Published August 15, 2014
Original Research
Field-Scale Sulfur Hexafluoride
Tracer Experiment to Understand
Long Distance Gas Transport
in the Deep Unsaturated Zone
Michelle A. Walvoord,* Brian J. Andraski, Christopher T.
Green, David A. Stonestrom, and Robert G. Striegl
A gas-tracer test in a deep arid
unsaturated zone demonstrates
that standard estimates of effective diffusivit y from sediment
properties allow a reasonable
first-cut assessment of gas contaminant transpor t. Apparent
anomalies in historic transport
behavior at this and other waste
disposal sites may result from factors other than nonreactive gas
transport properties.
A natural gradient SF6 tracer experiment provided an unprecedented evaluation of long distance gas transport in the deep unsaturated zone (UZ)
under controlled (known) conditions. The field-scale gas tracer test in the
110-m-thick UZ was conducted at the U.S. Geological Survey’s Amargosa
Desert Research Site (ADRS) in southwestern Nevada. A history of anomalous (theoretically unexpected) contaminant gas transport observed at
the ADRS, next to the first commercial low-level radioactive waste disposal
facility in the United States, provided motivation for the SF6 tracer study.
Tracer was injected into a deep UZ borehole at depths of 15 and 48 m, and
plume migration was observed in a monitoring borehole 9 m away at various depths (0.5–109 m) over the course of 1 yr. Tracer results yielded useful
information about gas transport as applicable to the spatial scales of interest for off-site contaminant transport in arid unsaturated zones. Modeling
gas diffusion with standard empirical expressions reasonably explained
SF6 plume migration, but tended to underpredict peak concentrations for
the field-scale experiment given previously determined porosity information. Despite some discrepancies between observations and model results,
rapid SF6 gas transport commensurate with previous contaminant migration
was not observed. The results provide ancillary support for the concept that
apparent anomalies in historic transport behavior at the ADRS are the result
of factors other than nonreactive gas transport properties or processes currently in effect in the undisturbed UZ.
Abbreviations: ADRS, Amargosa Desert Research Site; FEHM, Finite Element Heat and
Mass Transfer code; LLRW, low-level radioactive waste; MQI, Millington–Quirk I; MQII,
Millington­– Quirk II; UZ, unsaturated zone.
M.A. Walvoord, USGS, DFC Box 25046,
MS-413, Lakewood, CO 80225; B.J.
Andraski, USGS, 2730 N. Deer Run
Road, Carson City, NV 89701; C.T.
Green and D.A. Stonestrom, USGS,
345 Middlefield Road, Menlo Park, CA
94025; R.G. Striegl, USGS, 3215 Marine
St., Boulder, CO 80303. *Corresponding author ([email protected]).
Vadose Zone J.
doi:10.2136/vzj2014.04.0045
Received 24 Apr. 2014.
Open access
© Soil Science Society of America
5585 Guilford Rd., Madison, WI 53711 USA.
All rights reserved. No part of this periodical may
be reproduced or transmitted in any form or by
any means, electronic or mechanical, including
photocopying, recording, or any information storage and retrieval system, without permission in
writing from the publisher.
Gas diffusion is often considered to be the main mechanism of gas-species transport in
natural-gradient arid unsaturated zones (Thibodeaux, 1981; Weeks et al., 1982; CostanzaRobinson and Brusseau, 2006). Characterization of gas diffusion depends on the diffusive
and reactive properties of gas-phase constituents, which are typically well established, and
the physical characteristics of the soil (Jin and Jury, 1996). The porous medium characteristics together with the free-air diffusion coefficient define the effective diffusion coefficient
(Deff ), which is utilized to mathematically describe spreading along a concentration gradient over time (Webb, 2006). Sound estimates of Deff are important for many waste
site management applications that utilize predictive models, including risk assessments
of volatilization and migration of chemical waste and remediation performance estimates.
