| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
a USGS, 345 Middlefield Rd., Menlo Park, CA 94025
b EGI, University of Utah, 423 Wakara Way, Suite 300, Salt Lake City, UT 84108
c USGS, P.O. Box 2230, Idaho Falls, ID 83401
* Corresponding author (jrnimmo{at}usgs.gov)
Received 13 November 2001.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONSITE DESCRIPTIONMETHODSRESULTSDISCUSSIONCONCLUSIONSNOTESREFERENCES |
|---|
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONSITE DESCRIPTIONMETHODSRESULTSDISCUSSIONCONCLUSIONSNOTESREFERENCES |
|---|
The slow-flow generalization may not hold for geologically diverse unsaturated zones or for large, nonuniform inputs of water over the land surface. Preferential flow paths can transport water and contaminants horizontally to adjacent regions or vertically to the aquifer far sooner than might be predicted based on bulk medium properties and Richards' equation. Another important effect of preferential flow is that a relatively small fraction of the subsurface medium interacts with the contaminants, which limits adsorption and other attenuating processes.
Fluid transport that is difficult to explain by common unsaturated-flow concepts is suggested in several recent cases of contaminant detections in the unsaturated zone at substantial horizontal distances from the source of contamination. At a location about 100 m away from a waste burial site in the Amargosa Desert of Nevada, tritium has been detected in plants and in boreholes within the 108-m-thick unsaturated zone at much greater than ambient concentrations (Prudic and Striegl, 1995). Horizontal liquid movement along hypothetical preferential flow paths might plausibly cause such transport (Striegl et al., 1996; Prudic et al., 1997). In the Mojave River Basin of California, Izbicki et al. (1998) found evidence of unsaturated-zone horizontal movement of liquid water 45 m from a stream channel, on the basis of the detection of atmospherically derived tritium presumed to have infiltrated through the channel. At a nearby site, Izbicki (personal communication, 2000) noted evidence of gaseous chlorofluorocarbon transport, horizontally within the unsaturated zone to distances of 400 m from a landfill. On the Snake River Plain in southeastern Idaho near the Idaho National Engineering and Environmental Laboratory (INEEL), detection of chlorofluorocarbons, tritium, carbon-14, and chlorine-36 in the unsaturated zone suggests they may be transported vertically to the water table at about 200 m depth and horizontally as much as several kilometers in a few years (E. Busenberg and L.D. Cecil, personal communication, 1998). Contaminants in groundwater and inferred water recharge temperatures at this location suggest that recharge through the unsaturated zone can be rapid (Busenberg et al., 1993). In another INEEL study, a 1994 field experiment known as the Large-Scale Infiltration Test, water that infiltrated from a 200-m-diameter basin was found to move at a rate of 5 m d-1 (Burgess, 1995; Wood and Norrell, 1996; Dunnivant et al., 1998).
At waste-disposal or spill sites, tracer studies are commonly based on detection of contaminants that derive from the pollution source. Conclusions from such studies sometimes raise questions about other possible sources of contamination closer to the point of detection, other possible modes of transport (e.g., through the atmosphere or on contaminated waste-conveying vehicles), the uncertainty of timing, and the magnitude of contaminant release. Field experiments with deliberately introduced chemicals as tracers avoid some of these uncertainties and afford fairly precise knowledge of the position and temporal nature of the source.
This paper describes an artificial tracer experiment near the Subsurface Disposal Area of the Radioactive Waste Management Complex at the INEEL (Fig. 1). This facility has a variety of radioactive and hazardous chemical wastes buried at shallow depths on the eastern Snake River Plain. Some contaminants have traveled down to the large and heavily used Snake River Plain aquifer (Laney et al., 1988). At this semiarid location, seasonal streamflow, snowmelt, and local runoff episodically generate large quantities of infiltrating water for periods of days or weeks. Particular concerns arise from seasonal artificial diversions of the Big Lost River to enhanced depressions called spreading areas, which serve as infiltration ponds. Contaminants from the Subsurface Disposal Area are known to have migrated to considerable depths in the unsaturated zone, so this additional water may enhance their further migration toward the aquifer.
