CONCEPTUAL MODELS OF CHLORINATED SOLVENTS
IN KARST AQUIFERS
Chlorinated solvents generally enter the subsurface environment as DNAPL and migrate downward and laterally until local conditions favor their accumulation (Schwille, 1988; Cohen and Mercer, 1993; Pankow and Cherry, 1996). Such accumulation can occur in any of several phases: DNAPL, vapor, sorbed to solids, or dissolved in ground water. Wherever and in whatever phase chlorinated solvents accumulate in the subsurface, they function as secondary sources for dissolved contaminant plumes. In the dissolved phase, chlorinated solvents can enter drinking-water supplies and threaten human health.
Major controls of the movement and ultimate fate of chlorinated solvents in the subsurface are (1) the physical and chemical properties and mass of specific contaminants released, (2) the areal extent and rate of contaminant release, and (3) the nature of the hydrogeologic environment into which the contaminant migrates (Cohen and Mercer, 1993; Pankow and Cherry, 1996). For a chlorinated-solvent release in karst, specific factors that control the residence time of contaminant accumulations and the concentration and movement of dissolved-phase contamination include:
In principle, knowledge of the above factors would greatly improve prospects for assessing (1) the rate of release of dissolved-phase contaminant to major ground-water flow systems, (2) the distance and direction of dissolved-phase contaminant movement, (3) the degree of dilution at different points along the dissolved-phase contaminant flow path, and (4) whether contaminant concentrations in ground water are likely to increase or decrease through time. In practice, information on these controls is commonly sparse or absent. The identity and amounts of contaminant and details about how, when, and where the release occurred are poorly documented at many sites (Pankow and Cherry, 1996). Even presumably simple hydrogeologic settings can turn out to be difficult and expensive to characterize in detail (Cherry, 1996). Karst settings are almost never simple, and there are currently no widely disseminated and applied standards for characterizing DNAPL spills in karst (Barner and Uhlman, 1995).
A necessary first step toward improving site characterization for chlorinated-solvent spills in karst settings is the development of conceptual models of where the contaminants are likely to have accumulated and where and how they may be moving. A good conceptual model provides numerous working hypotheses that can be evaluated and refined as new information becomes available. The absence of an adequate conceptual model commonly leads to wasted effort and expense as data are collected which do little to illuminate the problem at hand. Standard techniques of site characterization developed for aqueous-phase contaminants or for porous granular media may provide irrelevant or erroneous results at DNAPL sites in karst settings (Quinlan and Ray, 1991; Cohen and Mercer, 1993; Barner and Uhlman, 1995).
This section presents five conceptual models of DNAPL accumulation in karst settings. The models emphasize DNAPL accumulation in different compartments of the subsurface environment. The emphasis on DNAPL emplacement reflects the preliminary nature of the five conceptual models and the complexity of chlorinated-solvent contamination and karst hydrogeology. An emphasis on the behavior of dissolved-phase contamination might be preferable in terms of contaminant delivery to drinking water (Sid Jones, Tennessee Department of Environment and Conservation, written commun., 1997). However, current understanding of the behavior of chlorinated solvents in karst is not sufficient to support generalizations about the movement and behavior of dissolved-phase chlorinated solvents without first considering the location of subsurface contaminant sources. The five conceptual models therefore assume that, subsequent to initial release, the distribution of DNAPL in the subsurface will largely determine the distribution and movement of other phases. The models were developed for the karst environments of Tennessee, but the concepts presented in this section are intended to be transferable to similar karst settings in adjacent states and applicable in other areas.
The five conceptual models of DNAPL accumulation in karst settings (fig. 12) are:
The conceptual models presented in this section are scale neutral. There is no minimum amount of DNAPL that could be stored in any of these environmental compartments, and the maximum amounts are a function of the size and nature of the release and the hydrogeologic character of specific sites. The models are mutually compatible in that more than one model may be applicable to a given site.
Numerous local and regional studies throughout Middle and East Tennessee indicate the regolith is highly variable and heterogeneous. The regolith typically consists of clays and silts that may be mixed and interbedded with sands, gravels, and rock fragments. Depending on the local setting, the regolith may be composed of residuum, alluvium, or colluvium, singly or in combination (Bradley, 1984; Hanchar, 1988; Moore, 1988; Webster and Bradley, 1988; Bradfield, 1992; Haugh and others, 1992; Haugh, 1996; Rutledge and Mesko, 1996).