In the absence of direct measurements, values of Deff are approximated using empirical
relationships developed from theoretical pore-size distribution models (Millington, 1959)
or more commonly derived from one-dimensional, laboratory-scale soil column experiments (Millington and Quirk, 1960; Moldrup et al., 2000). Little work has been done to
evaluate the suitability of empirical relationships of Deff for field applications involving
long distance transport through deep UZs. An exception is the use of trace atmospheric
gases to explore the deep UZ (Weeks et al., 1982; Weeks and McMahon, 2007). The atmospheric tracer method is limited to assessing diffusion in the vertical direction, typically
Vadose Zone Journal
perpendicular to bedding and sediment stratigraphy. The focus of
this study, in contrast, is on lateral transport of gases introduced
into the subsurface. As pointed out in a review by Werner et al.
(2004), additional experimentation is needed for diffusion theory
model validation at the field scale, particularly in heterogeneous
and structured porous media.
The ADRS in southwestern Nevada (Fig. 1a) serves as a field-scale
laboratory for investigating long-distance gas transport in arid
unsaturated soils and sediments. This site has been well characterized over many years of study (e.g., Fischer, 1992; Andraski,
1997; Andraski and Stonestrom, 1999). A low-level radioactive
waste (LLRW) disposal facility adjacent to the ADRS serves as
the source of numerous gas-phase contaminants. From 1962 to
1992, the facility buried commercial waste in a series of 22 unlined
trenches ranging from 2 to 15 m deep. Contaminants from the
buried waste have migrated through the thick UZ and have been
consistently measured in ADRS boreholes since measurements
first began about 20 yr ago (Stonestrom et al., 2004; Baker et al.,
2012; Maples et al., 2013). The ADRS deep unsaturated boreholes were drilled and instrumented to various sampling depths
(UZB-1, 47.9 m; UZB-2, 108.8 m; UZB-3, 103.9 m) in November
1992, September 1993, and December 1999, respectively (Fig. 1b).
Tritium concentrations in water vapor and core water collected
from the UZ at the ADRS (Prudic and Striegl, 1995; Prudic et
al., 1997) greatly exceed concentrations predicted by accepted
analytical tritium transport models (i.e., Smiles et al., 1995)
applied across the estimated ranges of geological, hydrological, and
chemical conditions at the site (Striegl et al., 1996). Simulations
conducted using multidimensional, multiphase flow and transport
numerical models (Walvoord and Stonestrom, 2004; Mayers et al.,
2005) also underestimate the measured tritium migration from
the LLRW facility. Several other gas-phase constituents derived
from buried waste, including chlorofluorocarbons, hydrocarbons,
and radioactive carbon dioxide, exhibit enhanced long distance
(>100 m) lateral transport at the ADRS (Striegl et al., 1996; Baker
et al., 2012). Vertical profiles of elemental mercury gas display
abrupt concentration changes with depth, suggesting preferential
lateral migration that is not captured by conventional modeling approaches based on simple transport by gas diffusion alone
(Walvoord et al., 2008).
Here we describe the results and analyses of a gas tracer test in
the deep UZ under natural-gradient conditions at the ADRS. The
temporal and spatial (both depth and longitudinal) scales investigated by this field study exceed those addressed by previous gas
transport experiments (Werner et al., 2004; Tick et al., 2007). The
primary objectives of this experimental study were to: (i) evaluate
the general utility of laboratory-derived empirical formulations
of diffusivity for field-scale application and (ii) improve the physical framework for interpreting both apparent anomalous historic
contaminant transport data as well as recent data collected via
regular monitoring. “Expected” tracer results (i.e., those displaying
Vadose Zone Journal
gas-phase Fickian diffusion well matched to model predictions)
would suggest that laboratory-derived empirical formulations of
diffusivity may be appropriately applied to field scales larger than
previously tested. Expected results would further suggest that historic apparent anomalies cannot be explained by current ambient
conditions, pointing to site history and contaminant-specific partitioning to explain transport evolution. “Unexpected” tracer results
would shed light on the limitations of standard diffusivity models
as applied to long-distance transport in the deep UZ. Unexpected
results could also help substantiate observed enhanced gas-phase
transport at the ADRS and provide insight into pervasive processes controlling anomalous transport at this and other LLRW
sites (Dirkes et al., 1999).
