| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
a Geological Institute, Copenhagen University, Øster Voldgade 10, 1350 Copenhagen, Denmark
b Geological Survey of Denmark and Greenland (GEUS), Øster Voldgade 10, 1350 Copenhagen, Denmark
c Previously at Environment & Resources, Technical University of Denmark, 2800 Lyngby, Denmark
* Corresponding author (apm{at}geol.ku.dk).
Received 27 May 2003.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
Abbreviations: FBA, fluorobenzoic acid
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
Different multiple tracing techniques have been developed for evaluating the influence of diffusion on the transport process (e.g., Jardine et al., 1999; Callahan et al., 2000; Becker and Shapiro, 2000; Meigs and Beauheim, 2001). For field conditions, the following methods are emphasized by Jardine et al. (1999): (i) apply different solutes with varying diffusion coefficients to the same experiment, (ii) apply different tracers with grossly different sizes, (iii) apply the same tracer in repeated experiments with different flow velocities, and (iv) use flow interruption during the tracer experiment. The first two methods take advantage of differences in the molecular diffusion coefficients between either different solute tracers or between solutes and colloidal tracers. Differences in breakthrough between the applied tracers are thus expected if diffusive exchange is important for the overall transport. The other two methods take advantage of the time dependence of diffusion processes, leading to concomitant differences in the breakthrough curves.
Previous studies of transport in fractured clayey till mostly considered saturated conditions (e.g., Sidle et al., 1998; McKay et al., 1999; Jørgensen et al., 1998, 2002). Seasonal groundwater fluctuations will, however, typically result in unsaturated conditions in the top part of the clay deposits resulting in different flow and transport conditions. Previous studies involving unsaturated fractured media have shown that flow patterns like film flow and fingering may create intermittent flow structures, even for steady-state boundary conditions (Tokunaga and Wan, 1997; Tokunaga et al., 2000; Glass and Nicholl, 1996; Glass et al., 2002; Su et al., 1999). These features further add to the complexity of describing and predicting flow and transport in unsaturated fractured clayey till. Only a few experiments examined the influence of unsaturated conditions on flow and transport in fractured clayey till. Experiments included small-scale laboratory experiments on undisturbed soil monoliths (e.g., Wildenschild et al., 1994), intermediate scale lysimeters (e.g., Vink et al., 1997; Schoen et al., 1999) and dye-tracer experiments (e.g., Forrer et al., 2000; Stadler et al., 2000), and large-scale tile-drain experiments (e.g., Villholth et al., 2000). The different approaches each have limitations in terms of characterizing flow and transport in fractured media. Column and lysimeter experiments are often limited in size, dye experiments fail to characterize the speed of the preferential flow, and tile-drain experiments fail to provide spatial resolution and often have large uncertainties in mass balance calculations.
In this study, a large isolated block of undisturbed clayey till (3.5 by 3.5 by 3.3 m) was used for multiple tracing experiments. The experimental setup gave full control of the water balance and allowed measurement of flux-averaged breakthrough curves for the entire block. A water table was maintained 3 m below surface resulting in a 3-m vadose zone with a saturated matrix and unsaturated fractures and macropores. Transport at three different steady-state flow rates was investigated, and the influence of diffusive exchange on the transport was examined by applying multiple tracing techniques.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
|
Experimental Setup
The tracer experiments were performed on an isolated block of clayey till 3.5 by 3.5 by 3.3 m (Fig. 2)