|
This evidence is inconclusive on the issue of significant horizontal flow, and on the issue of the ability of contrasting layers to inhibit vertical flow to the aquifer. Our investigation tests both of these issues, with emphasis on the hypothesis that short-term, large-volume infiltration may cause rapid, long-range transport through the unsaturated zone near the INEEL Subsurface Disposal Area.
|
SITE DESCRIPTION |
|---|
|
TOPABSTRACTINTRODUCTIONSITE DESCRIPTIONMETHODSRESULTSDISCUSSIONCONCLUSIONSNOTESREFERENCES |
|---|
The unsaturated zone near the Subsurface Disposal Area is highly stratified and about 200 m thick, as illustrated in Fig. 2 and 3 (Anderson and Lewis, 1989; Anderson et al., 1996). Additional scale drawings and other stratigraphic information are in the works of Barraclough et al. (1976), Rightmire and Lewis (1987a)(b), and Anderson and Liszewski (1997). The unsaturated zone comprises low-permeability layers of fine sediments or dense basalt, and layers of fractured or rubbly basalt that conduct water easily when wet. Depending on the magnitude and prevalence of critical hydraulic processes within these layers, they may accelerate or retard contaminant transport, concentrate or dilute contaminants, and establish the dominant movement as vertical or horizontal.
|
|
Basalt flows and other volcanic deposits combine into a basalt flow group, a complex assemblage of overlapping flows and deposits related to a single eruption. Three important basalt flow groups, called A, B, and C, are within about 70 m of the surface at the Subsurface Disposal Area.
Rubble zones are common along the rapidly cooled margins of basalt flows. Basalt rubble and scoria (cinders) have been described in some wells near the Subsurface Disposal Area, but have not been systematically characterized. The potential for scoria is greatest in basalt-flow group C because its vent is thought to be located in this vicinity (Anderson and Liszewski, 1997).
Between some individual basalt flows, and especially between flow groups, sedimentary interbeds consist of well sorted to poorly sorted deposits of clay, silt, sand, and gravel. These interbeds are named after the basalt flow groups they lie between, for example, AB, BC, and CD (Fig. 2). Their depths below land surface vary topographically, but the two most prominent interbeds under the Subsurface Disposal Area, the BC and CD interbeds, are traditionally labeled with depths of 34 and 73 m, respectively. Near the Subsurface Disposal Area, they tend to dip in an easterly direction, the BC interbed about 3.8 m km-1 and the CD about 4.7 m km-1 on average (Anderson and Lewis, 1989). Their thicknesses vary in accordance with the topography of underlying basalt flows, and in some places they pinch out to zero thickness. Immediately below a particular basalt flow, an interbed is likely to have a baked zone of sedimentary or volcanic material whose properties may have been altered by exposure to the heat of fresh lava.
Barraclough et al. (1976) noted that at the interface between basalt and sedimentary layers, the permeable openings into the basalt are partially filled by sediment, the nature of which has been further investigated by Rightmire (1984). A layer of basalt in which the fractures are filled with fine sediments may be particularly low in effective permeability.
Hydrology
Surface Water
The Big Lost River is the principal stream within the INEEL and a major source of recharge to the Snake River Plain aquifer. To minimize the likelihood of floods at INEEL facilities, a diversion was constructed on the river in 1958 and enlarged in 1984. During high flows, water can be diverted into four spreading areas designated SAA, SAB, SAC, and SAD (Fig. 1). The estimated total capacity is 2.8 x 107 m3. Barraclough et al. (1967) estimated characteristic ponded infiltration rates of 2.5 x 10-6 m s-1 (0.22 m d-1) for SAA and 9.1 x 10-6 m s-1 (0.79 m d-1) for SAB. As Fig. 4 shows, the average annual flow into the spreading areas from 1965 through 1999 was 4.9 x 107 m3. From 1977 to 1981, the spreading areas received little water, with none at all in 1978 and 1979. There also was no inflow from 1988 to 1994. During the wet years of 1982 through 1985, approximately two-thirds of the Big Lost River flow that entered the INEEL was diverted to the spreading areas (Pittman et al., 1988).