Fig. 1. (a) Location of the Amargosa Desert Research Site (ADRS)
and the low-level radioactive waste (LLRW) facility in southwestern
Nevada, as well as (b) the ADRS deep unsaturated zone boreholes
adjacent to the LLRW facility.
p. 2 of 10
In situ methods for evaluating gas-diffusion transport properties include, but are not limited to, single-well and cross-well
tracer tests with instantaneous and continuous point sources as
reviewed by Werner et al. (2004). Single-well gas tracer injection
tests (Werner and Höhener, 2003; Tick et al., 2007) are rapid,
require few soil gas measurements, and yield transport property
information for a small, usually poorly defined, volume of soil near
the well. Cross-well gas tracer tests (Kreamer et al., 1988; Nicot
and Bennett, 1998) offer a distinct advantage over single-well tests
in that a larger volume of subsurface material is investigated, but
they require more gas concentration measurements and a longer
monitoring duration. An instantaneous point-source test has operational advantages over a continuous point-source test in that a
slow, constant release rate of the tracer gas is difficult to maintain
over long periods of time.
For the purposes of better understanding field-scale gas transport
in the deep UZ, a cross-well instantaneous point-source approach
was selected, taking advantage of existing monitoring infrastructure at the research site. Sulfur hexafluoride (Table 1) was chosen
as the tracer gas due to its conservative physiochemical properties
(Rattray et al., 1995; Maiss et al., 1996). Sulfur hexafluoride is primarily anthropogenic and is found in relatively low concentrations
in the atmosphere (<10 pptv) and in the UZ under natural conditions. Concentrations can be detected at atmospheric levels with
high precision using a gas chromatograph equipped with an electron capture detector and following rigorous procedures to prevent
contamination and instrumental noise (Busenberg and Plummer,
2000). For measuring concentrations well above background levels,
procedures need not be as rigorous, and a gas chromatograph with
an electron capture detector that is relatively easy to transport can
be used for on-site analysis.
66Materials
Study Site
and Methods
The present climate in the Amargosa Desert is arid-thermic, with
precipitation averaging 108 mm yr−1 from 1981 to 2011 (Arthur et
al., 2012). The dominant vegetation is creosote bush [Larrea tridentata (Sessé & Moc. ex DC.) Coville], a highly water-efficient North
American desert shrub. Continuous monitoring at the ADRS since
1987 under native vegetation indicates that evapotranspirative processes confine infiltrating precipitation-induced wetting fronts to
within 1.2 m of the land surface (Andraski, 1997; Tumbusch and
Prudic, 2000). Native plants at the ADRS maintain extremely low
matric potentials at the base of their ?1-m-deep root zone (£−4
MPa), preventing deeper percolation. This shallow monitoring data
together with modeling of measured chloride, matric potential, and
stable isotope (2H and 18O in water) profiles of the deep UZ strongly
suggests that liquid fluxes below the root zone and above the water
table are negligible (Walvoord et al., 2004).
Vadose Zone Journal
Table 1. Properties of sulfur hexafluoride (SF6).
SF6 property
Molecular weight, g mol−1†
Density, g L−1†
Value
146.06
6.164
Free-air diffusion coefficient, m2 d−1‡
0.77
Vapor pressure at 22°C, MPa‡
2.26
Air–water partitioning constant (mol m−3 air/mol m−3 water)
at 10°C‡
105
Air–water partitioning constant (mol m−3 air/mol m−3 water)
at 15°C‡
126
† Tick et al. (2007).
‡ Werner and Höhener (2003).