. The block was isolated with great care to minimize disturbances. A steel plate (4 by 4 m) was inserted at the 3.55-m depth to constitute a base for the block. This was accomplished by excavating trenches on two sides of the block and using a hydraulic press to insert the steel plate horizontally into its position underneath the block. Afterwards, each side of the block was individually excavated and isolated with prefabricated bentonite plates (sodium bentonite, Volclay Panels Type 1C, 1.22 m by 1.22 m by 5 mm; CETCO Europe, Copenhagen, Denmark) and the trenches backfilled with gravel. During all excavations, wood bars protected the block to prevent stress relief of the clay. The block was initially designed for saturated experiments; however, because of minor leakage problems, the study was never performed. In the current study, the block and surroundings were kept under tension by maintaining a water table at 3 m depth both inside and outside the block, which circumvented leakage from the block. For monitoring and controlling transport through the block, four horizontal drainpipes were installed above the steel plate. The drainpipes ran the full length of the block and were evenly spaced at 70-cm intervals. An additional two sets of drainpipes were installed at the 1.5-m and 2.7-m depths; however, since drains are active only during saturated conditions, they were not used in this study. Previously, 25 cm of the topsoil was removed from the plot for the purpose of studying transport through the clayey till. Therefore, the topsoil was also absent in this study, resulting in a final thickness of the block of 3.3 m. To avoid effects from smearing of the block surface, the top layer (1–2 cm) was carefully chipped off and a 10-cm layer of small stones was placed on the surface to protect the block from algae growth and subsequent blocking. Four gratings where afterwards laid across the block to allow for manual inspection without disturbing the block surface. Eight tensiometers were inserted vertically in the block at depths of 0.5 and 1 m, respectively, to monitor the water tension during the different experiments. The tension was read using pressure transducers (measuring interval 0–3.5 m), and the data were recorded every 15 min using dataloggers (Tinytalk Volt Logger, Gemini Data Loggers UK Ltd., Chichester, UK).
|
A wood shelter was constructed over the block to protect the surface from natural rain. Instead, an automatic rainmaking device was designed to provide a uniform water flux to the block. Because of the previous problems with leakage, we chose to only apply water to the innermost 3 by 3 m of the plot, leaving the outer 0.25-m circumference around the block dry. The rainmaking device was constructed with nozzles traditionally used for agricultural chemical spraying. On a 3.5-m transverse metal beam, a spray boom was attached with six flat fan nozzles spread 50 cm apart (110° spray angle, 4110 Syntal, Hardi, Taastrup, Denmark). With a spray height of 35 cm and a spacing of 50 cm, the spray pattern had 50% overlap, which ensured uniform water distribution along the spray boom. A water tank supplied the necessary amount of water to the rainmaking device. Between the tank and the spray boom, a pump and a reduction valve were inserted to secure hydrostatic pressure of 0.2 MPa on the nozzles, and thereby constant flow velocity. The beam was driven back and forth by a computer-controlled belt-driven motor. The speed of the beam was set to 0.167 m s–1, whereby one trip back and forth lasted 36 s. The flow rate was adjusted by using different nozzles and by changing the length of the intermission between two successive runs. Experiments with flow rates of 4, 6.2, and 9.6 mm h–1 were conducted. These rates were obtained by using two different sets of nozzles (application rates of 0.25 and 0.6 L min–1) and two different lengths of intermission (56 and 96 s). The uniformity of the sprinkled water was controlled by collecting water in small containers placed on the soil surface and weighing the accumulated water after a specified time. The distribution was found to be uniform along the direction of the boom movement, with only slightly more water being discharged at the edges of the block, where the beam changed direction. Along the boom, the distribution was fairly uniform, with only small variations depending on the position of the nozzles. The constructed rainmaking device was thus capable of creating a constant flux to the block and thereby imposing steady-state conditions. The water used for the experiments was pumped from a nearby well in the limestone aquifer. The chemical composition of major ions (mg L–1) was: Ca2+ = 105, Mg2+ = 28, Na+ = 66, HCO–3 = 410, SO–4 = 82, Cl– = 72, Fetotal = 2.5, and pH = 7.5. To prevent Fe precipitation in the till block, the water was oxidized by sprinkling it over a rock bed and afterwards filtered through a provisional sand filter. As an extra precaution, a 50-µm capsule filter (Millipore, Glostrup, Denmark) was inserted before the nozzles to remove remaining small particles and iron oxides. This water treatment reduced the Fe content to near zero concentrations.