|
Perched water might enable high-permeability features such as fractures and rubble to divert flow horizontally. Such flow would probably parallel the axis of a basalt flow. Near the Subsurface Disposal Area, this direction would be generally northward because these basalt flows erupted from vents on a volcanic rift zone south of the Subsurface Disposal Area. Lateral flow within or adjacent to sedimentary interbeds is likely to move in the direction of dip of the interbeds. The horizontal continuity and dip of some of the interbeds suggests that perched water derived from the spreading areas could move eastward, toward and beyond the Subsurface Disposal Area.
Beneath the Subsurface Disposal Area, some perched water, possibly nearly all of it, may derive from the spreading areas and the Big Lost River rather than from local precipitation. Rightmire and Lewis (1987b), for example, found relatively light abundances of 18O and 2H in perched water beneath the Subsurface Disposal Area. This suggests a source at higher altitude than the Snake River Plain, consistent with the high-altitude streamflow present in the Big Lost River and spreading areas.
Properties and Behavior of the Aquifer
Rubble and fracture zones provide the main conduits for groundwater flow in the saturated zone at the INEEL. Anderson et al. (1999) found the effective hydraulic conductivity to range from about 3 x 10-8 to 8.4 x 10-2 m s-1. They suggested that values less than about 3 x 10-4 m s-1 are associated with dikes and thick (>9 m) basalt flows. Greater values are associated with rubble zones and various types of fractures. The prevailing direction of regional flow in the aquifer is toward the southwest.
Fluctuating recharge rates can temporarily alter the flow rate and direction in the aquifer. Barraclough et al. (1976) suggested that periods of high infiltration cause local reversals in water-level gradients near the Subsurface Disposal Area sufficient to move water toward the northeast, against the prevailing direction. They judged infiltration from the spreading areas to be the predominant influence in this process, exceeding the influence of infiltration from local precipitation and the Big Lost River. Rightmire and Lewis (1987a)(b) summarized related evidence concerning a fluctuating groundwater mound beneath the spreading areas that alters gradients in the saturated zone. By mapping the rise in water levels that occurred during the unusually wet period between July 1981 and July 1985, Pittman et al. (1988) found further support for this hypothesis.
Expected Behavior of Water from Spreading Areas
Based on available evidence from earlier investigations, and consistent with the understanding of hydrologists knowledgeable about this site, our initial conceptualization of the flow system was as follows (see also Fig. 2). We assumed that when spreading areas become filled, water infiltrates over the course of a few weeks, as long as ponded surface water remains. Initially much of the infiltrating water is taken up by unsaturated-zone pore space, including dead-end fractures. The water percolates downward, continuing for some time after surface water is gone. Vertical flow occurs mainly through highly conductive fractures, taking a few days to reach the water table. Some of the water perches in or on layers of lesser effective vertical conductivity. Most of the water goes down to create a mound, a few meters thick, on top of the aquifer. The mound spreads radially while the spreading areas have standing water, and for a few days or weeks after they empty. The mound dissipates into the aquifer and its water and solutes are carried along with the prevailing Snake River Plain aquifer flow. The perched zones may persist for a few months or longer, until horizontal and vertical flow spreads them enough to leave the porous materials unsaturated.
|
METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONSITE DESCRIPTIONMETHODSRESULTSDISCUSSIONCONCLUSIONSNOTESREFERENCES |
|---|
|
|
Field Application, Sampling, and Monitoring
The tracer was applied to the spreading areas from a pontoon-type power boat. Fifty-kilogram bags of tracer were partially opened at one end and suspended over the bow of the boat, allowing the material to gradually spill out and the propeller to disperse it as the boat moved forward. The course followed with the boat during this operation was intended to broadly cover the area of ponded water, so that the concentration might be fairly uniform throughout each spreading area. On 21 June 1999, 350 kg of tracer was applied in SAA. The following day, 300 kg was applied in SAB. On 23 June, 83 surface water samples were taken at various locations and depths throughout both spreading areas to assess mixing and approximate concentrations. After the sampling on 23 June, an additional 25 kg of tracer was applied to the northeastern lobe of SAB (Fig. 1), which had filled with water overnight.