Depth to groundwater at the ADRS ranges from 85 to 115 m
(Fischer, 1992). Sediments in the UZ consist of unconsolidated
and poorly sorted debris-flow, fluvial, and alluvial-fan deposits
(Nichols, 1987). Deep sediments include playa and lacustrine
deposits (Taylor, 2010). Beneath the surface soil, sediment composition is almost entirely sand and gravel, with minor amounts
of silt and clay. However, layers with more abundant fine-grained
sediment were observed at or near the surface and between depths
of 79 and 85 m. Borehole cores analyzed for texture were all classified as gravelly to very gravelly sand to sandy loam. A generalized
layered stratigraphy for the site was conceptualized (Fig. 2a) and
implemented for previous modeling studies (Mayers et al., 2005;
Walvoord et al., 2008). Volumetric water contents are relatively low,
ranging from 0.05 to 0.14 m3 m−3, and estimated porosities range
from 0.18 to 0.32 m3 m−3 (Mayers et al., 2005). Very low unsaturated hydraulic conductivity and thus negligible water flux (on the
order of 10−2 mm yr−1) in the deep UZ render gas-phase transport
an important mechanism for the movement of water vapor and
volatile constituents. In this study, we test the hypothesis that diffusion is the main mechanism of gas-phase transport at this site.
Experiment Design
The gas tracer study at the ADRS made use of existing instrumented boreholes located 9.00 m apart (Fig. 2b,c). UZB1, designed
for water potential monitoring, contained tubes with removable thermocouple psychrometers similar to those described by
Tumbusch and Prudic (2000). Borehole completion of UZB-1
enabled ground surface access to four discrete depths (Prudic et
al., 1997). UZB-1 air-sampling ports have specially constructed
stainless-steel screens that connect directly to 6-mm nylon tubing
reaching the ground surface and thus enabling access. Tracer was
injected at two of these port locations, 14.9 and 47.9 m below the
ground surface, via the psychrometer access tubes. The UZB-2
borehole was used to monitor concentration breakthrough curves
at available gas ports. UZB2, designed for gas sampling, was
described by Prudic et al. (1997). For the shallow injection port
in UZB-1, the closest horizontal monitoring ports in UZB-2 were
offset by ?3 m above (11.9 m) and below (18.0 m). For the deeper
p. 3 of 10
injection port in UZB-1, there was a UZB-2 monitoring port at
the same depth (47.9 m).
A 99.99% pure SF6 tracer was used to minimize the amount of
pressure disturbance to the system while introducing the desired
amount of tracer. Tracer was injected simultaneously at each of
the two UZB-1 ports on 27 Apr. 2005 at an approximate rate of
15.38 mL min−1 over 78 min to achieve total injection volumes of
1200 mL. Manual injection was accomplished using polypropylene
60 mL-syringes with three-way stopcocks connected to the 6-mm
access tubes associated with the two UZB-1 ports at depths of
14.9 and 47.9 m. Twenty syringes were needed for each injection
location to deliver the desired tracer amount. Immediately following tracer introduction, 120 and 240 mL of air were injected
at the shallow and deep injection ports, respectively, to clear the
psychrometer-access tubes of residual SF6 tracer and introduce all
tracer into the sediment.
run before every round of sample analysis to establish calibration
curves. The standards were chosen to span the expected range of
concentration values for the field samples and yielded a detection
limit of 1 nL L−1. Periodic blanks and duplicates were run for quality assurance.
Numerical Modeling Approach
To compare the SF6 tracer observations with expected results for
Fickian gas diffusion, a two-dimensional radial forward modeling
approach was implemented using the UZ framework developed
in previous ADRS studies and standard empirical formulations
for diffusivity calculations. The model design followed the methodology described in Walvoord et al. (2008) to simulate radial
transport of a conservative gas at the ADRS. The Finite Element
Heat and Mass Transfer Code (FEHM; Zyvoloski et al., 1997)
was used to simulate non-isothermal vapor and liquid movement
and SF6 transport in the 110-m thick UZ for 1 yr following tracer
injection at UZB-1. The model domain extended 110 m vertically
(the entire UZ) and 200 m in the horizontal (radial) direction
away from the injection borehole; variable grid spacing of 0.25 to
2 m was used, resulting in a 11,640 node (97 ´ 120), 11,424 rectangular elemental grid. Layered heterogeneity in soil properties
observed at the ADRS, as illustrated in Fig. 2a and listed in Table
2, was implemented in the model following Mayers et al. (2005)
and Walvoord et al. (2008). Initial conditions for the transport
simulations represent 16,000 yr of transient UZ drying based on
water potential and pore water concentrations of Cl−, 2H, and 18O
Before tracer injection, gas samples were collected from the
UZB-1 and UZB-2 ports to establish background concentrations.
