| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
a Department of Environmental Sciences and Energy Research, Weizmann Institute of Science, Rehovot 76100 Israel
b Department of Civil Engineering and Geological Sciences, University of Notre Dame, Notre Dame, IN 46556
* Corresponding author (brian.berkowitz{at}weizmann.ac.il).
Received 5 July 2003.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONDEFINITION OF THE CAPILLARY...FLOW WITHIN AND ACROSS...EXPERIMENTAL DEMONSTRATIONSNUMERICAL SIMULATION OF FLOW...CHEMICAL AND MICROBIAL TRANSPORT...DISCUSSION AND CONCLUSIONSREFERENCES |
|---|
Abbreviations: CF, capillary fringe • LNAPL, light nonaqueous phase liquids • PSF, partially saturated fringe • TCE, trichloroethylene • TDR, time domain reflectometry
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONDEFINITION OF THE CAPILLARY...FLOW WITHIN AND ACROSS...EXPERIMENTAL DEMONSTRATIONSNUMERICAL SIMULATION OF FLOW...CHEMICAL AND MICROBIAL TRANSPORT...DISCUSSION AND CONCLUSIONSREFERENCES |
|---|
Flow through the vadose zone is commonly characterized by a mean vertical flow, with substantial temporal and spatial variability. Within this zone, water pressures are less than atmospheric. Studies of flow in the vadose zone often focus on issues of fingering of water fronts and characterization of the relationships among moisture content, matric potential, and relative permeability. In contrast, flow in the groundwater zone is generally characterized by pores that are essentially saturated with water. The liquid within the pores is typically at pressures greater than atmospheric. Flow is no longer predominantly vertical, but reacts to local or regional recharge and discharge fluxes, including natural fluxes such as recharge from precipitation or discharge to surface water.
At the contact between these two zones is a relatively narrow region that serves as a transition between the vadose zone and the groundwater zone. Sediments immediately below the water table, for example, are characterized by pore water pressures greater than atmospheric pressure, and nearly full saturation by the liquid phase in most portions of the medium. Also present are local zones of entrapped air in the form of bubbles or small inclusions of sediments at low moisture content. The region immediately above the water table contains water that essentially saturates the pores, although the pore water pressures are less than atmospheric. This latter region, above the water table, is commonly known as the capillary fringe (CF). As discussed below, the CF combined with the region immediately below the water table may affect, far more significantly than is usually assumed, the natural geochemical and microbial conditions present in the region of transition from the vadose zone to the saturated groundwater zone. Thus, we argue that this interface is not necessarily a "simple" lower (upper) boundary for the vadose (groundwater) zone, which might be treated as one treats, for example, constant head, constant flux, and impermeable boundaries. Rather, the combination of the CF with the region immediately below the water table is a unique interface that deserves focused study.
Here, we examine underlying physical factors that control local flow and transport behaviors in and across the CF and discuss aspects of associated laboratory and field data. This contribution is not a review, inasmuch as the amount of published literature on the subject is relatively limited. Rather, we provide an exposition and critical discussion of the subject to speculate on the nature and relative importance of various factors controlling flow and transport in and across the CF. We focus our discussion on the region surrounding, above and below, the water table. As such, we shall consider not only the CF, but also the region with potentially significant amounts of air below the water table (see below). We note, however, that the significant body of existing literature on unsaturated and partially saturated flow and transport is not directly relevant here, and so will not be mentioned.
|
DEFINITION OF THE CAPILLARY FRINGE |
|---|
|
TOPABSTRACTINTRODUCTIONDEFINITION OF THE CAPILLARY...FLOW WITHIN AND ACROSS...EXPERIMENTAL DEMONSTRATIONSNUMERICAL SIMULATION OF FLOW...CHEMICAL AND MICROBIAL TRANSPORT...DISCUSSION AND CONCLUSIONSREFERENCES |
|---|
Although exact terminology varies, most textbooks refer to the CF as the thin zone above the water table with essentially 100% water saturation (as low as 75–85%, depending on the textbook), but with air-entry pressure less than that required to penetrate the water table (e.g., Bear, 1972; de Marsily, 1986; Domenico and Schwartz, 1990; Fetter, 1992). Generally, within homogeneous sediments the CF extends from the water table up to the limit of the capillary rise of water.
The thickness of the CF depends on the soil properties, in terms of soil type and uniformity of pore sizes, and can range from centimeters for coarse sands to several meters for geological formations rich in clays. The natural heterogeneity of most geological materials can therefore be expected to have a significant influence on the heterogeneity of the CF. Thus, while there may be a sharp interface between air and water in any given soil pore, the height of capillary rise within a porous medium is expected to be highly heterogeneous in space. Further, inputs that lead to frequent fluctuations in the water table height, such as frequent infiltration and/or drying events or pumping, also have a significant effect on the CF thickness. Because of these nonuniformities, the upper "surface" of the CF has an irregular shape. Further, it is expected that the CF, under field conditions, can contain pockets of air and gases of variable size and configuration (depending also on the degree of microbial activity present).
