HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Abstract Figures Only Full Text (PDF) Free Alert me when this article is cited Alert me if a correction is posted Services Similar articles in this journal Similar articles in ISI Web of Science Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via ISI Web of Science (1) Citing Articles via Google Scholar Google Scholar Articles by Wilson, G.V. Articles by Gillespie, D.R. Search for Related Content PubMed Articles by Wilson, G.V. Articles by Gillespie, D.R. GeoRef GeoRef Citation Agricola Articles by Wilson, G.V. Articles by Gillespie, D.R.

DIVISION S-1-SOIL PHYSICS

Recharge from a Subsidence Crater at the Nevada Test Site

^a Desert Research Institute, currently with the USDA-ARS National Sedimentation Laboratory, 598 McElroy Dr., Oxford, MS 38655 USA
^b Harry Reid Center for Environmental Studies, P.O. Box 4009, Las Vegas, NV 89119 USA
^c Desert Research Institute, Hydrologic Sciences Division, 755 E. Flamingo Rd, Las Vegas, NV 89119 USA

wilson{at}sedlab.olemiss.edu

	ABSTRACT

TOP ABSTRACT INTRODUCTION Objective Materials and methods Results and discussion Conclusions REFERENCES

 
Current recharge through the alluvial fans of the Nevada Test Site (NTS) is considered to be negligible, but the impact of more than 400 nuclear subsidence craters on recharge is uncertain. Many of the craters contain a playa region, but the impact of these playas has not been addressed. It was hypothesized that a crater playa would focus infiltration through the surrounding coarser-grained material, thereby increasing recharge. Crater U5a was selected because it represented a worst case for runoff into craters. A borehole was instrumented for neutron logging beneath the playa center and immediately outside the crater. Physical and hydraulic properties were measured along a transect in the crater and outside the crater. Particle-size analysis of the 14.6 m of sediment in the crater and morphological features of the crater suggest that a large ponding event of 63000 m³ had occurred since crater formation. Water flow simulations with HYDRUS-2D, which were corroborated by the measured water contents, suggest that the wetting front advanced initially by as much as 30 m yr^-1 with a recharge rate 32 yr after the event of 2.5 m yr^-1. Simulations based on the measured properties of the sediments suggest that infiltration will occur preferentially around the playa perimeter. However, these sediments were shown to effectively restrict future recharge by storing water until removal by evapotranspiration (ET). This work demonstrated that subsidence craters may be self-healing.

Abbreviations: ET, evapotranspiration • NTS, Nevada Test Site • PA, performance assessment

	INTRODUCTION

TOP ABSTRACT INTRODUCTION Objective Materials and methods Results and discussion Conclusions REFERENCES

 
AN INCREASED INTEREST in the hydrologic processes of arid region infiltration and recharge has been caused by the need for safe disposal sites for hazardous and radioactive waste. The NTS has served as a principal disposal site for low-level waste from Department of Energy facilities since 1961. Tyler et al. (1996) examined 10 boreholes (seven shallow series and three deep series) to characterize the thick vadose zone of Frenchman Flat (Area 5) on the NTS. Secondary Cl^- bulges were evidence of recharge occurring during the last two glacial maxima, when the climate was wetter and cooler. Based on the calculated Cl^- fluxes, a paleorecharge rate of 0.004 to 0.008 m yr^-1 was estimated for the three deep boreholes. Current recharge was found to be negligible under natural conditions and topography of the alluvial fans.

Other studies have shown that recharge in well-vegetated, stable-fan environments of the NTS has been minimal or nonexistent for at least 6000 yr (Fischer, 1992; Winograd and Doty, 1980). However, the anthropogenic impacts of the 828 underground nuclear detonations between 1951 and 1992 on the NTS may have altered the hydrology such that localized recharge may occur. Of these underground tests, 260 tests were conducted below or within 100 m of the static water table. The nuclear explosion vaporizes and melts 500 t of rock for each kiloton of device yield. As the temperature and pressure within the cavity eventually decreases, the overlying rock and soil collapse, forming a rubble chimney. If that rubble chimney reaches the ground surface, a subsidence crater results. More than 400 nuclear subsidence craters exist at the NTS (Tyler et al., 1992).

Several studies have been conducted to assess the role of subsidence craters on potential recharge. In the 1960s, the Plowshare program began looking at nuclear subsidence craters as a possible method for enhancing recharge in the extremely arid NTS. The positive impact of enhanced moisture flux through these artificial recharge basins, however, was outweighed by the possibility of radionuclide mobilization from the cavity to the groundwater (Teller et al., 1968). Tyler et al. (1992) investigated Crater U3fd in Yucca Flat of the NTS to address the potential for enhanced recharge. Two boreholes were drilled to determine water content with depth, one beneath the low point of Crater U3fd and one representing an undisturbed condition immediately outside of the crater. Results showed that effectively no recharge was occurring beneath the undisturbed area adjacent to the crater where moisture infiltrated during precipitation events was soon recycled back into the atmosphere through ET. In contrasts, a range of recharge rates from 0.02 to 35 m yr^-1 was estimated for the crater using traditional Darcy's law and saturated hydraulic conductivity measurements. Soil water tritium concentrations and mixing models, based on the assumptions of steady-state one-dimensional flow, yielded a reduced range of 0.55 to 0.70 m yr^-1 for the vertical flux below U3fd. It was concluded that surface topography, coarse soils along the crater bottom, and periodic ponding events due to contribution of the surrounding drainage basin allow for deep infiltration and eventual recharge to the groundwater system.

