of dissolved and adsorbed dye per unit volume of bulk soil are normalized vs. the expected total dye concentration when the initial soil solution is completely replaced by the applied solution and in equilibrium with the adsorbed dye concentration (Eq. [10]). Lines are depth profiles predicted by the convection dispersion model that is calibrated to the Cl- breakthrough. The amount of water that drained from the cores since the start of the leaching experiment is given in parentheses.

View this table: [in this window] [in a new window]   Table 3. Fitted CDM parameters: effective water content, eff, and dispersivity, .

The BTCs of the dyes indicate that BF is more mobile than SB, which is in accordance with the adsorption isotherms (Fig. 1). Yet, the different mobility of the two dyes can hardly be inferred from the concentration depth profiles. The dye breakthrough suggests that the dyes are more mobile in the 0.5- to 0.6-m than in the 0- to 0.1-m and 0.2- to 0.3-m soil layers. Especially for SB, this is in total disagreement with the dye mobility expected from the adsorption isotherms. A fast dye breakthrough goes along with a typical concentration depth profile in which total concentrations deeper in the soil core are low but remain constant with depth (leading tail): for example, Cores 1 and 2 from the 0.5- to 0.6-m layer, Core 2 from the 0- to 0.1-m layer, and to a lesser extent Core 3 from the 0.2- to 0.3-m layer. The slow dye breakthrough in Cores 1 and 2 from the 0.2- to 0.3-m layer is accompanied by a concentration depth profile in which total concentrations clearly decrease with increasing depth. Concentration depth profiles in Core 3 (0- to 0.1-m layer) and Core 3 (0.5- to 0.6-m layer) illustrate that dyes may be leached relatively deep into the soil core by a small amount of infiltrating water (Table 2).

In the cores from the 0.2- to 0.3-m layer, the CDM predicts almost no dye breakthrough during the experiment, which is in agreement with the measurements. Also the concentration depth profiles in these cores are, at least qualitatively, relatively well reproduced by the CDM. The CDM is not capable of reproducing the rapid dye breakthrough, nor the concentration depth profiles in the cores from the 0.5- to 0.6-m layer. The large dispersivities in the cores from this layer suggest that flow is very heterogeneous and that the assumption of macroscopically uniform flow and transport does not hold. Also the low _eff in Core 1 from this layer suggests that rapid flow occurred in a small part of the total pore volume. The heterogeneous flow and transport led to a poor contact between dye solution and solid phase. For the cores from the 0.2- to 0.3-m layer, the dispersivities were smaller and _eff closer to _s (Table 1), indicating a more homogeneous flow and transport. The relative good CDM predictions suggest that sorption in these undisturbed soil cores was similar to sorption in the batch experiments, in which the contact between dye solution and solid phase is optimized. In the following paragraphs, we will use observed concentration patterns and the x-ray CT data to elucidate further the high dye mobility in the 0.5- to 0.6-m layer.