Beginning 1 d after tracer injection, gas samples from the UZB-2
monitoring borehole were collected twice daily for the first 2 wk of
the experiment, after which the sampling frequency was gradually
reduced during the remainder of the year. Each 6-mm monitoring tube was purged by removing approximately two pore tube
volumes (individually calculated for the depth of each port) before
sample collection. Samples were collected in 20-mL nylon syringes
equipped with a three-way polycarbonate stopcock.
Sample syringes were flushed with gas from the collection depth four times before the actual sample was
retained. Positive pressure was applied to the closed-off
syringe containing the gas sample to minimize dilution
before gas chromatograph analysis. Duplicate samples
were collected. Gas samples were analyzed by on-site
gas chromatography for the first 2 wk of the experiment. Subsequent samples were shipped to the USGS
in Denver, CO and analyzed within 1 to 2 d of collection. All samples were analyzed using a PerkinElmer
gas chromatograph equipped with a dual electron
capture detector, which was transported to the site
before tracer injection and returned to Denver 2 wk
post injection. Ultra-pure N2 gas was used as the carrier gas. Gas passed through moisture and oxygen traps
before entering the chromatograph. Peak separation
of SF6 was accomplished by injecting 0.5 mL of the
collected sample or standards with the N2 carrier gas
at a rate of 30 mL min−1 through a 80/100 Porapak
Q 3.175-mm (1/8-inch) column. Settings included
an oven temperature of 35°C, injector temperature
of 100°C, and electron capture detector temperaFig. 2. (a) Generalized layered configuration of unconsolidated material in the
ture of 100°C, following those used by Rattray et al.
unsaturated zone at the Amargosa Desert Research Site and the vertical distribution
(1995). Standards of National Institute of Standards
of (b) injection points in UZB-1, (c) monitoring points in UZB-2, and (d) modeled
water content, air-filled porosity, and total porosity.
and Technology (NIST)-traceable SF6 in air were
Vadose Zone Journal
p. 4 of 10
measured at the ADRS (Walvoord et al., 2004). The resulting water
content and air-filled porosity profiles, together with total porosity,
are shown in Fig. 2d. Initial SF6 concentrations for the simulation
were set at 0 nL L−1 before tracer injection. Temperatures varied
linearly between 21.8°C at the surface to 26.8°C at the water table.
Two Millington–Quirk formulations were used to evaluate their
effect on simulation results. Effective diffusion coefficients (Deff )
for SF6 were calculated by the model for each model node at each
time step based on the Millington–Quirk I and II formulations
(Millington 1959; Millington and Quirk, 1960; Jin and Jury, 1996),
Millington–Quirk I (MQI): Deff = D0a (10/3)/j2
Millington–Quirk II (MQII): Deff
= D0 a2/j(2/3)[2]
and Discussion
SF6 Tracer Data
Background concentrations of SF6 in monitoring borehole UZB-2
were determined using samples collected before tracer injection in
UZB-1 (time = 0 d in Table 3), and values ranged between 3 and
18 nL L−1. These background values are about 600 to 3000 times
the 2005 atmospheric SF6 concentration, as a result of migration
from SF6 –containing waste in the adjacent LLRW area. Such
large background concentrations preclude applying the method
of Weeks et al. (1982) to discern soil diffusion parameters and
warrant the use of pure SF6 for tracer injection to easily distinguish
the injected tracer from ambient site concentrations. Measured SF6
concentration profiles in UZB-2 through time following simultaneous tracer injections at shallow (14.9 m) and deep (47.9 m) ports
in UZB-1 illustrate the migration of the shallow and deep plumes
(Fig. 3; Table 3). UZB-2 concentrations measured over the first
week following tracer injection showed no systematic increases.