Evaporation during the experiments was expected to be minor since the wood shelter protected the block surface from sun and wind. Furthermore, the actual flux was determined using small containers placed on the surface, such that evaporation was already incorporated in the measurements. The water balance was controlled on a daily basis by monitoring the water level in the accumulating containers and comparing the amount of discharged water with the intensity of the sprinkled water. At steady-state conditions, a water balance error of about 5% was obtained, suggesting that leakage from the block was minimal under the current setup and that accurate water and solute balances could be derived. Flow rates from 4 to 9.6 mm h–1 were used to establish different saturation scenarios in the block, with the highest application rate being close to the prevailing infiltration capacity of the block. Based on the retention characteristics of the clay, the matrix was expected to be fully saturated under the applied tension, and only preferential flowpaths like fractures and macropores were unsaturated. This was confirmed by the tensiometers that showed near-saturated conditions throughout the experimental period, independent of the different applied flow rates.
Tracer Characteristics
Multiple tracers were selected for the purpose of evaluating the flow and transport mechanisms in the block in general, and the diffusive exchange between fractures and matrix in particular. Four different types of tracers were selected for the experiments, including two halogen anions (Cl– and Br–), two FBAs (2,3-DFBA and 2,6-DFBA), two fluorescent dyes (uranine and sulforhodamine B), and one colloidal tracer(0.5-µm fluorescent latex colloids). The halogen anions and FBAs were selected as the primary solute tracers. In most situations, both Cl– and Br– are considered conservative groundwater tracers (Davis et al., 1985). Due to the high background concentration of Cl– at Avedøre (72 mg L–1), a relative high Cl– concentration of 4000 mg Cl– L–1 was applied as inlet concentration in the form of CaCl2. The background concentration of Br– was negligible (<0.4 mg L–1), and an initial concentration of 3000 mg Br– L–1 in the form of KBr was applied. The halogen anions increased the density of the applied tracer solution with approximately 0.6% compared with the water from the aquifer. However, since the main flow direction in the block was vertical, this increase in density added only insignificantly to the gravitational component of the driving force for water flow. The influence of density-driven flow was therefore not considered a major problem in the experiment.
Several of the FBA isomers have shown excellent water tracing qualities (Bowman and Gibbens, 1992; Benson and Bowman, 1994). For the experiments in Avedøre, 2,3-DFBA and 2,6-DFBA were selected as tracers because they have shown stable behavior during unsaturated conditions (Bowman and Gibbens, 1992). The sorption of several FBA in clay-rich saprolite was examined by McCarthy et al. (2000). They found that ionized FBAs are transported conservatively in a clay-rich saprolite, and concluded that they are useful as nonreactive groundwater tracers as long as the pH is approximately 2 pH units above the pKa for the FBAs. The pKa values for 2,3-DFBA and 2,6-DFBA are 3.29 and 2.85, respectively (McCarthy et al., 2000). Since the pH of the Avedøre clay was between 7 and 7.9, and the water from the limestone aquifer had a pH of 7.5, both 2,3-DFBA and 2,6-DFBA were expected to behave conservatively.
In addition to the halogen anions and the FBAs, two fluorescent dyes were used as tracers. Uranine and sulforhodamine B were selected for this purpose since they are the most conservative fluorescent dye tracers available and are easily detected without interfering with each other (Kasnavia et al., 1999; Käss, 1998). Both uranine and sulforhodamine B were expected to sorb slightly to the clay material, especially sulforhodamine B, and thus they would not act as conservative tracers. However, the main purpose of applying the fluorescent dyes was to obtain a continuous on-site measurement of the breakthrough curve using a flow-through fluorometer, and thereby use them as indicators of the breakthrough of the primary tracers. The fluorescent dye tracers were applied in concentrations of 5 and 10 mg L–1, respectively.