Dye Concentration Patterns and 3-D Soil Density Structure
Figure 4 shows 3-D reconstructions of the soil density structure and of the 20-mg L^-1 SB concentration isosurfaces in three cores. The soil density structure and concentration isosurfaces were reconstructed from 41 x-ray CT images and nine concentration patterns on horizontal cross sections from 15 up to 95 mm above the bottom of the core, respectively. In the core from the 0.2- to 0.3-m layer, 20% of the core volume was classified as larger pores or less dense regions which formed a dense network. This dense network goes along with a more or less uniform dye distribution over a horizontal cross section and a piston-like displacement of the initial solution by the applied dye solution. For the cores from the 0.5- to 0.6-m layer, only 5% (Core 1) and 4% (Core 2) of the core volume was classified as larger pores or less dense regions. Yet, the dye isosurfaces connect the top with the bottom surface in these cores. The similarity between the dye patterns and the larger pores in the x-ray CT images is evident, especially for Core 1 of the 0.5- to 0.6-m layer. It provides evidence that rapid dye breakthrough in these cores can be explained by preferential leaching through larger pores and a lack of lateral mixing with the soil matrix region. The lateral mixing, or the interaction between the matrix and large pore regions, depends on the proximity of pixels in the matrix region to the nearest stained large pore. Since only stained large pores contributed to preferential leaching, only the interaction between these pores and the matrix is relevant. In Fig. 5 , original and binarized x-ray CT images, overlays between binarized x-ray CT images, and concentration patterns in which stained large pores were identified, and maps of distances to the nearest stained large pore are shown for a cross section in Core 1 from the 0.5- to 0.6-m layer and in Core 1 from the 0.2- to 0.3-m layer. An average distance distribution (Fig. 6a) was calculated from a number of cross sections in which approximately the same area was stained by the dyes. For the cores from the 0.5- to 0.6-m layer, the stained area was more or less constant in all cross sections except the two top sections (100 and 95 mm). For the core from the 0.2- to 0.3-m layer, only the cross sections close to the top of the core were used to calculate the distance distributions (Fig. 3 and 4). In the core from the 0.2- to 0.3-m layer, the pixels in the matrix are considerably closer to the nearest large pore than in the cores from the 0.5- to 0.6-m layer (Fig. 5 and 6a) which explains the better lateral mixing and later breakthrough. The distance distribution also suggests a slightly stronger interaction between the preferential flow region and the matrix in Core 2 than in Core 1 from the 0.5- to 0.6-m layer, which is consistent with the slower breakthrough in Core 2 than in Core 1 (Fig. 2). The concentration profile vs. the distance from the nearest larger pore yields information about the mobility of the dyes in the matrix region. This profile represents the distance across which dyes moved laterally into the matrix during the leaching experiment and the time period between the end of the leaching experiment and the imaging, which was of the same order of magnitude as the duration of the leaching experiment. Average concentration profiles were calculated from overlays between distance maps and concentration images in all cross sections except the two top sections of Cores 1 and 2 of the 0.5- to 0.6-m layer (Fig. 6b). In Core 2, dyes diffused slightly further from the large pore region into the matrix than in Core 1. Since slices of Cores 1 and 2 from the 0.5- to 0.6-m layer were imaged at the same time, this suggests a larger interaction between the large pore and matrix regions due to larger lateral diffusion in Core 2. The concentration profiles of BF and SB are nearly identical, suggesting similar lateral mixing of both dyes despite the substantially different adsorption characteristics of the two dyes.

View larger version (111K): [in this window] [in a new window]   Fig. 4. Three dimensional reconstructs of the sulforhodamine B 20 mg L-1 total concentration isosurfaces and that of the large pores or less dense regions in two soil cores from the 0.5- to 0.6-m layer and one from the 0.2- to 0.3-m layer. The concentration isosurfaces were derived from concentration maps on horizontal core cross sections and the large pores or less dense regions from binarized x-ray computer tomography (CT) images.

View larger version (44K): [in this window] [in a new window]   Fig. 5. (a) Original grey value x-ray computer tomography image (darker values correspond to less dense regions) of a horizontal core cross section (Core 1 of the 0.5- to 0.6-m layer), (b) binarized and opened x-ray CT image, (c) overlay of concentration map and binarized CT image (same cross section) from which identified unstained large pores are removed, and (d) map of distances to the nearest large pore (darker values correspond with larger distances). (e–h): same as (a–d) for a horizontal cross section in Core 1 from the 0.2- to 0.3-m layer.

View larger version (26K): [in this window] [in a new window]   Fig. 6. (a) Average distribution of pixel distances from a large pore in a cross section. (b) Average normalized total dye concentration profile as a function of the distance from a large pore. Averages are taken of cross sections in a soil core with similar dye stained area. Dye concentrations are normalized by the average total concentration in the large pores.

View larger version (44K): [in this window] [in a new window]   Fig. 7. Concentration maps of total brilliant sulfaflavine (BF) concentrations (mass of dissolved and adsorbed dye per unit volume of bulk soil) in core cross sections from the 0.5- to 0.6-m layer and from the 0.2- to 0.3-m layer. Heights of the cross sections from the bottom of the soil cores are indicated. The triangles on the concentration scale indicate the expected total tracer concentration when the initial soil solution is completely replaced by the applied solution and in equilibrium with the adsorbed dye concentration. Closed triangle is for the 0.2- to 0.3-m layer and open triangle for the 0.5- to 0.6-m layer.