Pohll et al. (1996) dynamically linked an overland flow model with a subsurface, unsaturated flow model to enhance simulation of recharge beneath NTS craters. The inflow to the crater via surface flow was converted to a pond depth by an empirical stage–volume relationship. Darcian flow through the variably saturated medium was simulated with the numerical model SWMS_2D (Simunek et al., 1992), a two-dimensional finite element approximation of Richards' equation. The soil was assumed to be isotropic, and the effects of hysteresis were modeled by the simplified approach of doubling the parameter for the wetting branch of the van Genuchten soil characteristic curve (van Genuchten, 1980). The simulations suggested that the flux at 10 m below the surface was temporally stable (mean value = 0.75 m yr^-1). The mean flux at 5 m below land surface was 1.4 m yr^-1, which is close to the estimates by Tyler et al. (1992). Pohll et al. (1996) suggested that significant recharge will occur beneath craters that experience ephemeral ponding events.

In addition to the concern for the impact of craters on recharge, there is concern for the long-term impact on waste management practices. The Area 5 Radioactive Waste Management Site includes shallow burial of low-level waste in trenches and transuranic waste in boreholes. Decomposition of waste and infilling of voids between waste packages lead to concern about the impact of runoff into subsided areas of the waste cells. To address this concern, a numerical simulation was conducted by Crowe et al. (1998) in which 2 m of ponded water was placed along the top boundary for a period of 24 h and then the head allowed to drop as the infiltration continued. A 1-yr period of moisture redistribution was simulated before these boundary conditions were repeated for three consecutive years. After this period of intermittent ponding, influx of water at the surface ceased and moisture in the domain percolated downward until reaching the water table. Travel times to the water table (250 m below land surface) ranged from 190 yr (1.3 m yr^-1) to 140 yr (1.8 m yr^-1) depending on the degree of moisture in the waste package. The latter result closely matched the prediction of 146 yr to the water table reported in the Area 5 performance assessment (PA) (Shott et al., 1997). However, the Area 5 PA did not consider the potential for recharge below existing craters in Area 5.

Hokett and French (1998) investigated the recharge potential beneath the U5a crater in Area 5 (Frenchman Flat) of the NTS using the SCS curve number approach to predict runoff events and HYDRUS-2D to simulate subsurface flow. This crater was chosen because its physical attributes suggested that it intercepted significantly more runoff from the surrounding drainage area than other craters, thus serving as a worst case for recharge. Extensive phreatophytic vegetation (salt cedar [Tamarix ramosissima Ledeb.]), large erosional gullies (5 m deep), and a playa near the crater center were evidence that this crater received substantial storm runoff water. While a crater is a unique geographic setting for a playa, the Crater U5a center meets all of the following criteria for a playa: ephemerally flooded, a vegetatively barren basin floor, a thin veneer of fine-textured sediment, and functioning as a sink for drainage water (Soil Science Society of America, 1997). Scanlon and Goldsmith (1997) reviewed the role of playas as recharge features in arid and semiarid environments and listed these scenarios: (i) playas act as evaporation pans, (ii) playas restrict recharge to their annular region, and (iii) playas focus recharge directly beneath the playa surface.

The 210-km² watershed upstream of U5a is poorly defined, with extensive areas of anthropogenic disturbance and highly permeable soils. Thus, Hokett and French (1998) concluded that it was not likely to contribute flows to the crater. Crater U5a has two deeply incised channels linking the crater through the 14.5-km² alluvial fan portion to the upstream portion of the watershed, which was used by Hokett and French (1998) to represent the contributing area to U5a. Based on measured precipitation, they predicted 13 ponding events in U5A for a 32-yr simulation period following crater formation in 1965.

These ponding events were used by HYDRUS-2D for modeling the recharge potential for U5a based on hydraulic properties selected from the HYDRUS-2D menu. The flow domain had a radius of 20 m and depth of 200 m, with the playa represented by a 10-m radius of silt loam to 3 m and loam to 10 m. Infiltration of the 13 ponding events was modeled with a 0.3-m head that extended 2 m beyond the playa (12-m total length) for the upper boundary condition. Following infiltration of each pond event, the upper boundary was changed to an atmospheric boundary condition with an annual average potential evaporation rate of 8 mm d^-1, based on pan evaporation measurements in the area (Magnuson et al., 1992). In addition, precipitation events during this 32-yr period that did not result in ponding were accounted for by including an average monthly precipitation rate to the upper boundary. A uniform matric head of -10 m throughout the domain was used as the initial condition.

Modeling results suggested that the playa focused infiltration through the coarse-grained material around the playa perimeter (2 m outside the playa), thereby enhancing the recharge potential. The fine-textured playa material restricted flow, while the surrounding sandy material provided a highly conductive pathway for deep water movement. Simulations suggested that the predicted ponds percolated to a depth of 129 m within the 32-yr period. Wetting front advancement resulted in a potential recharge rate of 3.7 m yr^-1.