In the regolith, DNAPL can migrate through macropores, fractures, and intergranular pores. Flow through macropores and fractures is important in fine-grained layers where DNAPL may not be able to enter the intergranular pores (Helton, 1987; Cherry, 1989).
Several mechanisms work to trap DNAPL in the regolith. DNAPL can be retained by capillary forces as residual DNAPL under both unsaturated and saturated conditions. DNAPL can pool on top of layers that are lower in permeability relative to the overlying layer and provide a capillary barrier to further downward movement of DNAPL. Low-permeability layers can be perennially saturated, perennially drained, or alternately saturated and drained. If low permeability layers contain fractures which pinch out with depth, DNAPL may pool in these fractures.
Low permeability layers must be free of erosional or depositional gaps and fractures for pooling of DNAPL to occur. In most cases, the low permeability layers will deflect the downward movement of DNAPL, but will not be of sufficient lateral extent and composition to serve as a significant barrier to downward migration of DNAPL. Where these layers are discontinuous laterally, they can cause horizontal spreading of DNAPL, which can continue moving downward once the edge of the restrictive layer is reached, resulting in a complex DNAPL distribution. Typically, many small pools in discontinuous lenses and fractures would be expected to form. Contrasts in permeability can be important because even small scale variations play a major role in determining the DNAPL distribution in the regolith.
Capillary forces in any porous medium that DNAPL can enter will trap a certain amount as residual DNAPL. In the vadose zone, in the rare case of dry conditions in Tennessee regolith, DNAPL will be the wetting fluid and will be retained as films and wetting rings coating the media. The amount held as residual DNAPL will increase with decreasing intrinsic permeability, effective porosity, and moisture content (Cohen and Mercer, 1993). More commonly, the vadose zone will be partly saturated with water and DNAPL will be retained as nonwetting ganglia in the pore throats and bodies of the media. In the saturated zone, DNAPL will be retained as isolated ganglia in the large pore body spaces. Residual saturation will be less in the unsaturated zone because DNAPL drains more easily in the presence of air than in a water-saturated system.
Macropores and fractures may be important pathways for DNAPL movement in clay-rich regolith. As with porous media, residual DNAPL will be held in these fractures by capillary forces as disconnected blobs and ganglia. Values of residual saturation in fractured clays will be less than in porous media (Pankow and Cherry, 1996) and will increase with decreasing fracture aperture (Schwille, 1988). DNAPL pools will accumulate in fractures that pinch out sufficiently with depth to provide a capillary barrier.
Large DNAPL pools in the regolith are expected to be rare in carbonate settings. Many small pools in course-grained layers within fine-grained layers may be typical. The amount of product released along with the local thickness and composition of the regolith will determine how significant regolith trapping will be at a site. The regolith is typically thinnest in the inner Central Basin, Cumberland Plateau, and outer Central Basin and thickest in the Highland Rim and Valley and Ridge (table 4). Substantial pooling on a low permeability layer in the regolith has been documented in case study 26-505 (described later in this report), located in The Barrens area of the Eastern Highland Rim.
In the situation where DNAPL has pooled on top of a low permeability layer, the residence time and removal processes depend on where the pool is located. The two main processes that determine the residence time are volatilization and dissolution. The DNAPL may be located above the water table, below the water table, or within the zone where the water-table surface fluctuates. In the vadose zone, volatilization is the dominant process for removal of DNAPL. After volatilizing into the soil gas, the vapor will dissolve into infiltrating water to contribute to dissolved-phase contamination of ground water. Dissolution also will occur from the DNAPL pool into infiltrating water. In the saturated zone, dissolution into the ground water is the dominant process of DNAPL depletion. A DNAPL pool located in the zone where the water-table surface fluctuates would be depleted through both volatilization and dissolution. DNAPL pools trapped in the vadose zone have shorter residence times than pools trapped below the water table. However, residence times in the vadose zone may still be of sufficient duration to present a long-term source of dissolved-phase contamination. The same processes just described also apply to residual DNAPL in intergranular pores and fractures.