A number of definitions of the CF exist within the literature. All of these definitions accept the water table (p = gauge pressure = 0) as the lower limit on the CF. Differences among the definitions are generally related to the definition of the upper limit of the CF. For example, Ronen et al. (1997), who employed the approach of Freeze and Cherry (1979) and a simple extension of Bear (1979), defined the upper limit of the CF as the minimum height above the water table at which the moisture content equals the residual water content, as found in the upper (near soil surface) regions of the vadose zone. The choice of the "minimum height above the table" within this definition represents a convenient manner in which to deal with the irregularity of the "upper surface" of the CF. Further, the condition of "minimum height" can be justified if one assumes that horizontal flow within the CF will occur primarily at elevations for which there is a continuous horizontal phase of saturated pores (see discussion below). However, alternatives to this definition can also be justified. These might include defining the upper bound on the CF in terms of (i) the maximum height above the water table for which at least one group of pores remains connected to the water table by a continuous path of pores essentially at saturation (i.e., maximum capillary rise), (ii) the height above the water table for which the average moisture content reaches a predefined value between full saturation and residual moisture, or (iii) the height above the water table for which significant horizontal contributions to flow or transport are considered plausible.
While these definitions are generally precise and often physically relevant, they are subject to difficulties at the local scale and related to the heterogeneity of the relative degree of saturation within the CF. For example, Dunn and Silliman (2003) demonstrated the presence of entrapped air in the region bounding the water table. Hence, a definition of the CF that is applicable at the local scale and based on a measure of the height to residual moisture content (i.e., satiated in the air phase) becomes somewhat problematic from the standpoints of both measurement and description of important flow and transport processes.
As discussed more below, a more adequate definition of the region bounding the water table at the local scale can be developed in terms of the relative connectivity of the liquid and gas phases both above and below the water table. For example, the region bounding the water table for which the liquid phase is well connected horizontally, while the air phase is discontinuous (or nonexistent), will likely demonstrate both horizontal and vertical pore water velocities associated with the regional gradient and local heterogeneity. In contrast, the region above the water table for which the liquid phase is poorly connected horizontally, while the air phase is well connected to the air phase in the overlying vadose zone, will exhibit behavior more classically associated with the vadose zone. There is likely a transition zone between these two regions. We herein suggest that this transition zone be referred to as the partially saturated fringe (PSF) bounding the water table. This PSF is bounded from above by the lower portion of the unsaturated zone, where fluid movement is dominated by gravity driven flow and fingering, and from below by the upper portion of the groundwater zone, characterized by fully three-dimensional flow and a discontinuous air phase. It should be noted that the lower boundary of the PSF is not considered to be the water table because, as has been shown and will be discussed herein, partially saturated regions can exist below what has been defined as the water table. Thus, the definitions of the upper and lower boundaries of the PSF are in some sense arbitrary, making the definition of PSF subjective and based on the particular application under consideration. It is suggested, therefore, that currently an adequate definition describing the behavior of the CF at the local scale does not exist. Rather, definition of the PSF at the local scale, such as proposed here, effectively accounts for the connectivity of the liquid and air phases (and, potentially, additional phases such as light nonaqueous phase liquids [LNAPL] or dense vapor-phase contaminants) within the porous medium. As this paper is intended as an examination of the current state of knowledge of the CF, the term CF will be utilized throughout this manuscript. Further discussion of the PSF, which we consider to be a more inclusive description of local behavior in the vicinity of the water table, is presented in the Discussion and Conclusions section.
While definition of the transition zone at the local scale may be problematic, it is often sufficient to define the CF at larger scales as the zone immediately above the water table where the water content remains at or very close to saturation. In operational terms, the top of the CF is located where the pressure in the pore water is at the average air-entry value (or bubbling pressure) for the porous medium, as used in the equations of Brooks and Corey (e.g., van Genuchten, 1980; Simunek et al., 1999). This definition is particularly useful at the regional scale for two reasons. First, it is consistent with discussions of air entry in theoretical and numerical analyses of flow in partially saturated regions. Second, in many practical situations at both the laboratory and field scales, the region of transition from full saturation to residual saturation can be modeled as relatively sharp.
|
FLOW WITHIN AND ACROSS THE CAPILLARY FRINGE |
|---|
|
TOPABSTRACTINTRODUCTIONDEFINITION OF THE CAPILLARY...FLOW WITHIN AND ACROSS...EXPERIMENTAL DEMONSTRATIONSNUMERICAL SIMULATION OF FLOW...CHEMICAL AND MICROBIAL TRANSPORT...DISCUSSION AND CONCLUSIONSREFERENCES |
|---|
In contrast, in the technical literature, a number of authors have published research on the role of the CF as a unique region of the subsurface that impacts the hydraulics of the system surrounding a fluctuating water table. Work has focused on a number of processes in the subsurface, including response to pumping from groundwater wells, hydrologic response of streams, behavior of seepage faces, and exfiltration to excavations (e.g., Abdul and Gillham, 1984; Gillham, 1984; Mixon, 1984).