While these simulations compared favorably with the work of Crowe et al. (1998), the simulated water content profile did not match the measured values in the borehole below the playa center. The lateral spread of the water plume that preferentially percolated around the annular area of the playa did not reach the playa center where the borehole was located. The lack of agreement between the measured and simulated water content and matric heads at the borehole location could be due to assumptions of initial conditions, boundary conditions, or ponding events. However, the sensitivity of the wetting front advancement and lateral spread to these factors was not evaluated.

	Objective

TOP ABSTRACT INTRODUCTION Objective Materials and methods Results and discussion Conclusions REFERENCES

 
The objectives of this study were to determine the potential for crater recharge from the U5a crater and the impact of sediment properties (i.e., playa region). It was hypothesized that if the surface runoff covered only the crater playa area, the low permeability surface would restrict deep percolation. Additionally, as the lateral extent of the pond outside the playa is increased, the maximum depth of water movement for a given volume of water will be reduced. Thus, the sensitivity to the assumptions on the lateral extent of the pond events (i.e., boundary conditions) used by Hokett and French (1998) was evaluated along with assumptions on the ponding events.

	Materials and methods

TOP ABSTRACT INTRODUCTION Objective Materials and methods Results and discussion Conclusions REFERENCES

 
Hydrogeologic Setting
The NTS lies within the northern edge of the Mojave Desert, 105 km northwest of Las Vegas, occupying an area of 3500 km². The lithology of the NTS can be divided into valley-fill alluvium, clastics, tuffs, and carbonates (Shott et al., 1997). The alluvial deposits are typically unconsolidated gravelly, loamy sands (Anderson, 1995). The upper alluvium is Paleozoic detritus and Tertiary rocks; deeper alluvium is predominantly tuffaceous. Maximum thickness is 730 m with an unsaturated zone generally more than 350 m thick. Average annual precipitation ranges from 8 to 25 cm, depending on elevation. Recharge principally occurs in the higher elevations of the mountains where water migrates vertically through the carbonate bedrock, then horizontally into the alluvial valley-fill (Domenico et al., 1964). Potential ET rates of 150 to 200 cm yr^-1 in the basins greatly exceed precipitation.

Frenchman Flat occupies the southeast portion of the NTS in Nye County, Nevada. Frenchman Flat is a closed drainage basin surrounded on all sides by mountains. Average annual rainfall at the Well 5B precipitation gage 2700 m southwest of Crater U5a is only 12.4 cm. This extremely arid environment has no perennial streams or standing bodies of water. Surface water flow occurs only infrequently and channels are rarely deeper than 1 m.

The Wishbone test that formed Crater U5a was detonated in 1965 at a working point depth of 175 m below land surface. The static water table was estimated to be 206 m below land surface. A topographic survey of Crater U5a conducted in 1996 by Bechtel Nevada Corporation showed a depth of 12.8 m, representing 14.6 m of infilling during the last 30 yr (Fig. 1) . The current crater morphology consists of numerous large gullies that cut through the side slope on the northern side, with a crater bottom that contains a near circular playa area that is surrounded by vegetation.

View larger version (46K): [in this window] [in a new window]   Fig. 1 Contour map of Crater U5a with sample locations along the transect identified by open circles and the location of the instrumented borehole identified by an X. Playa area is shadowed and contour values are reported in meters

Surface soil properties were also determined along a transect across the crater surface that cut across the center of the playa region (Fig. 1). These included particle-size analysis, bulk density, water retention, and hydraulic conductivity at saturation and at -15 cm matric head. These properties were determined at 16 stations spaced 10 to 15 m apart along the transect and at three stations (A, B, C) adjacent to the crater. To characterize the subsurface, sites along the transect were augered using a Gidding's probe. Due to the hardness of the fine-grained layers encountered, continuous core samples were not recovered. Eight locations were drilled to a maximum depth of 6.1 m and the cuttings sampled. Only particle-size analysis was conducted on the subsurface samples.

Particle-size analysis of bulk samples was accomplished using the hydrometer and dry sieve methods as described by Gee and Bauder (1986). Following the hydrometer measurements, samples were sieved through a stack of 63-, 125-, 250-, 500-, 850-, 1,180-, and 2360-µm particle diameter sieves. The hanging water column and pressure plate methods were used for water retention analysis as outlined in Klute (1986) on two undisturbed soil cores (5.3-cm i.d. and 6 cm long) from each transect station. Water retention was determined at matric heads of -15, -50, -100, -150, and -200 by the hanging water column method and at -500, -750, -1000, and -3000 cm by the pressure plate method.

Single-ring infiltrometer measurements were conducted at each station along the transect and at the three stations adjacent to the crater to determine the saturated hydraulic conductivity, K_s. A metal ring (24.7-cm diam.) was driven into the ground 10 cm, and a constant head of 4 cm was maintained by a Mariotte device. Infiltration measurements were recorded until a steady-state condition was observed. The tension infiltrometer (also known as disc permeameter) was used to measure the unsaturated hydraulic conductivity, K(h) following the procedures for confined (1-D) infiltration as described by Wilson and Luxmoore (1988). Tension infiltrometer measurements were made immediately following the constant head ring infiltrometer procedure at each station along the crater transect and at the three stations adjacent to the crater under a constant matric head of -15 cm until a steady-state infiltration rate was achieved. For confined infiltration, the steady-state infiltration rate equals the hydraulic conductivity for the prescribed matric head.