A monitoring network of existing wells, both up- and down-gradient, was selected to characterize the vertical and horizontal extent of infiltrated spreading-area water. After positive detection of tracer in perched zones, additional wells near the Subsurface Disposal Area were added to the sampling scheme. In December, sampling began at additional Subsurface Disposal Area wells and at monitoring wells (B series and C series wells in Fig. 1) at the location of the Large-Scale Infiltration Test (Wood and Norrell, 1996).
Each well was sampled using a dedicated pump or bailer, or both, following U.S. Geological Survey protocol (Mann, 1996). Extreme precautions were taken in cleaning field equipment to avoid cross-contamination between wells (see Appendix). Samples were collected in clear plastic Nalgene 30-mL bottles, externally rinsed after collection, then labeled and shipped to the analysis lab at the University of Utah. Well conditions were noted in a field logbook.
Initial sampling frequency during June through September 1999 was weekly, biweekly, or monthly, depending on well location, available personnel, and anticipated travel times predicted on the basis of previous studies. After September, sampling was carried out monthly.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONSITE DESCRIPTIONMETHODSRESULTSDISCUSSIONCONCLUSIONSNOTESREFERENCES |
|---|
Substantial amounts of perched water were found in the unsaturated-zone wells. Those at the Large-Scale Infiltration Test site (south of the Subsurface Disposal Area and east of the spreading areas) were not sampled until 183 d after tracer introduction. Water was encountered in five of the 38 wells that were probed, and water samples were recovered from three of these (C09C11, C09C11A, and B12011). On the following sampling visit, 227 d after tracer introduction, water samples were recovered from all five of the wells with standing water. These five wells were completed in, and above, the base of the BC interbed, which comprises low-permeability silty-clay loam material. Wells that did not yield water were either completed in the basalts above the BC interbed or in the basalts below the BC interbed, so for any water that might cascade in, there was nothing to prevent immediate outflow through basalt fractures. Thus, the presence of standing water in five of the wells is consistent with there being lateral flow in the basalts from the spreading areas.
Measured tracer concentrations in the spreading-area surface-water samples averaged 16.6 µg L-1 (± 20.6 µg L-1 standard deviation, based on 8 samples) in SAA and 180.4 µg L-1 (± 99.0 µg L-1, based on 18 samples) in SAB. It is likely that concentrations in SAA were lower because of rapid infiltration between the time of application and the surface water sampling, and the fact that the SAA water was sampled 48 rather than 24 h after tracer introduction.
Tracer was detected in nine wells. Figure 7 shows measured concentrations vs. time. In the Snake River Plain aquifer at USGS-120, tracer was detected 9 d after introduction. The preceding sample, taken 6 d after tracer introduction, showed no tracer. The breakthrough curve for USGS-120 has a double peak, as is not uncommon for tracer studies involving preferential flow (e.g., Mohanty et al., 1998). In perched water from basalt immediately above the BC interbed at UZ98-2, tracer was detected 18 d after introduction. It also was detected in perched water from the BC interbed at the nearby well UZ98-1, 65 d after introduction, but in this well the tracer may have been accidentally introduced when the well's casing was temporarily pulled to conduct a cross-hole tomography study, and then reinstalled just before the sample was collected. In the five sampled BC-level wells at the Large-Scale Infiltration Test site, tracer was detected at relatively high concentrations (12–65 µg L-1) in all water samples taken. Limitations of the sampling schedule prevent precise estimates of arrival times at these five wells, but clearly it took less than 183 d to travel this 1.3-km distance.