View larger version (41K): [in this window] [in a new window]   Fig. 8. Concentration maps of total sulforhodamine B (SB) concentrations (mass of dissolved and adsorbed dye per unit volume of bulk soil) in core cross sections from the 0.5- to 0.6-m layer and from the 0.2- to 0.3-m layer. Heights of the cross sections from the bottom of the soil cores are indicated. The triangles on the concentration scale indicate the expected total tracer concentration when the initial soil solution is completely replaced by the applied solution and in equilibrium with the adsorbed dye concentration. Closed triangle is for the 0.2- to 0.3-m layer and open triangle for the 0.5- to 0.6-m layer.

View larger version (24K): [in this window] [in a new window]   Fig. 9. Predictions of the breakthrough of normalized brilliant sulfaflavine (BF) and sulforhodamine B (SB) concentrations, c = C/C0 (C is the concentration in the effluent; C0 is the input concentration) in the outflow from soil cores (soil core number is given in parentheses) using a stream tube model that is calibrated to the Cl- breakthrough.

For a first set of optimizations (set A in Table 4), we fitted the PNEM first to the Cl^- BTC to obtain _m, and determined subsequently the parameters f and from a fit of the PNEM to the BF and SB BTCs. In a second set of optimizations (set B in Table 4), _m was first derived from a fit of the PNEM to a BTC of SB or BF. The model fits to the BTCs are shown in Fig. 10 . For Set A, small _m are fitted (small _m set) whereas large _m are obtained for Set B (large _m set). The dye BTCs are better described for the large _m set whereas the Cl^- BTCs are better described using small _m values, although the difference in goodness-of-fit between the two parameter sets is small. For SB, the fitted f values are considerably smaller than the fraction _m/, which points at a smaller sorption capacity of the mobile pore region. For BF, the fitted f was similar to the _m/ fraction, suggesting a more homogeneous distribution of BF sorption sites. The fitted f values and the corresponding distribution of sorption sites are in general consistent with the BF and SB concentration maps (Fig. 7 and 8). Comparison of exchange rate parameters, , does not allow conclusive statements about differences in diffusion rates between the different tracers. But fitted are a factor 10 larger in Soil Core 2 than in Soil Core 1. The concentration decrease in the outflow from Core 2 after a sudden decrease in flow rate forced the optimization procedure to fit high exchange rates.

View this table: [in this window] [in a new window]   Table 4. Fitted and directly estimated PNEM parameters: mobile water content fraction, m/; fraction of sorption sites in mobile pore region, f; dispersivity in mobile pore region, m; and first-order exchange rate parameter, .

View larger version (23K): [in this window] [in a new window]   Fig. 10. Fits of the physical nonequilibrium model (PNEM) to the breakthrough of Cl-, brilliant sulfaflavine (BF), and sulforhodamine B (SB) concentrations (C) in the outflow from soil cores from the 0.5- to 0.6-m layer. Full lines are PNEM fits for the large m parameter set and dashed lines for the small m parameter set (see text and Table 4). The bottom right plot gives the outflow rate as a function of time in Core 2.

View larger version (30K): [in this window] [in a new window]   Fig. 11. Mass fraction of brilliant sulfaflavine and sulforhodamine B on a horizontal core cross section plotted vs. the minimal fraction of the total area in which the mass fraction is contained. Smooth lines are mass fraction vs. area fraction plots that are derived from concentration maps on core cross sections (55–15 mm above the bottom of the soil cores, with gray values decreasing with height of the cross section). Discontinuous lines are mass fraction vs. area fraction plots at 45 mm above the bottom of the soil core predicted by the physical nonequilibrium model (PNEM). Full lines are PNEM predictions for the large m parameter set and dashed lines for the small m parameter set (see text and Table 4).