Numerical Modeling Methods
While vapor flow and nonisothermal conditions are generally important under arid conditions as at the NTS, there are other complexities, like spatial variability and media properties of the crater collapse zone, which could not be accounted for that are more significant. Thus, we restricted this analysis to processes that could be parameterized. The HYDRUS-2D code by Simunek et al. (1996) was used to solve Richards' equation for variably saturated flow:

(1)

where h is the matric head (L); C() is the specific water capacity (L^-1), which is the slope of the water retention function (h); and K(h) is the hydraulic conductivity function (L T^-1). HYDRUS-2D uses the van Genuchten (1980) model to represent the hydraulic functions (h) and K(h) in Richards' equation. Several researchers (Smettem and Kirby, 1990; Smettem et al., 1991; Wilson et al., 1992; Mohanty et al., 1997) have proposed methods for modeling K(h) and (h) for dual-porosity soils (macropore–matrix system). HYDRUS-2D provides a means of incorporating the effect of macropore flow on the hydraulic properties by making K(h) a two-region function (Wilson et al., 1992, Mohanty et al., 1997). K_s represents the hydraulic conductivity when all pores are contributing and K_k is the hydraulic conductivity after the macropores empty such that

(2)

(3)

(4)

where h_s is the air-entry value and h_k is the matric head at the measured matching point, K_k. The unsaturated hydraulic conductivity determined by the tension infiltrometer, , was used as the K_k value to enable modeling the crater properties as a dual-porosity medium.

	Results and discussion

TOP ABSTRACT INTRODUCTION Objective Materials and methods Results and discussion Conclusions REFERENCES

 
Site Characterization
The initial deep drilling beneath the playa revealed, for samples below 24 m, an increase in volumetric water contents of 38% over the undisturbed location, with a vertically averaged volumetric water content of 0.119 cm³ cm^-3 for the entire depth. The matric head of the samples below the playa ranged from -679 m at the surface to -0.5 m with depth, whereas matric head outside the crater ranged from -665 to -336 m, indicating strikingly dissimilar conditions. A dramatic shift in matric head from -219 to -2.5 m was observed between 8 and 10 m beneath the playa. Bulk densities from the upper 5 m of the fine-grained playa material ranged from 0.9 to 1.33 g cm^-3. Below 5 m, in the coarser loamy sand to sandy loam, bulk densities ranged from 1.21 to 1.74 g cm^-3. Neutron moisture meter measurements (Fig. 2) were high near the surface due to water added during borehole completion. Monitoring revealed no appreciable change in water content with time, and values below the crater (Fig 2a) were considerably higher than observed outside (Fig. 2b). Below the crater there was a significant increase in water content at the 20-m depth, from an average water content of 0.10 above to 0.16 cm³ cm^-3 below this depth.

View larger version (27K): [in this window] [in a new window]   Fig. 2 Water contents measured with a neutron moisture meter at (a) immediately outside the crater and (b) the playa center in the crater

View larger version (29K): [in this window] [in a new window]   Fig. 3 Particle-size analysis profiles at transect locations (a) 30, (b) 70, and (c) 120 m. Locations at 30 and 70 m are within the playa

The crater subsurface outside the playa region was high in gravel and sand down to 2.5 m (Fig. 3). Below the 2.5-m depth, the particle-size distribution mimics the distribution at the crater center. The coarse upper 2.5 m to the north side of the playa suggests that deposition events of considerably lower magnitude occurred subsequent to the initial large deposition event. However, these appeared to be of sufficient energy, near the entrance of the gullies into the crater, to cut into and mix with the upper portion of the previous deposition, resulting in a gradation from coarse to fine surface materials with distance from the gully entrance. Upon entering the crater (and leaving the more confining channels), flow velocity decreased and the heavier, coarser material was preferentially deposited near the gully entrance.

Results of the ring infiltrometer and tension infiltrometer measurements conducted for saturated and unsaturated hydraulic conductivities exhibited the same spatial trend as the particle-size distribution (Table 1). The K_s and K(h) values decreased from the gully entrance to the playa center and then increased at the toeslope position south of the playa. Almost two orders of magnitude difference existed for K_s between the playa region (silty clay to clay textures) and the surrounding region (loamy sands). The average K_s value for the sandy loam and loamy sand soils within the crater was , whereas the fine-grained playa material had a mean .

Confined tension infiltrometer measurements of K(h) at -15 cm head were made immediately following the ring infiltrometer measurements of K_s. For the loamy sand to sandy loam textures, unsaturated hydraulic conductivities, K(-15 cm), decreased by 55% compared with K_s, with a mean of . Within the playa, unsaturated hydraulic conductivity decreased by 40% compared with the associated K_s values, with a mean K(-15 cm) for these fine-grained soils of .