With respect to response to pumping from wells in unconfined aquifers, Nwankwor et al. (1992) observed that temporal variation in the thickness of the capillary fringe could result in excess storage of water above the water table. Drainage of this water can have a significant effect on the hydraulics of the system, specifically in delaying the response of unconfined aquifers to pumping (e.g., Lee et al., 2001; Nwankwor et al., 1992; Akindunni and Gillham, 1992). Zhang et al. (1999) also noted that the capillarity of a system significantly influences the relationship between the pumping frequency (i.e., continuous or pulsed) and the fluid response.
The CF has also been shown to impact groundwater response to other transients influencing the groundwater system. Examples include the impact of the CF on the specific yield of an aquifer (Nachabe, 2002); water table dynamics, particularly in response to infiltration events (e.g., Rosenberry and Winter, 1997; Neilsen and Perrochet, 2000); hillslope hydrologic processes (Iverson et al., 1997; Torres et al., 1998); storm hydrographs in streams (e.g., Williams et al., 2002; Jayatilaka et al., 1996; Jayatilaka and Gillham, 1996); the response to wave run-up (Li et al., 1997); and groundwater response to tidal fluctuations (Ataie-Ashtiana et al., 2001).
Another aspect of the CF that has been investigated by a number of authors is the formation and function of seepage faces. Wise et al. (1994) showed that the location of the water table and the height of the seepage face are functions of the capillary forces exerted in the vadose zone. Discussions of the significance of capillary flow on modeling of the seepage face and flow patterns have also been reported (Shamsai and Narasimhan, 1991; Romano et al., 1999; Boufadel et al., 1999). It has also been observed that the height of the steady-state seepage face increases with a decrease in the ratio of the domain height (total depth of the subsurface system) to the height of the CF (Naba et al., 2002). Similar studies have investigated the impact of the CF on flow in the vicinity of structures that invade the subsurface environment, such as tunnels and irrigation canals (Kacimov and Nikolaev, 1995; Youngs, 2002; Kacimov, 2003, unpublished data).
This brief survey of the literature, while certainly not exhaustive, serves to illustrate the diversity of studies that have demonstrated the potential impact of flow within the CF on hydraulics in the subsurface. It is worth noting that these efforts were based predominantly on theoretical analyses, field characterizations, or laboratory characterizations focused on macroscale phenomena. Specifically, few of these studies have focused on the local-scale impact of the CF on exchange of water across the water table, or on the influence of heterogeneity of grain size in behavior at the microscale within the CF.
Further, few of these studies involved direct field measurements of flows within and across the CF. It is suggested that this lack of field measurement is in large part due to the difficulty of obtaining representative samples from the CF. Selected measurements from the CF (Ronen et al., 1997) of a relatively homogeneous (sandy, unconfined) aquifer in Israel showed that the vertical distribution of moisture contents, above the water table, were dramatically different in two boreholes separated by only 10 m. These measurements suggest that moisture contents and flow patterns in the CF are likely to be highly heterogeneous. In contrast, these authors completed highly detailed field measurements in a different region of the same aquifer. In this case, the CF was observed to be compact and relatively uniform for 13 boreholes (12 of which were drilled within a radius of 5 m).
|
EXPERIMENTAL DEMONSTRATIONS |
|---|
|
TOPABSTRACTINTRODUCTIONDEFINITION OF THE CAPILLARY...FLOW WITHIN AND ACROSS...EXPERIMENTAL DEMONSTRATIONSNUMERICAL SIMULATION OF FLOW...CHEMICAL AND MICROBIAL TRANSPORT...DISCUSSION AND CONCLUSIONSREFERENCES |
|---|
The dimensions of the first flow cell (Fig. 1) provided for a sand filling volume 0.27 m long by 0.17 m wide by 0.21 m tall. The "homogeneous" porous medium was packed under saturated conditions, and controlled water levels in the inlet and outlet reservoirs allowed a prescribed head drop across the length of the cell. Flow was allowed to reach steady-state conditions, as indicated by a lack of change in the rate of flow through the medium, with constant water levels in the reservoirs. The sands along the inflow and outflow reservoirs, above the level of water in the reservoirs, were confined using screens and were open to the atmosphere. The upper surface of the cell was covered to reduce effects of evaporation from all exposed sand surfaces. The sand itself was medium-grained and unwashed, available locally from the region around Rehovot, Israel. The sand was partially saturated with a water table approximately at the elevation shown in Fig. 1. Here, the CF extended from the water table to the upper surface of the sand.
|
Figure 1 shows the progressive positions occupied by these individual spots of dye. Seven dye spots were released in the cell, with all but the lower-most spot being within the CF. Flow was, on average, from left to right. It is emphasized that the difference in water levels between the inlet and outlet reservoirs caused the driving force for all flow in this cell; no driving force was applied in either reservoir above these water levels. It is clear that the dye spots at and below the water table moved in the direction of the mean hydraulic gradient, and that the images show strong evidence of vertical and horizontal motion of the dye spots situated within the CF. Of particular interest are the dye spots near the inlet reservoir in Fig. 1. They illustrate a significant upward component to flow—compare, for example, the spots near the left boundary in Fig. 1a through 1c. Thus, near the inflow boundary, the pore water velocity has a significant vertical (upward) component both in the region bordering the water table and in the CF at heights up to the upper surface of the sand. Vertical components to the pore water velocity within the CF are also observed—compare Fig. 1c and 1d where the flow field converges toward the saturated zone at the outflow boundary. We again emphasize that these vertical components to the pore water velocity are driven purely by the difference in head between the two bounding reservoirs (and in the absence of infiltration).