Transport of solutes in fractured media is generally influenced by diffusive processes, and as such depends on the aqueous molecular diffusion coefficient of each solute. Table 1 lists the diffusion coefficients for the selected tracers. The halogen anion tracers have the highest diffusion coefficients, followed by the two FBAs that both have the same diffusion coefficient. The fluorescent tracers have the lowest diffusion coefficients and are therefore expected to be less subject to diffusive exchange.
|
Experimental Procedure
Experiments at three different flow rates were conducted on the isolated block. The experiments took place at steady-state conditions using constant water application rates of 9.6, 6.2, and 4 mm h–1, respectively. Throughout each experiment, the desired flow rate was maintained constant to ensure that the same fractures were active, thus avoiding changes in the physical conditions for diffusion. Before the tracer application, water from the limestone aquifer was sprinkled at the desired flow rate for a minimum of 7 d. Water balance calculations were conducted to verify that steady-state conditions were obtained; no tracers were applied before an acceptable water balance between applied and discharged water was obtained (water balance error <10%). Tracer solutions were applied at the same rates as those for water, in short pulses of either 10 or 120 min. Immediately after the tracer pulse, sprinkling with water from the limestone aquifer was continued. Tracer applications thus took place without changing the flow paths through the block.
The conducted experiments are presented in chronological order in Table 2. In each experiment a tracer from each of the considered tracer groups was applied, that is, one halogen anion, one FBA, and one fluorescent dye. Tracers with different diffusion coefficients were used in each experiment to reveal differences in the diffusion processes. Since 2,3-DFBA and 2,6-DFBA have the same aqueous molecular diffusion coefficient (Table 1), the FBA applied in each experiment acted as a reference tracer, thereby allowing for comparison between different experiments. For tracers that were applied several times, the invoked differences in pulse length and flow rate assured that the tracer was present in only small amounts during the first application, thereby preventing high background concentrations during a second application. The tracer application method varied depending on the pulse length. In the experiments with a 10-min tracer pulse, the tracer solution was sprinkled manually using a can with a 50-cm spreader. When a 120-min pulse was applied, the tracers were mixed in a 200-L container connected to the automatic rainmaking device and thereby spread on the block with the spray boom.
|
The water samples were afterwards analyzed for the halogen anions and the FBAs. Chloride concentrations were determined on an auto analyzer (Technicon Autoanalyzer II, Tarrytown, NY) with a detection level of approximately 0.2 mg L–1, while Br– was analyzed on an ion chromatograph (DX-120, Dionex Denmark, A/S, Rodovre, Denmark) with a detection level of about 0.1 mg L–1. A new method was developed for analyzing the FBAs by applying liquid chromatography–tandem mass spectrometry with a detection limit of 1 µg L–1 (Juhler and Mortensen, 2002). As described above, the fluorescent dye tracers were measured in the field with the flow-through fluorometer; however, uncertainties were expected because of the pH sensitivity of the uranine fluorescence (Käss, 1998; Smart and Laidlaw, 1977). Since the pH during the experiments changed from approximately 6 in the tracer solution to approximately 7.5 in the effluent water, the concentration of inlet and outlet water were measured at varying pH conditions. Additional measurements of the uranine fluorescence were therefore conducted on the collected samples to remove this uncertainty. Five drops of NaOH were added to the 30-mL samples to adjust the pH to approximately 11, where maximum fluorescent intensity is found, after which the samples were reanalyzed using the fluorometer. Sulforhodamine B does not show the same pH dependence, and additional measurements were not deemed necessary. The concentration of the fluorescent colloids was determined by filtration and epifluorescent microscopy. The collected water samples were vacuum filtered through 25-mm, 0.22-µm, black, polycarbonate, membrane filters (Isopore Membrane, Millipore). The amount of water filtered ranged from 1 to 30 mL (the total water sample) depending on the amount of colloids in the respective sample. The filters were mounted on glass plates and 30 random fields were point-counted in the microscope (Axioplan Microscope, Zeiss, Oberkochen, Germany).