Residual DNAPL will have a shorter residence time than DNAPL in pools because the higher ratio of surface area to volume of residual DNAPL increases the rates of the processes that deplete the DNAPL. Results of laboratory and modeling experiments show that residence time on the order of months to years can be expected for residual DNAPL. Residence time for thin, horizontal pools can be expected to be on the order of decades to centuries. DNAPL pools in clay fractures can be depleted by matrix diffusion in timeframes on the order of weeks to a few years (Pankow and Cherry, 1996).
If the regolith is sufficiently thin and the contaminated area limited, excavation and removal of the DNAPL is effective. If pools are found on top of low permeability layers at shallow depths, DNAPL can be successfully excavated, significantly reducing the source mass. More typically, the DNAPL will be distributed in many small pools that will be difficult to find. Patterns of residual DNAPL will be similarly complex, making location and remediation difficult. Solvent that has diffused into low-permeability layers will be slow to diffuse back out.
DNAPL contamination at land surface and in the shallow soil zone can migrate down through the regolith to the top of the underlying bedrock. In this report, "top of rock" refers to the surface between the regolith and bedrock. In karst settings, this surface is commonly irregular, highly weathered, and variable in depth. The top of rock is part of a transitional zone (the subcutaneous zone or epikarst) that includes weathered rock fragments in the regolith and dissolution openings within the upper part of the bedrock. The transitional epikarst zone is commonly 3 to 10 m thick and extends above and below the top of rock (Quinlan, 1989).
For DNAPL to accumulate at the top of rock, the DNAPL must pass through the regolith and encounter a low-permeability pooling site at the bedrock surface. Even small volumes of DNAPL will in general have the potential to migrate down to the top of rock (Cohen and Mercer, 1993; Pankow and Cherry, 1996). At top of rock, DNAPL may accumulate in pools where differential weathering or structure has created irregularities in the bedrock surface.
For a given DNAPL release, the relative importance of pooling at top of rock will be influenced by the thickness and physical properties of the regolith, the bedrock lithology, and the geologic structure. Thin, permeable regolith will allow DNAPL to reach top of rock more easily than thick regolith with high residual saturation or numerous impermeable layers. Rocks with low secondary porosity or in which dissolution openings in epikarst pinch out with depth (Williams, 1983, 1985) will trap DNAPL more effectively than rocks with efficient hydraulic connections between their surface and underlying bedrock aquifers. DNAPL is more likely to be trapped by flat-lying or gently dipping rocks than by steeply dipping rocks, especially in cases where secondary porosity develops preferentially along bedding planes.
All of the carbonate formations of Tennessee have the potential to trap at least some DNAPL at their upper surfaces. In the Central Basin, the thin soils, gently dipping strata, and numerous outcrops of confining units, such as the Hermitage Formation, are conducive to top-of-rock pooling. Areas of the Highland Rim also have high potential to trap DNAPL at the bedrock surface, especially where point-recharge features are locally uncommon. DNAPL pooling at top of rock is less likely along the coves and escarpments of the Cumberland Plateau where high sinkhole density (Crawford and Veni, 1986) and close hydraulic integration between sinkholes and karst conduits (Crawford, 1987) provide efficient pathways for water and contaminants to move below the regolith-bedrock interface. Similarly, the steeply dipping strata and dissolution-enlarged bedding planes of carbonate units in the Valley and Ridge have high potential to transmit DNAPL beneath the top of rock.
The top of rock can be perennially saturated, perennially dry, or alternately wet and dry. For perennially saturated DNAPL pools, dissolution is the main mechanism of depletion. A DNAPL pool has a lower surface-area to volume ratio than has residual DNAPL. Only the pool surface and margins can come into contact with flowing water. Flow velocities through the regolith and the rate at which the DNAPL-water boundary layer is flushed are likely to be limiting factors for DNAPL dissolution. Residence time for DNAPL pools that are permanently below the water table are likely to be relatively long (years to decades). Top-of-rock pools that are exposed to air will be depleted mainly by volatilization. In most cases, the vapor is likely to remain in the subsurface as a persistent source of aqueous contamination. In areas such as the Eastern Highland Rim where the top-of-rock zone is an important aquifer (Haugh and others, 1992), DNAPL pooling at the bedrock surface could have a substantial effect on ground-water resources.