Thus, while this experiment is quite simple, it demonstrates at least two key characteristics. First, the active exchange of dye between the CF and the groundwater zone illustrates the complex interplay between the CF and the hydrologic zones bounding it. Second, the rapid horizontal advection of the dye within the CF illustrates the potential for the CF to be an active contributor to the horizontal migration of chemicals, whether they be contaminants, nutrients, or other chemical species of interest.
While largely ignored over the last half century, this behavior—lateral flow above the water table, and transport (in both directions) across the water table—was in fact recognized many years ago. A series of papers (Luthin and Miller, 1953; Day and Luthin, 1954; Luthin and Day, 1955) examined dynamics of fluid flow in the water table region, using flow cell experiments and theory. On the basis of measured moisture contents along 122-cm-long soil columns that had been saturated and then allowed to drain, and together with indirect field evidence, Luthin and Miller (1953) first concluded that the CF region "... may be of considerable influence." In particular, they emphasized the effect of the CF on drainage near ditches and tiles. A series of sandbox experiments (Day and Luthin, 1954) then confirmed that seepage and drainage to open channels is partially controlled by the CF. Building on this understanding, Luthin and Day (1955) used a sand-filled flow cell to demonstrate explicitly the existence of lateral flow in the CF, above a sloping water table. An accompanying theoretical and numerical analysis yielded a streamline map that confirmed fluid flow from the inlet reservoir into the CF, lateral flow through the CF, and fluid flow back into the outlet reservoir (exactly as seen in Fig. 1).
A second laboratory flow cell was used to demonstrate infiltration patterns. The cell provided for a volume of sand 0.8 m long by 0.10 m wide by 0.89 m tall. The sand used was a commercially available, presieved, washed silica sand (0.30–0.425 mm grain size, 40/50 mesh sand) that has been used in a number of previous studies of fluid flow and chemical transport in both the vadose and saturated zones (e.g., Schroth et al., 1996; Fry et al., 1997; Silliman et al., 2001, 2002; Dunn and Silliman, 2003; Levy and Berkowitz, 2003). The pressure–moisture relationship for these sands appears to follow the van Genuchten (1980) model (Schroth et al., 1996), with an air-entry head of approximately 0.20 m.
Similar to the other flow cell, the porous medium studied was packed under saturated conditions, with the water level later lowered in the inlet and outlet reservoirs to establish a prescribed head drop across the length of the tank. Flow was allowed to reach steady-state conditions, as indicated by a constant elevation of the upper boundary of the region of high moisture content (and easily identified through visual inspection, as illustrated in Fig. 2) . The sands along the inflow and outflow reservoirs, above the level of water in the reservoirs, were horizontally confined using screens and were open to the atmosphere.
|
The Impact of Heterogeneity
To illustrate the impact of heterogeneity on behavior bounding the water table at the local scale, we briefly discuss results obtained from a series of experiments reported in Dunn and Silliman (2003). In this effort, inclusions of coarse sand were added to an otherwise homogeneous fine sand matrix within a sand tank 0.51 m long by 0.16 m wide by 0.39 m tall. The sands used for these experiments were the same as those used in the experiments described above with the fine sand being the same grain size as used in the experiments shown in Fig. 2 and the coarse sand having a grain size of 0.85 to 1.55 mm (with an air-entry pressure of approximately 6 cm). These experiments were instrumented with time domain reflectometry (TDR) probes to monitor moisture content and a pressure transducer to monitor pressure at a single location within the medium.
The distribution of sediments is shown in Fig. 3 . It is noted that the coarse sand in the center of the tank is overlain by fine sand, whereas the coarse sand to the right in this figure is exposed directly to the atmosphere. It is the comparison of the behavior of these two zones that is the focus of our discussion. Details of all experiments performed can be found in Dunn and Silliman (2003). It is important to note that both zones consist of the identical coarse sand packed under saturated conditions. Two experiments were performed. In the first, performed immediately after packing under saturated conditions, the water table was slowly lowered until the two reservoirs were empty. Following this experiment, the second experiment was initiated by slowly raising the water table until it was once again at the upper surface of the tank. Moisture content was monitored at a series of TDR probes (as indicated by the numbers in the figure).
|
|
|
Theoretical Analysis of Horizontal Flow within the Capillary Fringe
The flow behavior shown in Fig. 1 and 2 strongly contrasts the common conceptualization of predominantly downward vertical fluid flow through the unsaturated zone, with transition to fully three-dimensional flow only below the water table. The examples provided above also demonstrate clearly the potential importance of the interplay of the CF with fluid above and below the water table. Moreover, these results challenge traditional definitions of the "unsaturated zone," in that complex distribution of soil moisture, as well as horizontal flows, are seen to occur both above and below the water table.