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
|
|
The 4 mm h–1 experiment displayed a clearly different breakthrough behavior than the experiments at the higher flow rates. Both 2,6-DFBA and Br– showed double peak breakthrough curves (Fig. 4A). The 2,6-DFBA tracer peaked first after 4.8 h (19.2 mm water applied) at a concentration of 1.35%, while a second more smeared peak appeared after 62.5 h (250 mm of water applied) with a concentration of 0.78%. Bromide appeared later with the first peak after 6.1 h (24.4 mm of water applied) at a concentration of 1.08%, and a second peak after 88.5 h (354 mm of water applied) with a concentration of 0.79%. Due to the double peak, the cumulative mass recovery did not reach a plateau as seen at the high flow rates, but increased almost linearly throughout the experimental period (Fig. 4B). Recoveries after 187 h (748 mm of water applied) were 54.8% for 2,6-DFBA and 53.0% for Br–. The breakthrough curve for the uranine tracer did not display the same distinct double peak behavior. Uranine peaked after 4.8 h (19.2 mm of water applied) at a concentration of 0.59%, while the cumulative recovery rate was 8.5%.
The 0.5-µm latex colloids, applied as a particle tracer in the 4 mm h–1 experiment, were expected to be less influenced by diffusive exchange compared with the solutes because of their larger size. The obtained breakthrough curve for the colloids was also remarkably different, as seen in Fig. 4, where the breakthrough is compared with the Br– data. The colloids peaked earlier than Br– (2.7 h compared with 6.1 h for Br–), but with a much lower relative peak concentration (0.23% for the colloids). The colloids were relatively quickly removed from the leaching solution without any distinct tailing and without the double peak behavior found for Br–. Total recovery after 748 mm was only 0.47%, considerably lower than the 53.0% found for Br–. A large amount of the colloids is likely retained in the 10-cm stone layer on top of the clay surface and also in the block itself. Much of this can be attributed to settling and physical straining (Cumbie and McKay, 1999; Becker et al., 1999), but also in part to depositing of colloids on the air–water interfaces (Wan and Wilson, 1994; Corapcioglu and Choi, 1996) and by film straining (Wan and Tokunaga, 1997).
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
The tailing in the 4 mm h–1 experimental data was different for both the Br– and 2,6-DFBA tracer, which displayed a double peaked breakthrough. The double peaks can be a result of several processes. First, different flowpaths can transport the tracers with different velocities, such that the slower flowpaths will lead to a second breakthrough. For some structured media, a second breakthrough could also occur by advective transport through the matrix. However, the relatively low hydraulic conductivity of the clay matrix in our experiments will result in an advective travel distance in the matrix of only about 3 mm, and therefore cannot explain the second peak. Second, the double peak can be explained by extensive diffusion into the matrix, which would cause the breakthrough curve to have a small first peak followed by a second more smeared peak as the tracers are washed out of the matrix. This hypothesis is partly supported by the uranine and colloid tracers. The fluorescent tracer, uranine, did not show the double-peak breakthrough behavior as seen for both Br– and 2,6-DFBA. This may be explained by uranine having less exchange with the matrix because of a smaller diffusion coefficient (Table 1), but also by sorption of the fluorescent tracer. However, the two processes are difficult to separate. The fact that uranine does not show the double peak behavior suggests that differences among the various breakthrough curves, at least in part, are a result of differences in the diffusion characteristics of the three tracers. The difference between the colloid breakthrough and the solute tracers shows the importance of solute size on the transport process. The smaller Br– molecule is strongly influenced by diffusive exchange, which leads to more matrix diffusion and extensive tailing of the breakthrough curve. The colloids are larger in size and are thus less affected by diffusion into the matrix and other areas with stagnant water. Additionally, the colloids are mainly transported in the middle of the flow paths where the velocity is higher, resulting in accelerated transport of the colloids as compared with the other solute tracers (Becker et al., 1999). Size exclusion due to the relative small pores of the clayey till matrix and straining leads to low peak concentrations and poor cumulative mass recovery rates.