|
Measurements low enough that their difference from zero might not be significant, or where it is not obvious that their measurement uncertainty can be attributed to analytical uncertainty rather than variability in the measured population, may require more formal decision criteria before detections can be confirmed as true positives. Data from USGS-92 and USGS-120 fell into these categories and were evaluated as described in the Appendix.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONSITE DESCRIPTIONMETHODSRESULTSDISCUSSIONCONCLUSIONSNOTESREFERENCES |
|---|
Several hypotheses may account for the tracer's apparent up-gradient (northeastward) movement over the 0.2 km or more to USGS-120 from the spreading areas. First, significant movement of tracer within the unsaturated zone may have provided an up-gradient source to migrate downward and then back down-gradient to the well. Second, highly localized mounding, as previously hypothesized, may have temporarily reversed the prevailing flow at this location. There are not enough nearby aquifer wells to adequately identify small, localized changes in gradient and flow direction.
The lack of positive detections in the other 12 sampled aquifer wells (Table 1) limits the possible generalizations concerning aquifer flow. These 12 points of nondetection are 0.7 to 2.9 km away from the nearest point of tracer introduction. Two of these, USGS-9 and USGS-109, are close to SAC and SAD, which did not contain surface water during the experiment. From SAB, USGS-9 is 1.5 km down-gradient and USGS-109 is 1.8 km across gradient. The other locations of aquifer nondetection are near the Subsurface Disposal Area, and thus are in the direction opposite the prevailing aquifer flow. Therefore, this experiment does not confirm the hypothesized spreading-area-related reversal of groundwater flow (Barraclough et al., 1976) over distances >0.2 km. However, the possibility of considerable and rapid dilution of spreading area–derived water within the Snake River Plain aquifer makes for a significant probability that any tracer-tagged water that had traveled 0.7 km or more through the aquifer was diluted to below detection limits. Therefore the observed pattern of aquifer nondetections at horizontal distances >0.2 km does not refute the flow-reversal hypothesis.
|
At the CD interbed level, tracer was detected from one of the three sampled wells, USGS-92, directly under the Subsurface Disposal Area 1.3 km northeast of SAB and 2.2 km east of SAA. The detected concentrations (0.7 µg L-1 and less) are low but significant (see Appendix). Because they declined after the first detection, the concentration may have peaked at this point in <91 d. They suggest an average horizontal flow rate of at least 14 m d-1. This detection may indicate an isolated case of flow through a single channel. The experiment cannot rule out more widespread horizontal spreading at this depth, however, because so few CD wells were available for sampling. The other two CD wells, also under the Subsurface Disposal Area, may also have contained tracer before the sampling of these wells began or at undetectable concentrations. The results from USGS-92 demonstrate that perched water beneath the Subsurface Disposal Area comes in part from subsurface flow of spreading-area water.
Of possible driving forces for flow, if conditions are nearly saturated, gravity would be expected to dominate. Besides the average easterly dip noted above, the pattern of observed unsaturated-zone detections is consistent with the known elevation of interbed contacts at various points (Anderson et al., 1996). The average dip of the interbeds gives a modest driving force, being a component 0.004 times vertical gravity. Thus, if the component of gravity parallel to the interbeds is the chief driving force, the observed long-range horizontal transport suggests an overwhelming effective anisotropy in unsaturated-zone properties.
It is unlikely that horizontal convective transport rates >10 m d-1 in the unsaturated zone occurred directly through the fine-textured portions of the interbeds because of the low saturated hydraulic conductivity of these materials (ranging from 10-7 to 10-4 cm s-1, Perkins and Nimmo, 2000). It is more likely that the interbeds caused perching that extended upward into some portion of the overlying basalt. Rapid transport may be possible through interconnected rubble zones, tension cracks, or similar features if they are nearly saturated. There must also be substantial impediments to vertical flow, such as interbeds or dense basalt that cause perching that extends upward into the highly conductive layer or flow path. Given the observed lengths of individual basalt flows and volcanic rift zones, such horizontal movement of water over distances of a kilometer or more is plausible.