Regarding the experiments shown in both Fig. 1 and 2, it might be conjectured that in the absence of the artificial inlet and outlet boundaries, no driving force acts to carry the tracer from below the water table up into the CF. However, Silliman et al. (2002) demonstrated that (in similar experiments) heterogeneity can lead to fluid movement, distinct from the effects of the boundaries, from below the water table into the CF.
So, referring to Fig. 1 and 2, what indeed causes the flow patterns? We note that the CF region, while at negative pressure head for the water phase, contains high moisture content to a certain height above the water table. A high moisture content ensures that the water phase in the porous medium remains connected throughout the CF, thus allowing relative permeability to remain high within the CF. In the case of the flow cell shown in Fig. 2, for example, the height of this high moisture content region is about 20 cm; in the Fig. 1 flow cell, the CF extends from the water table to the top of the sand (and the moisture content in this cell was measured to be as high as 84% near the surface). The high relative permeability of the sands above the water table combined with the boundary conditions present in both reservoirs (i.e., constant head up to the height of the water in the reservoirs, and free atmospheric, functionally no-flow boundaries above these water levels), leads to relatively simple numerical models demonstrating flow patterns similar to those shown in the figures. Indeed, flow patterns similar to those observed in the experiments were demonstrated numerically in the work by Luthin and Day (1955) and in more recent numerical work that we have performed (Silliman et al., 2002) as discussed below.
In addition, early Russian literature also recognized the possibility of flow in the CF. In particular, the textbook by Gersevanov and Pol'shin (1948) described capillary siphoning above the core of a soil dam. They focused on the case when there is no ambient groundwater flow and a barrier (dam core) blocks groundwater passage from one reservoir to another. Another comprehensive textbook by Vedernikov (1939) develops a mathematical method (essentially, a two-dimensional Green–Ampt model), which can be used to quantify bulk saturated flow below the water table, together with horizontal flow in the CF. A discussion of the Vedernikov model is given in (the English version of) the book by Polubarinova-Kochina (1962). More recently, Kacimov and Nicolaev (1995) integrated the Vedernikov (analytical) model with a numerical approach (based on a finite element formulation of the Richards equation) to quantify such flows. Also, following the results shown in Fig. 1 and 2 (Silliman et al., 2002), Kacimov (unpublished data, 2003) derived a simple solution in terms of the Vedernikov model that aims to quantify the amount of water from the feeding reservoir that is transported from the groundwater zone into the CF.
Having recognized the flow dynamics that can occur in the vicinity of the CF, the next important question to ask is, Are these flow dynamics of any real significance in real field situations? In fact, considerations from analytical and numerical modeling indicate that at the field scale, the CF generally constitutes a relatively minimal contribution to regional volumetric flow. For example, Bouwer (1978)(p. 243) estimated that for a 50-m-thick, unconfined aquifer consisting of medium sand, lateral flow in a 20-cm-thick CF would constitute only 0.4% of the flow below the water table, but he also noted that lateral flow in the CF could be as much as 30% of that below the water table, when considering flow toward a tile drain in a 2-m-thick agricultural loam with a 60-cm-thick CF. In agreement, numerical modeling by Silliman et al. (2002), extrapolating the systems shown in Fig. 1 to the field scale, indicated that the CF contributes a relatively minimal, or even negligible, amount to volumetric flows at the regional scale (see the following section). Hence, from the standpoint of considering the flux of water moving through the CF, one might conclude that in many situations, regional horizontal transport within the CF is commonly negligible. From the viewpoint of considering water flux at the local scale, contaminant transport, and the potential for bioremediation in the vicinity of the water table, however, these results indicate that the CF may play an important and unrecognized role.
|
NUMERICAL SIMULATION OF FLOW PROCESSES IN THE CAPILLARY FRINGE |
|---|
|
TOPABSTRACTINTRODUCTIONDEFINITION OF THE CAPILLARY...FLOW WITHIN AND ACROSS...EXPERIMENTAL DEMONSTRATIONSNUMERICAL SIMULATION OF FLOW...CHEMICAL AND MICROBIAL TRANSPORT...DISCUSSION AND CONCLUSIONSREFERENCES |
|---|
Most relevant to dynamics such as those shown in Fig. 1 and 2 are numerical models that use a fully saturated–unsaturated model of flow within the combined vadose zone, CF, and saturated groundwater zone. As an example of one such model, Silliman et al. (2002) used the Hydrus-2D software package (Simunek et al., 1999), with the flow system for the experiment shown in Fig. 1, to reproduce numerically the resulting flow field. Hydrus-2D treats a finite element solution of the Richards equation for variably saturated porous media. In the simulations, the CF was treated as a tension-saturated zone with a hydraulic conductivity value equal to that of the saturated zone. This simulation exhibited qualitative behavior quite similar to that shown in Fig. 1, including the strong vertical flux near the inlet and outlet boundaries, and horizontal and vertical migration of the tracer within the CF.