The differences in breakthrough for the colloids and especially the solutes confirm the influence of diffusive processes in the 4 mm h–1 experiment. Our results support the hypothesis that retardation due to matrix diffusion was part of the observed differences in the breakthrough curves for the different solutes. Other studies using multiple tracing techniques have shown results that were both consistent and inconsistent with the premise that diffusive exchange will affect breakthrough tailing. For example, multiple tracing experiments by Jardine et al. (1999), Callahan et al. (2000), and Hu and Brusseau (1995) confirmed that matrix diffusion is a significant part of the overall transport process in fractured media, whereas experiments by Becker and Shapiro (2000) showed no significant differences in breakthrough for the different tracers. They concluded that more complex mechanisms were causing the tailed breakthrough, such as heterogeneity and non-Fickian transport.
Our results revealed relatively large differences between the breakthrough curves conducted at different flow rates. The influence of flow rate and degree of saturation on transport is best seen by comparing the breakthrough responses from the FBA tracers, which have similar diffusion coefficients. In saturated fractured media, a reduction in flow velocity generally results in a longer solute residence time in the fractures and thereby an increase in diffusive exchange with the matrix. During unsaturated conditions, however, different flow rates will result in different flow patterns and different saturation profiles, and potentially considerable differences in local-scale diffusion processes. In view of the measured retention curves, the clayey till matrix at Avedøre should be fully saturated during all experiments. The different applied flow rates will therefore only change the saturation of the fractures, resulting in different active flow pathways. At the higher flow rates, more fractures will be fully saturated, thus enabling the solutes to be transported quickly through the large fractures where matrix diffusion has less influence. At the low flow rates, the large fractures will partly drain and flow may become more dominated by film flow or finger flow. Solutes transported by film flow and/or through the small fractures should be subject to more diffusion into the matrix. Hence, at the high flow rates, tracers are expected to peak earlier and with a higher peak concentration, followed by less pronounced tailing than in experiments at lower flow velocities.
The observed breakthrough curves for FBAs were found to be very different for the three experiments. The 6.2 and 4 mm h–1 application rates, both with tracer pulses of 120 min, produced two distinct tracer breakthroughs (Fig. 3 and 4). The tracers at the 4 mm h–1 flow rate showed considerable retardation, which is reflected by both the delayed peak and the lower peak concentration, and also by the double peak behavior. The latter, however, resulted in a relatively high recovery rate of 54.8% after 748 mm of water was applied. This value compares well with the recovery found for the 6.2 mm h–1 experiment, where a similar water application rate (748 mm corresponding to 120 h) resulted in a recovery of 56%. The 9.6 mm h–1 experiment was conducted with only a 10-min tracer pulse, which produced, as expected, a much lower peak concentration than the 6.2 mm h–1 experiment. The short pulse and the high intensity would imply less influence by matrix diffusion in the 9.6 mm h–1 experiment. However, the log-log slope of the FBA breakthrough curve was more gradual than for the 6.2 mm h–1 experiment (–0.78 at 9.6 mm h–1 and –0.98 at 6.2 mm h–1), which indicates more pronounced tailing. The reason for this deviation is likely an input pump failure we experienced during the 9.6 mm h–1 experiment. The pump failure happened approximately 115 min after the application, which is immediately after the peak, and resulted in water not being sprinkled on the block for approximately 2 h (until 3.9 h after start). During this unintended flow interruption the fractures were expected to have drained, thus changing the circumstances for diffusion. Flow interruption with drainage of the preferential flow paths has been shown to increase solute leaching (e.g., Cote et al., 2000). For the 9.6 mm h–1 experiment, drainage during the flow interruption likely increased the leaching of the applied tracers, resulting in a more asymmetrical breakthrough curve. The effect of the drainage can also be found in the shape of the breakthrough tail. For the 6.2 mm h–1 experiment, a sharp decrease in concentration occurred immediately after the peak, corresponding to advection-dominated transport in the fractures. As diffusion becomes limiting for transport, there is a small change in the log-log slope curve, especially for the 2,3-DFBA tracer. This change in slope is not seen for 2,6-DFBA of the 9.6 mm h–1 experiment where the advection-dominated part was disturbed by the flow interruption.