This evidence as a whole supports several particular hydrologic phenomena.
1. There can be rapid vertical transport under the spreading areas.
2. Substantial perching of water in the unsaturated zone results directly from the percolation of water from the spreading areas.
3. Over distances of a few kilometers and durations of a few months, horizontal transport occurs in perched water in or above both the BC and CD interbeds.
4. The horizontal spreading in perched zones is not uniform with direction.
These hydrologic phenomena suggest certain subsurface features that determine the character of flow. Sedimentary interbeds and impeding features of the basalts are imperfect barriers that permit substantial vertical flow, possibly because of interbed discontinuities and basalt fractures. This is consistent with earlier evidence showing that the interbeds impede vertical flow where they are continuous, for example, the Large-Scale Infiltration Test (Wood and Norrell, 1996c; Dunnivant et al., 1998). In spite of existing vertical fast flow mechanisms, at least some of the flow-impeding features are sufficiently effective and continuous that they divert substantial flow horizontally for considerable distances. Nonuniformities in both basalt and interbed topography are likely causes of the irregular spatial patterns of tracer detection.
The long-range horizontal flow inferred from unsaturated-zone tracer detections requires alteration of previous conceptualizations of subsurface contaminant transport that did not include this sort of flow. Horizontal movement by itself does not necessarily imply a greater likelihood that water-borne contaminants will be transported to the aquifer. However, it does put more water under the Subsurface Disposal Area than would occur without spreading-area input. This water may facilitate contaminant transport to the Snake River Plain aquifer by intercepting water-borne contaminants that would otherwise percolate into and be absorbed by the underlying sedimentary interbeds; it may also transport them downward through preferential flow paths that are active because of the greater amount of water present. The horizontal flow is likely to dilute contaminants and disperse them over wider regions of the unsaturated zone. This could provide greater opportunity for sorption, especially within the sedimentary interbeds, by reducing competition for sorption sites, increasing the availability of sorption sites, and changing solution chemistry to favor sorption. Therefore some effects of the inferred horizontal and nonuniform flow are likely to enhance and others to hinder the transport of contaminants to the aquifer.
Similar behavior concerning fast, long-range flow both horizontally and vertically in the unsaturated zone may occur at other locations with the necessary site characteristics. The INEEL tracer detections were at depths of 30 m and more, suggesting emphasis on arid and semiarid areas where the unsaturated zone is deep. It is important that the unsaturated zone be highly stratified, which is common, but usually not to the obvious extreme of the Snake River Plain. The unsaturated zone must have preferential flow paths such as fractures or coarse granular material, also common at many sites. Long-range horizontal continuity of individual preferential paths and impeding layers has importance that is hard to assess; such continuity could exist at the INEEL, though if essential, the number of such paths must be strikingly large to permit the multiple BC-level tracer detections (Fig. 1). Ponded water is needed to provide a large volume of inflow to the unsaturated zone, and episodic occurrence is important so that conditions are normally unsaturated. Features that might play this role, for example, ephemeral streams and ponds, playas, and artificial recharge ponds, are fairly common on plains and valleys, though less so in mountains.