As discussed above, extrapolation of the systems shown in Fig. 1 to the field scale indicates that CF flow, while of local importance, contributes insignificantly to regional volumetric flow. This can be demonstrated quantitatively using, for example, Hydrus-2D. Silliman et al. (2002) simulated horizontal flux through a coupled saturated and partially saturated medium for a series of (homogeneous) sediment types ranging from sand to clay. The flow domain was 100 m in the horizontal direction, while the vertical dimension of the grid varied as a function of the desired saturated zone thickness. All boundary conditions were identical to those for the experiment in Fig. 1, except that the same mean horizontal gradient was applied in both the partially saturated and saturated zones. The total horizontal flux above the water table was estimated and compared with the flux observed below the water table for aquifer thicknesses of 1 to 100 m. The calculated percentages of total horizontal flux in the CF, relative to the region below the water, was as high as approximately 6%, for a 1-m-thick loam aquifer, and as low as about 0.04% for a 100-m-thick sand aquifer. Thus, these calculations confirm that, in general, and at the regional scale, lateral flow within the CF does not contribute significantly to horizontal flow through an aquifer system. However, as discussed above, lateral flow in the CF can be significant at local scales and in the presence of aquifer heterogeneities. Moreover, as we emphasize below and as has been demonstrated by stochastic studies of the impacts of the CF on flow and transport in heterogeneous porous media (Lu and Zhang, 2002, 2003; Zhang and Lu, 2002), even a relatively minimal contribution to flow in the CF can translate into significant transport processes within the CF.
It is less obvious whether continuum models based on the Richards equation could be successfully employed to simulate the air-entry effects illustrated in Fig. 3 through 5. It is recognized, for example, that the Richards equation is based on an assumption of free movement of the gas phase within the medium. As such, it is apparent that the Richards equation is not capable of simulating the details of the movement (and entrapment) of water and air in complex heterogeneities such as shown in Fig. 3. However, it is common practice to include hysteretic effects in the pressure–moisture relationships used in association with the Richards equation (e.g., Simunek et al., 1999). It remains unclear whether addition of hysteresis at the local scale would be sufficient to allow prediction of regions of trapped air below the water table and the maintenance of high moisture contents in coarse sands above the water table related to heterogeneities.
Another combined experimental and numerical analysis to examine the movement and interaction of saltwater and freshwater in coupled saturated–partially saturated porous media was given by Thorenz et al. (2002). In this study, the software package RockFlow (Kolditz et al., 1999) was used for the numerical simulations. RockFlow is based on a hybrid finite element method (FEM), which enables the use of one-, two-, and three-dimensional elements in three-dimensional space. RockFlow uses a dynamic hierarchical grid adaptation strategy that readjusts the grid during runtime; it consists of multiple kernels for full or partially saturated single-phase flow, multiphase flow, and transport of one or more components. In the application of Thorenz et al. (2002), particular emphasis was placed on flow phenomena in the CF, in the presence of a saltwater–freshwater interface. Both laboratory experiments and numerical simulations demonstrated clearly, in both qualitative and quantitative agreement, that significant lateral flows, and coupled density-driven flow effects, may take place within and across the CF. Further, the inclusion of multiphase flow within this modeling approach allows the potential for explicit analysis of effects such as air-entry barriers.
An alternative to continuum-scale models such as those discussed above is offered by network models. Network-scale models are particularly useful for investigating small-scale (or pore-scale) flow dynamics in porous media, including processes such as drainage and imbibition. Ronen et al. (1997) used a detailed network model by Blunt and Scher (1995) to capture key features of the CF. The simulations accounted for capillary-controlled fluid displacement processes in the presence of a gravitational field. Structures generated by the simulations illustrated the irregularity of the upper "surface" of the CF and regions of trapped air within the water phase above the water table. Moreover, analysis of horizontal cuts through the three-dimensional CF structure demonstrated that "islands" of water and trapped air can coexist in the CF; connections among islands of water permit horizontal flow in the CF. In fact, the simulations indicated that island coverage of about 50% leads to a connected phase, in agreement with percolation theory considerations (Zallen and Scher, 1971). The simulations were in qualitative agreement with field measurements reported by Ronen et al. (1997).
These network model simulations also agree qualitatively with the experimental systems represented in Fig. 1 through 5. Specifically, the network model permits not only the prediction of horizontal flow within the CF (and guidance on the definition of the maximum height above the water table at which horizontal flow is likely to be important), but also the explicit prediction of trapped air phases below the water table and the maintenance of saturated large pores at significant distances above the water table in the presence of heterogeneity in pore throat diameters.
|
CHEMICAL AND MICROBIAL TRANSPORT THROUGH THE CAPILLARY FRINGE |
|---|
|
TOPABSTRACTINTRODUCTIONDEFINITION OF THE CAPILLARY...FLOW WITHIN AND ACROSS...EXPERIMENTAL DEMONSTRATIONSNUMERICAL SIMULATION OF FLOW...CHEMICAL AND MICROBIAL TRANSPORT...DISCUSSION AND CONCLUSIONSREFERENCES |
|---|
As a result of such flow phenomena, the evidence provided above suggests that the CF might be a highly active zone in terms of the geochemistry and microbiology of the subsurface. Fluid fluxes within this zone will involve a complex combination of flux across the upper region of the CF (e.g., downward flux related to infiltration or upward flux due to evapotranspiration), horizontal flux due to regional hydraulic gradients, and flux across the lower boundary of the CF (upward or downward flux dependent upon the motion of the water table and local heterogeneity). Chemical fluxes within this zone are even more dynamic, including the pathways identified above plus diffusive exchange in the liquid and gas phases within both the groundwater zone and the vadose zone.