While our experiments were performed with great care, results still may depend somewhat on the experimental setup. The clayey till at Avedøre is extensively fractured, generally more so than elsewhere in Denmark (McKay et al., 1999). The establishment of the isolated block, and particularly the installation of the steel plate at 3.55-m depth, may have increased the fracturing of the till near the bottom of the block. But since the lower 30 cm was kept saturated and used for drainage of the block, the influence was probably minor. Stress relief during excavation of the sides may also have disturbed the block and caused fractures to collapse. Furthermore, lateral flow was reduced because of the sidewalls. The large amount of water supplied to the block during the experimental period may also have slightly changed the hydraulic properties of the block. While the Fe content of the inlet water was reduced to limit Fe precipitation on the surface and on the fracture walls, the large water application may still have changed the fracture system. For example, during the experimental period we experienced some problems with clogging. This was alleviated by letting the block dry for 2 to 3 d, and then removing the top few centimeters of the clay surface. The clogging problem was mostly a surface problem, likely caused by a combination of small particles being released from the surface, bacterial growth, and maybe Fe precipitation. Some clogging along the fracture walls may also have taken place, which would imply less interaction between the fractures and the matrix, and thus higher peak concentrations and less tailing. Scraping off the surface resulted in the stone layer being removed from the plot. The final 6.2 mm h–1 experiment was hence conducted without the top layer of stones. However, the precise influence of the stone layer on observed transport was difficult to establish for our dataset. Finally, the four bottom drains that collected all water from the block affected the measured breakthrough curves somewhat since they increased mixing and dilution of the tracers. Regardless of these limitations to the experimental setup, the data obtained show that fast preferential flow in unsaturated fractured clayey till is to be expected. Results indicate that transport is very much influenced by the applied flow velocity, and that diffusive exchange has a significant effect on the transport process, especially at the lower flow rates.
In subsequent work we plan to analyze the experimental data in terms of several transport models. Preliminary results using a simple one-dimensional double-porosity model as embedded in CXTFIT 2.1 (Toride et al., 1999) showed that the obtained breakthrough curves, including the double peaked curve found for the 4 mm h–1 experiment, could be fitted fairly well; however, the estimated parameters were not all physically realistic (Mortensen, 2001). Similar limitations when applying simple double-porosity models to complex fracture transport problems were noted by Hu and Brusseau (1995) and Haggerty et al. (2001), among others. To describe the experimental results from Avedøre, more advanced modeling tools may be needed, including a better description of the heterogeneity of the system, the fracture distribution, and the bottom drainpipes.
|
CONCLUSIONS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
|
ACKNOWLEDGMENTS |
|---|
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  A. E. Rosenbom, V. Ernstsen, H. Fluhler, K. H. Jensen, J. C. Refsgaard, and H. Wydler Fluorescence Imaging Applied to Tracer Distributions in Variably Saturated Fractured Clayey Till J. Environ. Qual., March 1, 2008; 37(2): 448 - 458. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S. Dasgupta, B. P. Mohanty, and J. M. Kohne Impacts of Juniper Vegetation and Karst Geology on Subsurface Flow Processes in the Edwards Plateau, Texas Vadose Zone J., October 3, 2006; 5(4): 1076 - 1085. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. Persson, S. Haridy, J. Olsson, and J. Wendt Solute Transport Dynamics by High-Resolution Dye Tracer Experiments--Image Analysis and Time Moments Vadose Zone J., August 16, 2005; 4(3): 856 - 865. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. Kjaer, P. Olsen, M. Ullum, and R. Grant Leaching of Glyphosate and Amino-Methylphosphonic Acid from Danish Agricultural Field Sites J. Environ. Qual., March 1, 2005; 34(2): 608 - 620. [Abstract] [Full Text] [PDF]
|
|
| |||||||||||||||||||||||||||||||||||||