Where all the relevant features are present in some degree, this sort of rapid, long-range flow might be expected. Alluvial valleys (e.g., Mojave desert, basins of basin and range topography) have many of these characteristics, though commonly with an unsaturated zone that contains only granular material. Horizontal flow is most likely where some of the granular materials are coarse, or underlain by fractured rock. Several sites of major contaminant transport concerns are of interest. At the Hanford Site in Washington, much of the unsaturated zone has depth in the 60- to 90-m range. It has less pronounced layer contrasts than the INEEL site and no significant fractured rock in the unsaturated zone, though there are strong textural contrasts of silt to gravel size, and caliche. These differences may not be very significant for transport rates. The Nevada Test Site has an extremely deep unsaturated zone in many places, and more diversity in layers than the Hanford site. Subsurface behavior there might easily be comparable to that found at INEEL, though ponded water would be more unusual. Yucca Mountain likewise has an extremely deep unsaturated zone and clearly lacks a major source of ponded infiltration. Its stratigraphy is analogous in many ways to the Snake River Plain. For example, the layer referred to as "paintbrush tuff" is less fractured than its surrounding layers and so has been hypothesized to play a role similar to that hypothesized for the INEEL interbeds in interrupting preferential flow and inhibiting downward solute transport. In both cases, these hypotheses may be invalid; evidence for such effects in paintbrush tuff is lacking (Flint et al., 2001), while our study shows that preferential flow can move downward through or past INEEL interbeds. Our experiment suggests that even if the impeding layer can allow substantial downward movement, it does not prevent long-range horizontal flow.
At smaller scales, similar fast-flow behavior may occur in unsaturated zones a few meters thick, as commonly exist in humid as well as arid climates. Small-scale stratification and preferential paths commonly exist, so it is entirely possible to have analogous phenomena over the shorter distances. Nimmo (2002) found that field data from a wide variety of field sites do show similar behavior and have transport speeds similar to those in this study.
|
CONCLUSIONS |
|---|
|
TOPABSTRACTINTRODUCTIONSITE DESCRIPTIONMETHODSRESULTSDISCUSSIONCONCLUSIONSNOTESREFERENCES |
|---|
Besides testing the chosen combination of tracer and techniques, this study was to assess long-range subsurface flow paths that might result from episodic ponded infiltration. The geologic nature, hydraulic properties, spatial distribution, source localities, and other characteristics of these flow paths remain of great interest and are poorly known. Results of the present experiment suggest several improvements for field experiments to explore these characteristics and processes further. The observed long distances and high speeds of travel suggest an extensive network of sampling wells, and a schedule that has even the most distant of these sampled early in the experiment. Multiple tracers, for example other naphthalene sulfonates (Fig. 6), are desirable for different portions of the surface water. With three tracers one could distinguish water from individual spreading areas A and B, and the Big Lost River. This would answer questions such as whether tracer in USGS 92 came from spreading area A or B. Repeat sampling of the tracer-containing surface water would give a better estimate of the initial concentration, for evaluation of attenuation and dilution. Besides offering a better understanding of subsurface preferential flow paths and their effect on solute transport, an improved study of this sort would allow the testing of important hypotheses concerning the permeability, continuity, and dip of interbeds, the role of rubble and fractures, and the partitioning of vertical and horizontal flow in the unsaturated zone.
This experiment shows that with a large input of water, the overall structure of the unsaturated zone of the Snake River Plain can generate rapid transport both vertically and horizontally. At least under ponded infiltration, fine-textured interbeds and layers of particularly dense basalt do not prevent the rapid, high-volume, vertical flow that is expected through fractured basalt. Some impediments to vertical flow, however, are effective in causing substantial perching and horizontal diversion. The observed rapid horizontal flow can persist over distances >1 km, entirely within the unsaturated zone, 100 m and more above the aquifer. Possible conduits for this horizontal flow include basalt fractures, tension cracks, lava tubes, or rubble zones positioned above low-permeability layers of basalt or interbeds. These observed behaviors are likely to enhance the vertical and horizontal spreading of contaminants in the subsurface, though by exposing the contaminants to a greater volume of sedimentary materials they may also have effects that tend to reduce aquifer contamination. Because the rapid vertical and horizontal transport in the unsaturated zone seems to be caused by natural features responding to a common sort of ponded infiltration, these results suggest that similar phenomena may occur at a variety of sites.
Quality Assurance
Samples were collected and analyzed according to the guidelines of the Quality-Assurance Program at the USGS INEEL Project Office, which includes: (i) analytical methods used by the laboratories, (ii) quality-control samples, (iii) review of analytical results, (iv) audits of field performance, (v) corrective actions to resolve problems with field and laboratory methods, and (vi) reporting of data. A chain-of-custody record tracked samples from collection to delivery.