The RETC code (van Genuchten, 1980) was used to model the water retention data by estimation of , m, and n. The residual water content was set to equal the lowest measured value and the Mualem model was used to fit for the and n parameters (Table 1). The sandy loam was found to have a slightly higher average value of 4.4 m^-1, as compared with 3.8 m^-1 for the loamy sand. Shott et al. (1997) and Bechtel Nevada (1998) found values for Area 5 soils of 1.8 and 1.2 m^-1, respectively, which were much lower than both the values provided by HYDRUS-2D, which were derived from Carsel and Parrish (1988), and those measured. The discrepancy may be due in part to the experimental approach. The first matric head applied to the core samples of the Shott et al. (1997) and Bechtel Nevada (1998) studies was -40 cm, while the initial value applied in this study was -15 cm. The van Genuchten n parameter showed good agreement between the two coarser soil textures, with n equal to 1.75 and 1.74 for the loamy sand and sandy loam, respectively, while Bechtel Nevada (1998) reported a value of 1.75. The n parameter in Shott et al. (1997) was 1.9, and HYDRUS-2D has n equaling 1.89 for sandy loam and 2.28 for loamy sand. The and n parameters for the fine-grained playa sediments matched expected results even more closely. The values ranged from 1.8 to 2.3 m^-1, which is close to values of 1.9 to 2.7 m^-1 provided by HYDRUS-2D for these textures. The shape of the water retention curve proved, as expected, to be far more gradual in these fine textures than the sands, with average n values equaling 1.28. This result matches closely the HYDRUS-2D range of 1.23 to 1.31.

Numerical Modeling: Sensitivity Analysis
The effects of varying the top boundary condition on the depth of water movement were examined for a 40 m radius by 40 m deep domain (Fig. 4) with radial symmetry about the left boundary. The no-flow boundaries along the vertical sides, free drainage along the bottom boundary, and media properties used by Hokett and French (1998) were employed. Pond depths tested were 30 and 90 cm, and lateral extents tested were 5, 12, 15, and 20 m. Each pond was allowed to infiltrate until the prescribed volume of water was infiltrated. Following pond infiltration, an ET boundary condition was applied as before until the next ponding event. The first pond volume equaled 271 m³, followed by 990 d of drainage and ET, followed by a second pond volume of 419 m³.

View larger version (20K): [in this window] [in a new window]   Fig. 4 Diagram of the flow domain depicting boundary conditions, soil properties, and initial head, hm. The flow domain is radially symmetric around the left border, which corresponds with the location of the neutron probe borehole. Hydraulic properties for the loamy sand (LS), silt loam (SiL), and loam (L) are reported in Table 3

The difference between wetting front depths for varying lateral extent of ponding was clearly evident after only two ponding events (Plate 1) . The preferential infiltration through the annular coarse-grained material was also evident in the times required to infiltrate the pond volume. For Pond Event 2 with a 0.3-m head, the infiltration times were 0.55, 0.175, and 0.085 d for 12-, 15-, and 20-m lateral extents, respectively. In the case of the 5-m lateral extent, the infiltration simulation, shown in Plate 1, was stopped after 30 d despite the desired volume of water not being fully infiltrated. The confining layers of silt loam and loam limited water movement into the playa, but a small volume of water was able to penetrate below the playa by the end of the second pond event. Given the substantially longer simulation time for the 5-m case, the depth of the wetting front shown in Plate 1 is not a fair comparison of the depth of movement to the other cases.

View larger version (42K): [in this window] [in a new window]   Plate 1 Water content (m3 m-3) profiles at the end of infiltration of Pond Event 2, depicted for the surface to 20 m deep, predicted by HYDRUS-2D for 0.3-m heads extended laterally 5, 12, 15, and 20 m. The fine-grained playa material that extends 10 m laterally and 10 m deep is the yellow material, which is distinguished from the remaining coarse-grained material (no color) due to differences in water retention characteristics

Physical characteristics of Crater U5a support the hypothesis that an extremely large ponding event occurred after crater formation that was not represented in the 13 ponding events of the simulations by Hokett and French (1998). The failure to predict this large event is likely due to their assumption of only the 0.65-km² portion of the 210-km² watershed contributing to runoff into Crater U5a. The three evidences for a large event are:

1. A thin veneer of depositional material is clearly evident at a uniform elevation of 20 m above the current crater floor. Stage-volume relationships of the initial crater dimensions resulted in a volume of 63000 m³ for such a pond.
   
2. Large erosional gullies up to 5 m deep suggest an intense high-energy surface runoff event. Distinct channels were tracked for 8 km back towards the mountain ridges, which suggest extremely large events created a watershed area much greater than the 0.65 km² used in runoff prediction.
   
3. Particle-size analysis of the deposited material provides strong evidence of a large-scale, one-time deposit of sediment as soil texture was gradually coarser with depth, with a clay-textured playa at the surface grading to sands and gravels.
   

To improve the estimation of the recharge potential of Crater U5a, the large event was modeled using field and laboratory measurement of soil properties. One complication with modeling the large ponding event is the inability of HYDRUS-2D (and other numerical models of Richards' equation) to handle a change in the flow domain geometry during the simulation period. Thus, the deposition of 14.6 m of sediment during infiltration of this large event could not be modeled as a continuous infiltration–redistribution process with sedimentation. To adequately capture this process, the final modeling effort was separated into two independent parts: (i) infiltration of the large pond followed by 32 yr of drainage using the original crater surface and (ii) infiltration and drainage of 13 smaller ponds during a period of 32 yr using the current crater features and the water content profile at the time the pond volume had infiltrated.