As shown in Fig. 2, infiltration patterns through the vadose zone are altered significantly as advancing fluid reaches the CF. Detailed analyses of wetting front instabilities and finger propagation through the partially saturated zone, and finger divergence, stabilization, and dissipation through the CF, have been documented in laboratory-scale experiments by Cho and de Rooij (2002) and Simmons et al. (2002). The CF has also been shown to influence and be influenced by the transport of solutes from the surface to the groundwater (e.g., Kinouchi et al., 1991). Lehman et al. (1998) argued that variations in water content and matric potential in the CF region influence the mixing regime of solutes during transport through the CF. Other experiments have demonstrated features of trichloroethylene (TCE) transport across the CF (Jellali et al., 2003), effects of changes in surface tension induced by surfactant as it moves vertically and horizontally through the CF into the saturated zone (Henry and Smith, 2002), and the nature of transverse dispersion across the CF (Klenk and Grathwohl, 2002). In this latter study, it was found that dispersivities for TCE are larger in the CF than in fully saturated porous media because of increased tortuosity caused by entrapped air.
Gas-phase transport above the water table and within the CF has also been investigated. For example, McCarthy and Johnson (1993) discussed the transport of volatile organic compounds across the CF and concluded that a one-dimensional approximation of the vertical transport across the CF can be used, with the dominant vertical transport mechanism being molecular diffusion. Caron et al. (1998) worked with the mass transfer of CO2 gas across the CF and suggested that the CF offers significant resistance to mass transfer, affecting the fate of CO2 transport. Affek et al. (1998) suggested that high CO2 concentrations in the CF may indicate a decrease in the diffusivity of the CF relative to the unsaturated zone. This may be attributed to the phase occupancy of the CF (e.g., water vs. gas in the unsaturated zone).
There is ample field evidence (by direct sampling of water and sediments, as well as circumstantially), supported by laboratory studies, that the CF plays an important role with respect to chemical transport within and below the water table. For example, the CF has been shown to significantly affect degradation processes. Ostendorf et al. (1995) suggested that microbiological activity near the CF had the effect of reducing the O2 demand of a deicing agent (calcium magnesium acetate) on groundwater. Zaidelman et al. (1997) reported that maximum biochemical decomposition of organic matter within peaty soils treated with a cover farming system was observed within the CF. Lahvis et al. (1999) studied the aerobic biodegradation and volatilization rates of gasoline hydrocarbons near the water table by extrapolating gas transport rates through the capillary zone. The results of this study indicate that aerobic biodegradation and volatilization of gasoline hydrocarbons within the CF can be an important pathway to the natural attenuation of hydrocarbons at gasoline spill sites.
Direct field measurements have also demonstrated how the high moisture content sustained in the CF can aid movement of pesticides along preferential pathways (Haria et al., 2003), how the CF can act to concentrate Cr and result in contamination of groundwater (Khan and Puls, 2002), and how salt can become concentrated in the CF region due to irrigation and upward water flux from the water table (Slavich et al., 2002). From these studies, it is clear that the CF actively impacts movement of dissolved chemicals and chemicals in the gas phase, both below and above the water table.
We note here peripherally that there has been extensive work that examines how LNAPLs can become distributed within the CF. Clearly, variations in the structure of the CF can influence the distribution and displacement of the LNAPL (e.g., Jawitz et al., 1998). The buoyancy forces can act to carry a lighter-than-water cosolvent into the CF during displacement of the resident groundwater. Subsequent flooding can then cause these buoyancy forces to trap the cosolvent in the CF, creating problems with the removal of the cosolvent from the aquifer. In parallel, circumstantial field evidence suggests that heterogeneous flow and transport processes in the CF are likely to form local traps for LNAPLs in the liquid phase and dense volatiles in the gas phase (Ronen et al., 1997).
Although less well documented within the literature, it is apparent to those who have drilled in relatively shallow unconsolidated sediments that the CF is quite active biologically and geochemically—one of the common means of detecting the depth of the mean water table during drilling is through observation of the change in color of the cuttings in the region of the CF and water table. This color change implies the presence of a significant change in the geochemical and/or microbial characteristics within this part of the CF. Moreover, substantial quantities of gas bubbles present in the immediate vicinity of the water table, resulting from geochemical and/or microbial activity, have also been reported (e.g., Ronen et al., 1989). Hendry et al. (2001) measured microbial respiration rates through a 3.2-m-thick "mesocosm," and found nonnegligible respiration rates in the CF, which increased upwards into the partially saturated zone.