Wells equipped with dedicated submersible pumps require decontamination of detachable sampling equipment. At the wellhead, a stainless-steel discharge pipe is equipped with a sample port and a gate valve to control the flow rate. The sampling equipment, discharge pipe, and fittings were rinsed with deionized water before and after installation at the wellhead. Subsequent flushing with purged water in amounts of about 2 to 30 m3 further reduced the possibility of cross-contamination with water from previously sampled wells. Additionally, before each sampling, at least three well-bore volumes of water were pumped from the well to remove any stagnant water. This purged water was disposed of to the ground. This is not expected to influence the subsurface movement of tracer because the amount of water was small (on the order of 103 m3) and was distributed over a substantial area of vegetated land under high-evapotranspiration conditions. For wells in which a bailer was used to collect a water sample, careful decontamination procedures were followed to prevent cross-contamination between well bores by rinsing with deionized water and wiping with lab tissues. Bailed samples were poured directly from the bailer into sample bottles after three consecutive rinses of the bailer with sample water. After sampling, the bailer was rinsed thoroughly with several liters of deionized water and placed in a clean plastic bag to prevent possible contamination during transport.
The quality-assurance program includes archival retention of replicate samples for evaluation of erroneous results, occasional submission for analysis of blind replicates (duplicate samples with different sample identification numbers), blank samples (deionized water labeled as a regular sample), equipment blanks (rinsate collected during decontamination procedures), and spiked samples (of known concentration). In general, about 15 to 20% of the samples that were analyzed for tracer concentration were dedicated to quality assurance.
This study included determinations of the analytical precision or uncertainty associated with the measurement of 1,5-naphthalene disulfonate concentrations using the high performance liquid chromatograph with reverse-phase column separation. The column was a reverse phase 50 by 4.6 mm Keystone BetaBasic-18 with 3-µm particle size (Thermo Hypersil-Keystone, Bellefonte, PA).1 The mobile phase consisted of a solution of 3.17 mM Na2HPO4, 6.21 mM KH2PO4, and 5.0 mM TBAP in 30:70% methanol/water. Test standards, consisting of nominal solution concentrations ranging from 0.1 to 100.0 µg L-1, were prepared from the industrial grade tracer that was used in the field experiment.
Table 2 and Fig. 8 present the measured results for determination of the analytical precision or uncertainty. The absolute error of a measurement, taken as the standard deviation (
) of eight replicate measurements of the same laboratory-grade standard, is affected primarily by the gain or sensitivity setting of the instruments. Between 0.1 and 0.5 µg L-1 it is essentially constant at about 0.07 µg L-1. Between 1 and 10 µg L-1 it also is essentially constant, at about 0.13 µg L-1. At higher ranges, higher gain settings result in higher absolute errors, as shown for nominal tracer concentrations of 50 and 100 µg L-1. To indicate relative error, absolute error estimates are normalized (/
) with respect to the measured means. Figure 8 shows that in the range of 0.1 to 100 µg L-1, the log of the relative error is nearly linear with respect to the log of measured concentrations, and decreases as the concentration increases. At 0.15 µg L-1, the lowest measured concentration, the relative error is 47% of the measured mean.
|
|
|
|
|
NOTES |
|---|
|
TOPABSTRACTINTRODUCTIONSITE DESCRIPTIONMETHODSRESULTSDISCUSSIONCONCLUSIONSNOTESREFERENCES |
|---|
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONSITE DESCRIPTIONMETHODSRESULTSDISCUSSIONCONCLUSIONSNOTESREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  C. L. Duke, R. C. Roback, P. W. Reimus, R. S. Bowman, T. L. McLing, K. E. Baker, and L. C. Hull Elucidation of Flow and Transport Processes in a Variably Saturated System of Interlayered Sediment and Fractured Rock Using Tracer Tests Vadose Zone J., November 20, 2007; 6(4): 855 - 867. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