Location and removal of DNAPL at top of rock is complicated by increased depth to bedrock, topographic complexity of the bedrock surface, and distance between the initial release and the site of subsurface accumulation. The average depth to bedrock and, perhaps more important, the complexity of the top-of-rock geometry are two critical factors that can be evaluated in the early stages of a site investigation. Where epikarst is poorly developed, maps of top-of-rock topography, contaminant concentrations, and geologic structure may be useful tools for locating DNAPL. Epikarst development is commonly greatest in relatively pure limestones and dolomites, but impure carbonates such as the Warsaw Limestone and Fort Payne Formation of the Highland Rim can have well-developed epikarst (Burchett and Hollyday, 1974; Haugh and others, 1992). If DNAPL is found in a relatively simple depression in the bedrock surface beneath relatively shallow regolith, direct excavation of the contaminated regolith is an effective mitigation method.
In many cases, thick regolith, highly developed epikarst, or migration along capillary barriers will interfere with the location or removal of DNAPL. In such cases, evaluating contaminant concentrations in ground-water or soil samples and techniques such as soil-gas analysis may be effective methods for estimating DNAPL location. If DNAPL has not migrated far from the site where it was spilled, it may be possible to find part of the DNAPL mass using these techniques. If located, DNAPL may be at least partly removed by direct pumping, air-sparging, or other techniques, depending on local conditions.
A diffuse-flow zone in bedrock occurs where many small fractures are present but where dissolution is minor, or where dissolution-enlarged fractures are filled with granular material. A typical environment for this situation is fractured shales and carbonates or conduits that have been filled with sediment washed in from land surface. Ground water moves through a diffuse network of small fractures or through granular material within conduits rather than through discrete conduits. Flow through the diffuse network may converge on larger conduits with more concentrated flow. In this situation, DNAPL present in a diffuse-flow zone could provide a source of aqueous-phase contamination to the more rapid flow in a large conduit.
DNAPL will migrate down through the network of fractures until the capillary resistance becomes too high for continued downward movement. DNAPL will then pool in the fractures. Within a network of small fractures, large vertical accumulations of DNAPL are possible.
Fractured shales in the Valley and Ridge form locally important aquifers (Hollyday and Hileman, in press). Zones of diffuse flow may also occur within the shales, sandstones, and carbonates of the Cumberland Plateau, Highland Rim, and Central Basin. Case study 82-516 (described later in this report) provides an example of a site where DNAPL has been documented to be pooled in a fracture network in the Sevier Formation in the Valley and Ridge. At this site, most ground-water flow occurs through the fracture network. Chieruzzi and others (1995) describe a site in southwestern Kentucky where DNAPL is being recovered by wells in the Ste. Genevieve and St. Louis Limestones from a diffuse-flow zone of pores and vugs in permeable beds or bedding planes. At this site, diffuse-flow blocks are believed to be bounded by a poorly connected system of conduits.
The residence time of DNAPL pooled in zones of diffuse flow is determined by dissolution and matrix diffusion. Dissolution into the actively flowing ground water is slow because of the small surface area available for DNAPL pooled in fractures. The rate of ground-water movement also limits the amount of DNAPL that can be depleted by dissolution. In cases where DNAPL is pooled in a network of fractures, a much larger surface area is available for matrix diffusion. Matrix diffusion can be an important process if the rock matrix has significant primary porosity. In Tennessee karst, significant primary porosity is likely only in a zone surrounding fractures and bedding planes where a significant width of dissolution has left a broad band of insoluble residue within an impure soluble rock. DNAPL depletion by matrix diffusion from all but the smallest (less than 0.1 mm) aperture fractures may take years or longer (Pankow and Cherry, 1996). After DNAPL has been depleted from fractures, solvent dissolved in the matrix pore water diffuses back into the fracture, serving as a persistent source of dissolved-phase contamination.
When found in fractures, some of the DNAPL may be removed by pumping, thereby reducing the source mass. However, the complex nature of DNAPL distribution in a fracture network makes the DNAPL mass difficult to find. Even when found, some of the DNAPL may be in dead-end fractures or fractures removed from active ground-water flow.