Numerical Modeling: Large Pond Scenario
At the time of the large pond, the currently observed playa surface and depositional sequence would not have existed. The flow domain for the original crater surface is shown in Fig. 5a . Radial symmetry about the crater center was used with a 110 m wide by 200 m deep flow domain with the top boundary represented by the original crater topography (before deposition). No-flow boundaries were assigned along the vertical sides and free drainage along the bottom boundary. The subsurface consisted of homogeneous media of coarse-sandy loam texture (Table 3) and the initial condition was set in equilibrium with the water table along the bottom boundary. During infiltration of the pond, a pressure head boundary condition was applied along the crater surface to represent a pond with a maximum height of 20 m above the crater floor. The remainder of the top boundary was a no-flow condition. Following infiltration of the 63000 m³ of water, the entire top boundary was changed to an atmospheric condition. Potential evaporation rates for the 32 yr of redistribution following infiltration were taken from the Area 5 PA (Shott et al., 1997). The monthly average potential evaporation was divided by 30 d and applied uniformly throughout the month, which gave a range of 1.5 to 8.0 mm d^-1. Since the impact of the 14.6 m of deposition was not taken into account, this greatly overestimates evaporation, thereby providing a conservative approach. Infiltration of the pond was accomplished incrementally to capture the change in head that would occur as the water level initially rises and then falls. The head was increased and then decreased, with the greatest increase along the crater bottom in which the head changed from 1 to 20 m. After 13.2 d, simulated infiltration of 63000 m³ was completed and the top boundary was changed to an atmospheric condition.

View larger version (76K): [in this window] [in a new window]   Fig. 5 Diagrams showing the domain and mesh for the simulations representing (a) original crater surface to the water table, and (b) present crater surface with sediment soil properties shown in (c) in which numbers refer to hydraulic properties listed in Table 3. Both flow domains are radially symmetric around their left borders, which correspond with the location of the neutron probe borehole

View larger version (95K): [in this window] [in a new window]   Plate 2 Wetting front advancement illustrated by water content distribution (m3 m-3) for selected times following the infiltration of the 63000-m3 ponding event

View larger version (20K): [in this window] [in a new window]   Fig. 6 Water content profiles corresponding with the borehole location (15–30 m depths in Fig. 2) for selected times following the infiltration of the 63000-m3 ponding event

Boundary conditions, soil types, and model domain were quite different from any previously discussed and most closely represented actual conditions observed in the field. A smaller modeling domain (110 m wide and 20 m deep at crater and 34 m deep outside crater) was created to investigate the hydraulic behavior of the depositional material (Fig. 5b). The gradual coarsening of the sediment was mimicked by incorporating seven layers of progressively coarser material (Fig 5c, Table 3). Additionally, the K(h) functions for the fine-grained playa material and the surrounding coarse-grained material were represented by dual porosity models (Table 3).

The thirteen separate ponding events predicted by Hokett and French (1998) occurred 2 yr apart over the 32-yr period simulated. The surface boundary condition used to represent these small ponds was a 0.6-m constant head that, according to the surface topography, extended out 40 m from the crater center. These ponds were placed on the crater bottom, allowed to infiltrate their respective volumes of water, then followed by their associated redistribution period.

Complete infiltration occurred extremely rapidly; even the largest pond (1369 m³) required only 0.03 d. This was a result of the pond extending across the coarse-grained surface material. The playa was observed (Plate 3) to restrict percolation, while the coarse material in the annular region served as a preferred flow region since the pond was extended beyond the playa. When this occurs, the entire pond volume is essentially infiltrated through the peripheral area, causing deeper movement of the wetting front. This was evident by the presence of a wetting front extending to 7.5 m around the periphery of the playa region following the 1369-m³ ponding event (Plate 3c). This form of preferential movement should not be ignored; however, unlike the deep (130 m) movement through the homogenous media used by Hokett and French (1998), this zone of preferential movement exhibited a relatively limited depth of penetration in the layered simulations. Additionally, the depth of penetration of this preferential flow zone stabilized with time and slowly lost water to ET (Plate 3d).

View larger version (49K): [in this window] [in a new window]   Plate 3 Wetting front advancement illustrated by water content distribution (m3 m-3) of the multilayered sediment in the crater at (A) time 0, and after redistribution of (B) Pond 2, (C) Pond 6, and (D) Pond 13

During redistribution periods, much of the moisture loss occurred in the area adjacent to the crater. The subsurface outside the crater sediment exhibited a net loss of 1700 m³ during the 32-yr period. Immediately following ponding events, the upper layers of the crater sediment increased in water content as expected. However, despite the addition of 7673 m³ of water to the crater, the upper 1.5 m of the sediment exhibited a net loss of 27 m³ of water in the 32 yr. Percolation into the deeper sediment contributed to this loss as the remaining portion of the sediment exhibited a net gain of 1040 m³. Simulation results shown in Fig. 3 graphically depict the penetration of the annular flow wetting front into the deeper sediment layers but not out of the sediment. During the 32-yr redistribution period, no water percolated out of the sediment. Thus, the soil layering limited the percolation of water to the upper 15 m. Model results suggest that deposition of fine-grained sediments in this crater can, with time, act to prevent future recharge. This hydraulic barrier was provided by the increased storativity of the fine-grained deposits as the gradual coarsening was probably not conducive to a capillary break effect The size and frequency of the simulated ponds may greatly exceed the actual occurrence, yet the simulated crater with layered sediments performed adequately in prohibiting deep percolation.