Finally, a series of recent laboratory experiments has demonstrated the ease with which bacteria can be transported from the groundwater zone into the CF by advective motion of the pore water (Dunn et al., 2003, unpublished data). Figure 6 illustrates one result from this work within a laboratory flow cell similar to those shown in Fig. 1 and 2. Utilizing fluorescing bacteria (the bright areas within Fig. 6), it was shown that microbes introduced into the inflow reservoir (i.e., at and below the water table) actively migrated into the CF under horizontal hydraulic gradients. In addition, the migration of the bacteria was observed to be substantially influenced by the presence of a trapped air phase below the water table and a coarse sand lens above the water table that remained at high moisture content due to the effect of air-entry barriers.
|
|
DISCUSSION AND CONCLUSIONS |
|---|
|
TOPABSTRACTINTRODUCTIONDEFINITION OF THE CAPILLARY...FLOW WITHIN AND ACROSS...EXPERIMENTAL DEMONSTRATIONSNUMERICAL SIMULATION OF FLOW...CHEMICAL AND MICROBIAL TRANSPORT...DISCUSSION AND CONCLUSIONSREFERENCES |
|---|
Several studies have documented the hydraulic impact of the CF on such processes as fluctuation of the water table and watershed response to precipitation events. Furthermore, a number of studies have demonstrated (on simplified systems), or implied, significant impact of the CF on chemical fluxes between the groundwater zone and the vadose zone. However, to date, there has been only limited study of flow and transport mechanisms specifically in the CF and specifically at the local scale. We suggest that physical heterogeneity in the sediments within the vicinity of the water table can lead to (i) increased flow and exchange of chemical constituents between the CF and the underlying saturated zone, (ii) preferential transport of chemicals moving into the CF during infiltration events, (iii) enhanced horizontal chemical flux above the water table, providing an opportunity for significant horizontal motion of contaminants without the possibility for sampling via groundwater piezometers, and (iv) increased contact between gas (both trapped and free flowing) and liquid phases in the region bounding the water table. Such phenomena may drive a number of transport and reaction processes, including (i) nutrient and O2 availability to microbes in both the CF and the upper portion of the groundwater zone, (ii) geochemical alteration of the CF and the shallow regions of the groundwater zone, and (iii) delivery of volatiles within the groundwater zone into contact with the air phase in the vadose zone.
As has been shown in many studies, the vertical motion of a fluid front through the vadose zone tends to form strong fingers with the associated chemicals distributed unequally at the upper surface of the CF. The process of redistributing that flux at the upper surface of the CF will have significant impact both on the distribution of biological activity within the fringe and on the flux of water and chemicals entering the water table. As the CF represents one of the most active zones of biodegradation of chemical contaminants, such redistribution of the contaminant, nutrients, O2, and bacteria become of critical concern in evaluating the probable flux of the contaminant across the water table.
Given the importance of the behaviors discussed in this paper with respect to a number of processes near the water table, it is worth reconsidering, briefly, the definition of the zone bounding the water table, at the local scale. While defining the CF is relatively straightforward for a homogeneous medium, an appropriate definition of the CF appears to be somewhat elusive in the presence of heterogeneity and trapped air and water phases above and below the water table. For example, a definition based on minimum height to residual saturation (as presented above) is problematic in the presence of trapped air phases on both sides of the water table. Specifically, zones of sediments may be at residual moisture content at or below the water table (due to air-entry barriers), thus making this definition inadequate for heterogeneous systems in the presence of a fluctuating water table. Similarly, a definition of the CF based on maximum height to residual moisture content will incorporate, within the CF, zones for which horizontal connectivity of the liquid phase (and thereby potential for horizontal flow) is essentially absent.
Upon reflection, it is believed that two critical defining characteristics of this zone bounding the water table, as described herein, are (i) the potential for horizontal flow and (ii) the potential for the coexistence of two phases (gas and liquid) within the pores. Therefore, we suggest that the partially saturated fringe, or PSF, may be a beneficial concept in these discussions. The upper limit of the PSF may be defined as the maximum height above the water table at which horizontal connectivity of the liquid phase is sufficient to allow for the possibility of horizontal flow. Further, the definition of the PSF must recognize explicitly or implicitly the potential for incorporation of discontinuous regions of trapped gas in the regions immediately above and below the water table. Of the definitions of the CF presented to date within the literature, it would seem that definitions based on connectivity of saturated pores, such as provided by network models, provide the most promise for describing the PSF in terms of fluid, chemical, and microbial interactions in the vicinity of the water table.
In the final analysis, it is becoming quite clear that, regardless of exact definition, the local-scale behavior of the CF, or more appropriately the PSF, plays a major role in defining the geochemistry and microbiology near the water table as it represents a zone of active mixing among waters, chemicals, and microorganisms derived from recharge, influx boundaries, hydraulic exchange with groundwater, diffusion, and gas exchange with both the vadose and saturated zones. Thus, the CF or PSF may have far greater impact than previously perceived on sample collection and interpretation, transport of contaminants and nutrients near the water table, and design of remediation strategies. These assessments are critical to management and exploitation of aquifers.
|
ACKNOWLEDGMENTS |
|---|
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONDEFINITION OF THE CAPILLARY...FLOW WITHIN AND ACROSS...EXPERIMENTAL DEMONSTRATIONSNUMERICAL SIMULATION OF FLOW...CHEMICAL AND MICROBIAL TRANSPORT...DISCUSSION AND CONCLUSIONSREFERENCES |
|---|