Flow in dissolution-enlarged openings is one of the most characteristic and best studied features of ground water in karst terranes. Much effort has been devoted in recent decades to the establishment of general principles governing this complex and variable phenomenon (White, 1969, 1977, 1988; Ford and Williams, 1989; Quinlan and others, 1992). In general, the best understood aspects of karst-conduit flow concern flow in large conduits (caves) directly accessible to researchers (White and Deike, 1989). The hydraulics of the much more numerous small (millimeter to centimeter scale) conduits are known only in the most general terms (Ford and Williams, 1989, ch. 5-6), though there is no basis for assuming that flow in small conduits is less important than flow in explorable caves in terms of overall aquifer hydraulics or contaminant transport. For many karst aquifers, with few or no explorable caves, small conduits represent most or all of the conduit-flow system. Empirical methods, such as dye tracing and spring sampling, are the most powerful tools currently available for characterizing karst-conduit flow systems. These methods are directly applicable to contaminants in aqueous solution (Quinlan and Ewers, 1985; Quinlan and others, 1992; Field, 1993), but provide only a limited basis for assessing DNAPL movement and storage within conduits.
The characteristic size range of karst conduits, typically on the order of millimeters to tens of meters (Ford and Williams, 1989, ch. 7), is too large for capillary forces to significantly restrict DNAPL movement. Thus, DNAPL will freely flow into most of the open conduits it encounters. Once it enters a conduit, DNAPL will flow along the conduit floor, collecting in cracks, pits, or other depressions. The movement, transformation, and persistence of a given DNAPL mass in a karst conduit depends on such case-specific factors as the size and shape of the conduit, the topography of the conduit floor, the degree of residual or sedimentary fill, and the position of the conduit relative to water-table fluctuations. All of these factors exhibit enormous variation in karst terranes (White, 1988; Ford and Williams, 1989).
Karst conduits develop along preferential pathways between areas of ground-water recharge and discharge, and are enlarged by dissolution (White, 1988). Karst landforms, such as sinkholes, that concentrate recharge may be closely integrated with conduits and thus provide direct routes for contaminants to conduit flow systems (Quinlan and others, 1992; Field, 1993).
Once DNAPL enters a conduit, any irregularity or obstruction in the conduit floor or inflection in conduit orientation will provide a place for the DNAPL to pool. Studies of springs show that karst conduit systems retain significant volumes of easily displaced vadose storage (water) in pools (Joseph Meiman, U.S. National Park Service, oral commun., 1997). The same pools will also hold DNAPL and may have enough volume to contain large DNAPL spills. Low spots along the floor of a conduit where DNAPL can pool (sumps) may be without capillary cracks, leaving all the pooled DNAPL exposed to flow. Such conduits can develop along the tops of confining units, or through massive limestone. Other conduits have sumps coinciding with cracks in the conduit floor through which DNAPL could migrate downward and out of the conduit-flow system.
All karst settings in Tennessee contain conduit-flow systems capable of storing DNAPL. The best developed conduit systems are in the Central Basin, the sinkhole plains of the Highland Rim, the coves and escarpments of the Cumberland Plateau, and the limestone and dolomite outcrops of the Valley and Ridge.
DNAPL pools in karst conduits can be perennially submerged in water or periodically exposed to air. DNAPL pools exposed to air will be depleted through volatilization. The fate of the resulting vapor will depend on the air-flow characteristics of the conduit. In many cases, vapor-phase chlorinated solvents may be as persistent in karst conduits as in other parts of the ground-water system. On the other hand, cave systems with high air flows (Bruce Zerr, Oak Ridge National Laboratory, oral commun., 1996) may efficiently route chlorinated-solvent vapors to the surface.
Flow velocities in conduits are high relative to ground-water flow rates in other settings (White, 1988; Quinlan and others, 1992). Recurrent inputs of fresh water tend to flush aqueous-phase contaminants and maintain a high concentration gradient close to the DNAPL pool. Dissolution is more likely governed by the maximum rate of dissolution into pure water than by the replacement of saturated solution in contact with the DNAPL. Frequent flushing of the DNAPL-water boundary layer would encourage relatively rapid dissolution and a short residence time (from weeks to years). However, Field (1993) notes that karst conduits can rapidly deliver significant quantities of contaminant to a discharge point yet still retain enough contaminant in storage to result in long-term ground-water contamination.