	Conclusions

TOP ABSTRACT INTRODUCTION Objective Materials and methods Results and discussion Conclusions REFERENCES

 
Areally distributed recharge through the alluvial fans of the NTS has been considered to be negligible. However, localized recharge through nuclear subsidence craters may be significant. Crater U5a in Area 5 was selected to represent a worst case for surface runoff catchment and therefore a worst case for recharge potential of craters that contain a playa region. Previously reported simulations using default hydraulic properties based on soil texture resulted in recharge rates of 1.18 m yr^-1 when the pond was extended just 2 m beyond the playa material. Variations in the boundary condition representing the pond revealed that extending the pond to 2 m outside the playa provided the highest estimates of recharge potential. Ponds restricted to the playa failed to fully infiltrate, while ponds extended further than 2 m outside the playa showed decreasing depth of penetration. However, boundary condition modifications were not able to reproduce the measured water contents at the playa center.

Previous modeling of 13 small ponding events failed to consider three pieces of physical evidence that support the hypothesis that an extremely large ponding event occurred after crater formation. These evidences were surficial coating indicating a high-water mark, extremely large deeply incised gullies indicating a high-energy runoff event, and continuous gradual coarsening of sediment with depth without the alternating-layers stratigraphy that would occur with intermittent deposition events. The hypothesis of a single, large ponding event was corroborated by the model simulations as water contents simulated with a large pond matched closely those determined in the field. Simulations suggest that the large pond readily infiltrated through the coarse-grained media of the original crater surface with the wetting front reaching the deep water table in 32 yr. The wetting-front advanced at a rate of 30.0 m yr^-1 during the first few years, but the recharge rate was estimated to be 2.5 m yr^-1 at the time the water table was reached.

Future recharge is controlled by the hydraulic properties of the nearly 15 m of sediment that consist of a gradually coarsening sequence of clay to silt to sandy loam material with depth. Model simulations revealed that recharge from subsequent ponding events would be effectively prohibited. The hypothesis of preferential infiltration through the coarser-grained material surrounding the playa was supported by the model simulations and by the observed plant community distribution. However, based on the predicted infrequent small ponding events, the deposited sediment effectively stored the infiltrated water until it was removed by ET, thereby creating a natural crater closure.

It is clear that nuclear subsidence craters can serve as focused recharge basins even under the extreme arid conditions of the NTS. However, unlike previous studies, this work demonstrated that these craters may be self-healing. Sediments deposited in the craters and the establishment of deep-rooting vegetation may form natural hydraulic barriers. This work also demonstrated the importance of site characterization for proper performance assessment.Smettem Kirkby 1990

	ACKNOWLEDGMENTS

 
Financial support for this study was provided by the Nevada Risk Assessment/Management Program (NRAMP) under U.S. DOE Grant DEFG 01 96 EW 56093 and the DOE/NV Environmental Restoration Division under contract DE-AC08-95NV11508. The support of Bob Bangerter of the DOE/NV Underground Test Area program and Don Baepler of the Harry Reid Center for Environmental Studies was much appreciated.

Received for publication July 12, 1999.

	REFERENCES

TOP ABSTRACT INTRODUCTION Objective Materials and methods Results and discussion Conclusions REFERENCES

 