[11]

where R_m (-) is the retardation coefficient in the preferential flow region, v_m (L time^-1) is the advection velocity in the preferential flow region, and _s (time^-1) the first-order adsorption rate coefficient. The retardation factor R_m was estimated as 1 + _b f K_d/_m, and for the _s, we took a conservative estimate of _s = 0.25 h^-1, which is likely to be smaller than the actual _s since the dye adsorption in the batch experiments already reached equilibrium after 2 h. The effective dispersivities _eff that were obtained using this conservative estimate of _s were smaller than 10 mm. Comparing this _eff with the fitted _m (Table 4) or (Table 3), it can be concluded that rate limited sorption did not significantly contribute to the observed spreading of the dye breakthrough and cannot explain the rapid dye breakthrough.

	SUMMARY AND CONCLUSIONS

TOP NOTES ABSTRACT INTRODUCTION METHODS RESULTS SUMMARY AND CONCLUSIONS REFERENCES

 
The mobility of BF and SB in undisturbed soil cores is not unequivocally determined by their adsorption characteristics since dye breakthrough occurred the earliest in the deeper soil layer where dyes adsorbed the most. As a consequence, information about the breakthrough of an inert tracer in combination with the adsorption isotherm of a sorbing tracer does not suffice to predict leaching of the sorbing tracer in this soil. Comparison of a 3-D reconstruction of the dye-stained regions with a 3-D reconstruction of large pores revealed that the dye mobility can be explained by soil structure. The dense pore network in the core from the 0.2- to 0.3-m layer facilitated lateral mixing and resulted in homogeneous flow and transport which could be fairly well predicted by the CDM. In the cores from deeper layer, the density of the large pore network was much smaller and the applied tracer solution was preferentially leached through a few larger pores that connected the inflow and outflow surfaces without laterally mixing with the initial soil solution. The CDM, which assumes perfect lateral solute mixing, could not predict the rapid dye transport. However, neither could a STM, which does assume no lateral solute mixing and a homogeneous distribution of sorption sites. Inspection of SB concentration maps in cores from the deeper layer revealed that local total SB concentrations were considerably lower than the expected total dye concentrations. This points at a lower sorption site concentration in the preferential flow region. Using a PNEM to predict dye breakthrough confirmed that the rapid SB breakthrough was explained by the fraction f of sorption sites in the mobile pore region being smaller than the mobile pore volume fraction, _m/. Therefore, rapid SB breakthrough resulted from a combination of preferential flow in a small fraction of the total core volume and a lower sorption site concentration in the preferential flow region. Besides the physical structure of the soil, the heterogeneity of chemical soil properties also determines leaching. Further evidence of significant differences in chemical and biological properties between preferential flow and matrix regions in this particular soil is given by Bundt et al. (2001) for the same soil. However, the degree of chemical heterogeneity cannot be inferred from BTCs alone. To estimate f from PNEM fits to dye BTCs, the mobile water content fraction _m/, had to be estimated independently using dye concentration maps. This suggests that a combination of breakthrough data and patterns of local concentrations in the soil provides information that is necessary to identify governing transport processes and model parameters. An immediate consequence of this statement is the need for observation tools with a high spatial resolution to capture distribution processes at the subgrid-scale, information that is usually averaged out and lost in structured media.

	ACKNOWLEDGMENTS

 
We wish to thank Dr. A. Flisch of the EMPA in Dübendorf, Switzerland, for the x-ray CT scans of the soil cores. We are also indebted to Dr. J. imnek of the U.S. Salinity Laboratory who provided us with an adapted version of the HYDRUS_1-D model. The assistance of Dr. S. Bujukova, H.-P Läser, J. Leuenberg, and H. Wydler with the experimental set-up, the measurements of the breakthroughs, and the analysis of the fluorescence images was essential and very much appreciated. The reviewers and the associate editor are acknowledged for their constructive comments that helped us to improve a first version of this paper.