The configuration of DNAPL pools in conduits determines the surface area exposed to flowing water, and, in turn, the residence time of the pool. Karst conduits can be long and narrow or nearly round in cross section; they can be nearly level or vertical; they can be open or filled with unconsolidated material (White, 1988; Ford and Williams, 1989). All of these factors will affect residence time and movement of DNAPL in karst conduits. In narrow passages with appreciable dip, DNAPL will drain down-dip to fill the lowest depressions. Narrow, deep pools of DNAPL exposed to flow only along their upper edge will dissolve more slowly than wide, shallow pools.
The amount and pore-size distribution of unconsolidated fill in a conduit influence the ability of DNAPL to enter the conduit and the distribution and eventual fate of any DNAPL that does enter. Unconsolidated fill may largely exclude DNAPL from some parts of a conduit system. Conversely, infiltration of DNAPL into coarse-grained cave sediment or residuum, burial of a DNAPL pool by sediment, or the formation of DNAPL sludges in combination with organic matter will reduce the pool's exposure to flowing water (Palmer, 1984, 1986), potentially increasing residence time to decades or longer.
Pooling of DNAPL in active karst conduits can result in delivery of dissolved contaminants to springs and to any wells that intercept the conduits downgradient of the pools. Because of the high flow velocities and convergent flow patterns of karst conduits, such contamination may be concentrated in relatively few wells or springs. Where DNAPL residence time is short because a wide, shallow DNAPL pool is exposed to large rapid flow, the mass flux of aqueous-contaminant will be correspondingly high. Dilution by uncontaminated water from converging conduits may partly attenuate concentrations in such cases, but concentration may still exceed drinking-water standards.
Ground-water levels in karst conduit systems can fluctuate several meters to tens of meters during periods as short as a few hours or days (Milanovic, 1981, p. 128-132; Quinlan and Ewers, 1985; Wolfe, 1996a, b). Many karst conduit systems have several levels of conduits and discharge points whose relative importance varies with fluctuations in ground-water levels (Palmer, 1986; White, 1988). At high water levels, discharge is commonly routed to overflow springs that remain dry at lower ground-water levels. Ground-water mixing may deliver dissolved contaminant to these overflow springs. Such mixing will probably reduce contaminant concentrations, but not necessarily below acceptable levels. Only dye traces and sampling under a variety of hydrologic conditions, based on careful inventory of springs, seeps, and other karst features, will reveal the presence of multilevel discharge points.
The complexity and high localization of karst conduit systems make the prospects of finding a DNAPL pool in conduit storage unlikely. One approach to remediation may be to identify discharge points and potentially affected parties using dye traces and water sampling and to provide alternative water supply until concentrations fall below drinking water standards. Depending on the details of the contaminant release and the conduit system, such reduction may take years or decades. Any conduit system that has transported or stored DNAPL in the past may still contain DNAPL pools. High-discharge events may exhume such isolated pools and deliver additional contaminants to discharge points.
Most karst conduits ultimately discharge at springs. Contaminated springs may be distant from the source of contamination or in nonintuitive locations. Comprehensive karst inventories, dye-tracing studies, and long-term spring sampling under different hydrologic conditions offer the best chance of determining the destination of dissolved contaminants (Quinlan and Ewers, 1985; Quinlan, 1989). Temporal patterns of contaminant-mass flux at springs may provide information about DNAPL distribution relative to the water table (Sid Jones, Tennessee Department of Environment and Conservation, written commun., 1997).
The high densities and low viscosities of chlorinated solvents cause these compounds to migrate downward until they encounter openings too small to enter. Under certain conditions, this downward migration can take DNAPL to fractures that are relatively isolated from major ground-water flow zones. In contrast to the previous cases discussed in this section, pools of DNAPL in isolated fractures have minimal interaction with flowing water. Reduced exposure to flowing water has major implications for DNAPL residence time, mitigation, and delivery to drinking-water supplies.