• Anderson T.W. Summary of the Southwest Alluvial Basins, regional aquifer-system analysis, south-central Arizona and parts of adjacent states. U.S. Geological Survey Professional Paper 1406-A. Reston, VA: U.S. Geological Survey, 1995.
• Bechtel Nevada. 1998. Hydrogeologic characterization of the unsaturated zone at the Area 3 Radioactive Waste Management Site. Data Report, U.S. Dep. of Energy, Nevada Operations Office, DOE/NV/11432-196, Las Vegas, NV.
• Carsel R.F., Parrish R.S. Developing joint probability distributions of soil water retention characteristics. Water Resour. Res. 1988;24:755-769.
• Crowe, B., W. Hansen, and M.J. Sully. 1998. The consequences of disposal of low-level radioactive waste from the Fernald Environmental Management Project. Rep. of the DOE/Nevada Independent Panel, Los Alamos National Laboratory. LA-13453-MS, UC-840.
• Domenico P.A., Stephenson D.A., Maxey G.B. Groundwater in Las Vegas Valley. Technical Rep. 7. Las Vegas, NV: Desert Research Inst, 1964.
• Fischer, J.M. 1992. Sediment properties and water movement through shallow unsaturated alluvium at an arid site for disposal of low-level radioactive waste near Beatty, Nye County, Nevada. U.S. Geological Survey, Water Resources Investigations Rep. 92-4032. U.S. Geological Survey, Reston, VA.
• Gee G.W., Bauder J.W. Particle-size analysis. In: Klute A., ed. Methods of soil analysis. Part 1, 2nd ed Madison, WI: ASA and SSSA, 1986:383-411 Agron. Monogr. 9..
• Hokett S.L., Gillespie D.R. Preliminary evaluation of recharge potential at subsidence crater U5a in Frenchman Flat. Las Vegas, NV: Desert Research Inst, 1996 Nevada Test Site. DOE/NV 11508-15, Publ. no. 45147. Water Resources Center.
• Hokett S.L., French R.H. Evaluation of recharge potential at crater U5a (Wishbone). DOE/NV/11508-02, Publ. no. 45160. Las Vegas, NV: Water Resources Center, Desert Research Institute, 1998.
• Klute A.J. Water retention: Laboratory methods. In: Klute A., ed. Methods of soil analysis. Part 1, 2nd ed Madison, WI: ASA and SSSA, 1986:635-662 Agron. Monogr. 9..
• Magnuson S.O., Maheras S.J., Nguyen H.D., Rood A.S., Sipos J.I., Case M.J., McKenzie-Carter M.A., Donahue M.E. Radiological performance assessment for the Area 5 Radioactive Waste Management Site at the Nevada Test Site. Idaho Falls, ID: INEL, 1992.
• Mohanty B.P., Bowman R.S., Hendrickx J.H., van Genuchten M.Th. New piecewise-continuous hydraulic functions for modeling preferential flow in an intermittent-flood-irrigated field. Water Resour. Res. 1997;33:2049-2063.
• Pohll G.M., Warwick J.J., Tyler S.W. Coupled surface–subsurface hydrologic model of a nuclear subsidence crater at the Nevada test site. J. Hydrol. (Amsterdam) 1996;186:43-62.
• Scanlon B.R., Goldsmith R.S. Field study of spatial variability in unsaturated flow beneath and adjacent to playas. Water Resour. Res. 1997;33:2239-2252.
• Shott, G.J., L.E. Barker, S.E. Rawlinson, and M.J. Sully. 1997. Performance assessment for the Area 5 RWMS at the Nevada Test Site, Nye County, Nevada. Revision 2.1. U.S. Dep. of Energy, Nevada Operations Office, DOE/NV/11432–196, Las Vegas, NV.
• Simunek J., Sejna M., van Genuchten M.Th. Simulating water flow and solute transport in two-dimensional variably saturated media. International Ground Water Modeling Center, IGWMC-TPS 53. Golden, CO: Colorado School of Mines, 1996.
• Simunek J., Vogel T., van Genuchten M.Th. The SWMS_2D code for simulating water flow and solute transport in two-dimensional variably saturated media. Version 1.1, Research rep. no. 126. Riverside, CA: U.S. Salinity Lab., USDA-ARS, 1992.
• Smettem K.R.J., Chittleborough D.J., Richards B.G., Leaney F.W. The influence of macropores on runoff generation from a hillslope soil with a contrasting textural class. J. Hydrol. (Amsterdam) 1991;122:235-252.
• Smettem K.R.J., Kirkby C. Measuring the hydraulic properties of a stable aggregated soil. J. Hydrol. (Amsterdam) 1990;117:1-13.
• Soil Science Society of America. Glossary of soil science terms. Madison, WI: SSSA, 1997.
• Teller E., Talley W., Higgins C., Johnson C. The constructive uses of nuclear explosives. New York: Mcgraw-Hill, 1968.
• Tyler S.W., Chapman J.B., Conrad S.H., Hammermeister D.P., Blout D.O., Miller J.J., Sully M.J., Ginanni J.M. Soil-water flux in the southern Great Basin, United States: Temporal and spatial variations over the last 120,000 years. Water Resour. Res. 1996;32:1481-1499.
• Tyler S.W., McKay W.A., Mihevc T.M. Assessment of soil moisture movement in nuclear subsidence craters. J. Hydrol. (Amsterdam) 1992;139:159-181.
• van Genuchten M.Th. A closed-form solution for predicting the hydraulic conductivity of unsaturated soils. Soil Sci. Soc. Am. J. 1980;44:892-898.[Abstract/Free Full Text]
• Wilson G.V., Jardine P.M., Gwo J.P. Modeling the hydraulic properties of a multiregion soil. Soil Sci. Soc. Am. J. 1992;56:1731-1737.[Abstract/Free Full Text]
• Wilson G.V., Luxmoore R.J. Infiltration, macroporosity, and mesoporosity distributions on two forested watersheds. Soil Sci. Soc. Am. J. 1988;52:329-335.[Abstract/Free Full Text]
• Winograd I.J., Doty G.C. Paleohydrology of the southern Great Basin with special reference to water table fluctuations beneath the Nevada Test Site during the late (?) Pleistocene. U.S. Geological Survey Open-File Rep. 80-569. Reston, VA: U.S. Geological Survey, 1980.

This Article Abstract Figures Only Full Text (PDF) Free Alert me when this article is cited Alert me if a correction is posted Services Similar articles in this journal Similar articles in ISI Web of Science Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via ISI Web of Science (1) Citing Articles via Google Scholar Google Scholar Articles by Wilson, G.V. Articles by Gillespie, D.R. Search for Related Content PubMed Articles by Wilson, G.V. Articles by Gillespie, D.R. GeoRef GeoRef Citation Agricola Articles by Wilson, G.V. Articles by Gillespie, D.R.