The corresponding author is grateful to the Belgian Fund for Scientific Research, Flanders, which funded his research stay at the Soil Physics Laboratory of the Institute of Terrestrial Ecology (ETHZ). While carrying out this research, the corresponding author was a post-doctoral research assistant of the Belgian Fund for Scientific Research, Flanders.

	NOTES

TOP NOTES ABSTRACT INTRODUCTION METHODS RESULTS SUMMARY AND CONCLUSIONS REFERENCES

 
(deceased).

Received for publication November 10, 2000.

	REFERENCES

TOP NOTES ABSTRACT INTRODUCTION METHODS RESULTS SUMMARY AND CONCLUSIONS REFERENCES

 

• Aeby, P., J. Forrer, C. Steinmeier, and H. Flühler. 1997. Image analysis for determination of dye concentrations in sand columns. Soil Sci. Soc. Am. J. 61:33–35.[Abstract/Free Full Text]
• Aeby, P., U. Schultze, D. Braichotte, M. Bundt, F. Moser-Boroumand, H. Wydler, and H. Flühler. 2001. Fluorescence imaging of tracer distributions in soil profiles. Environ. Sci. Technol. 35:753–760.[Medline]
• Altfelder, S., T. Streck, and J. Richter. 1999. Effect of air-drying on sorption kinetics of the herbicide Chlortoluron in soil. J. Environ. Qual. 28:1154–1161.[Abstract/Free Full Text]
• Bejat, L., E. Perfect, V.L. Quisenberry, M.S. Coyne, and G.R. Haszler. 2000. Solute transport as related to soil structure in unsaturated intact soil blocks. Soil Sci. Soc. Am. J. 64:818–826.[Abstract/Free Full Text]
• Booltink, H.W.G., and J. Bouma. 1991. Physical and morphological characterization of bypass flow in a well-structured clay soil. Soil Sci. Soc. Am. J. 55:1249–1254.[Abstract/Free Full Text]
• Bundt, M., F. Widmer, M. Pesaro, J. Zeyer, and P. Blaser. 2001. Preferential flow paths: Biological ‘hot spots’ in soils. Soil Biol. Biochem. 33:729–738.
• Durner, W., and H. Flühler. 1996. Multi-domain models for pore-size dependent transport of solutes in soils. Geoderma 70:281–297.[ISI]
• Feyen, J., D. Jacques, A. Timmerman, and J. Vanderborght. 1998. Modelling water flow and solute transport in heterogeneous soils: a review of recent approaches. J. Agric. Eng. Res. 70:231–256.
• Flühler, H., W. Durner, and M. Flury. 1996. Lateral solute mixing processes—A key for understanding field-scale transport of water and solutes. Geoderma 70:165–183.
• Flury, M., H. Flühler, W.A. Jury, and J. Leuenberger. 1994. Susceptibility of soils to preferential flow of water. Water Resour. Res. 30:1945–1954.
• Flury, M., J. Leuenberger, B. Studer, and H. Flühler. 1995. Transport of anions and herbicides in a loamy and a sandy field soil. Water Resour. Res. 31:823–835.
• Forrer, J., R. Kasteel, M. Flury, and H. Flühler. 1999. Longitudinal and lateral dispersion in an unsaturated field soil. Water Resour. Res. 35:3049–3060.
• Forrer, J., A. Papritz, R. Kasteel, H. Flühler, and D. Luca. 2000. Quantifying dye tracers in soil profiles by image processing. Eur. J. Soil Sci. 51:313–322.
• van Genuchten, M.T., and J.C. Parker. 1984. Boundary conditions for displacement experiments through short laboratory soil columns. Soil Sci. Soc. Am J. 48:703–708.[Abstract/Free Full Text]
• van Genuchten, M.T., and P.J. Wierenga. 1976. Mass transfer studies in sorbing porous media I. Analytical solutions. Soil Sci. Soc. Am. J. 40:473–480.[Abstract/Free Full Text]
• 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]
• Ghodrati, M., and W.A. Jury. 1992. A field study on the effects of soil structure and irrigation method on preferential flow of pesticides in unsaturated soil. J. Contam. Hydrol. 11:101–125.<