Every karst aquifer has a lower boundary below which flow is greatly reduced. In general, smooth, abrupt lower boundaries are probably much less common than are rough, gradational boundaries. Karst develops through the interaction of atmospheric water with soluble rock (White, 1988). In many cases, dissolution and the resulting secondary porosity are concentrated in the upper parts of a carbonate rock unit and decrease with depth (Ford and Williams, 1989, p. 158-162). The base of karstification is typically a zone in which the karst-conduit system propagates downward through the progressive enlargement and integration of discrete voids which initially have only poor interconnection. This zone may be at considerable depth below the zone of major flow within the aquifer (Ford and Williams, 1989, p. 177-178). The network of conduits and fractures above the base of karstification provides potential flow paths for DNAPL through the major flow zone to the smaller, more isolated voids below.
Regionally, epikarst systems may be thought of as karst aquifers which have been partly exhumed through weathering and erosion. Near their lower limits, epikarst systems exhibit a decrease in the frequency and size of dissolution openings analogous to the pattern described above (Williams, 1983). Dissolution openings in epikarst can store significant quantities of ground water, in part because relatively few such openings are directly connected to underlying aquifers (Williams, 1983, 1985). Blind epikarst openings that decrease in width toward their lower limits may store DNAPL in isolation from ground-water flow, provided the openings are not filled with regolith that is impermeable to DNAPL.
Similar to karst aquifers, noncarbonate fractured-rock aquifers commonly have secondary porosity and associated ground-water flow that decrease with depth (Davis and Turk, 1964; Davis, 1988). In the case of fractured, noncarbonate rocks, forces such as tectonic stress initiate fracturing, but progressive unloading through erosion and uplift is a major mechanism through which these incipient fractures are sufficiently enlarged to convey significant ground-water flow (Trainer, 1988). DNAPL entering the more permeable, highly fractured part of a fractured-rock aquifer has potential to migrate downward to the smaller, less numerous fractures at the base of the aquifer where it will be isolated from flow.
Migration of DNAPL to fractures isolated from ground-water flow is most likely where ground-water flow is concentrated in the upper part of the aquifer and secondary porosity extends to considerable depth below the major flow zone. These conditions are met in the several karst aquifers of the Central Basin, the carbonates and fractured shales of the Valley and Ridge, the cavernous dolomites of the western toe of the Blue Ridge, the fractured sandstones of the Cumberland Plateau (Brahana, Macy, and others, 1986), and the fractured crystalline rocks of the Blue Ridge (Brahana, Mulderink, and others, 1986). In the Highland Rim, potential for pooling in isolation from ground-water flow depends largely on the local thickness of the Mississippian carbonate sequence, which decreases where the Chattanooga Shale confining unit is near the surface. The high hydraulic gradients and well-developed conduit systems of the Cumberland Plateau escarpments make isolation of DNAPL pools from ground-water flow less likely than in other karst settings.
A DNAPL pool isolated from ground-water flow will have a long residence time (on the order of decades or longer). The major mechanism for removal will be diffusion into adjacent fractures and primary pores. The low rate of local flow will limit flushing of the aqueous phase, resulting in a relatively low concentration gradient near the DNAPL mass and a correspondingly low rate of diffusion and dissolution. Migration of aqueous phase to zones of higher ground-water flow will occur through diffusion. Depending on the flow system, the rate of diffusion may be small relative to the flow, resulting in greater or lesser attenuation by dilution. Whatever the attenuation achieved by dilution, aqueous-phase contamination of ground water from DNAPL pools isolated from ground-water flow is likely to persist for many decades.
Isolation from major ground-water flow does not preclude severe local impacts on ground-water resources. In areas where ground water is scarce, such as in parts of the Cumberland Plateau and Blue Ridge, water-supply wells are commonly drilled into low-yielding units that would be considered confining units elsewhere. Pooled DNAPL in such settings might contaminate existing wells, further depleting an already scarce resource. DNAPL masses that sink below zones of ground-water flow are unlikely to be located. If found, mitigation will be limited by the same factors that operate in the case of DNAPL in bedrock diffuse-flow zones discussed previously.
| Next | Title Page | Table of Contents | List of Illustrations | List of Tables | Conversion Factors and Vertical Datum | Glossary |