Mean concentrations for Days 1 through 7 are listed in Table 3.
At Day 8, the SF6 concentration increased by about an order of
magnitude above background concentration at the UZB-2 11.9-m
depth, which is vertically offset from the UZB-1 shallow injection
port by 3.0 m. A similar increase at the 18.0-m monitoring depth,
which is vertically offset from the shallow injection port by 3.1 m,
was not detected until Day 13. The earlier arrival of the shallow
Vadose Zone Journal
Property
Silty sand
with gravel
Gravel
Sandy gravel
Porosity, m3 m−3
0.32
0.18
0.24
Permeability, m2
5.1 e-13
1.23 e-11
3.8 e-13
Residual saturation, m3 m−3
0.001
0.001
0.001
Maximum saturation, m3 m−3
0.91
0.91
0.91
van Genuchten a, m−1
3.6
714.8
43.6
van Genuchten n
1.29
1.18
1.15
[1]
where D 0 is the vapor diffusion coefficient for SF6 in air:
0.77 m2 d−1 (Werner and Höhener, 2003); a is air-filled porosity;
and j is total porosity. Sulfur hexafluoride was modeled as a conservative gas species. Sulfur hexafluoride was injected as a rate of
15.38 mL min−1 over a 78-min injection period at depths of 14.9
and 47.9 m, commensurate with the field experiment. Simulations
were run on a 0.1-h time step for 1 yr after the injection period to
correspond with the duration of the tracer test monitoring period.
66Results
Table 2. Soil properties used in model simulations. Values are from
Mayers et al. (2005), based on data in Andraski (1997) and Andraski
and Jacobson (2000).
plume at the 11.9-m versus 18.0-m monitoring depth suggests that
SF6 tracer movement was more strongly affected by sediment stratigraphy than density effects by the time the plume reached UZB-2.
As expected, concentration changes in UZB-2 associated with the
UZB-1 shallow (14.9 m) injection were most pronounced at the
11.9- and 18.0-m sampling depths throughout the monitoring
period. Similarly, concentration changes in UZB-2 associated with
the deep (47.9 m) injection were most prominent in the sample
depths of 47.9 m and 57.6 m throughout the monitoring period.
Breakthrough curves for UZB-2 monitoring depths are illustrated
in Fig. 4a. The maximum shallow-plume SF6 concentration measured at 11.9 m was 927 nL L−1 at 42 d, and the maximum at 18.0
m was 761 nL L−1at 70 d. The sampling frequency during this
period imposes about a 1-wk margin of error on the peak concentration arrival estimate. Deep plume concentrations began
increasing at the UZB-2 47.9-m depth ?12 d after the UZB-1
tracer injection. The peak deep-plume SF6 concentration at 47.9
m reached 1312 nL L−1 at 70 d and that for the 57.6-m depth was
249 nL L−1 at 89 d.
Gas samples from the UZB-1 tracer injection ports at 14.9 and
47.9 m were also collected and analyzed beginning 56 d after injection. Samples from UZB-1 were not withdrawn before that time to
minimize near-field disturbance of the tracer test. During the post
injection period from Days 56 and 357, a total of seven gas samples
at each injection port were collected and analyzed for SF6 concentration (Fig. 4b, Table 4). Although equivalent SF6 volumes were
injected simultaneously at the two injection ports at the beginning
of the tracer test, the Day 56 concentration for the deep port (54,373
nL L−1) was nearly three times that for the shallow port (18,542
nL L−1). Greater concentrations for the deeper port (factors of 2.3
to 4.1) persisted throughout the UZB-1 sample collection period.
By the end of the tracer test, SF6 concentrations had decreased to
174 and 471 nL L−1 at 14.9 and 47.9 m, respectively.
Comparisons of the UZB-2 breakthrough curves (Fig. 4a) and
the UZB-1 declines in SF6 at the injection ports (Fig. 4b) support
the interpretation that spreading of the shallow plume is faster
p. 5 of 10
Table 3. Sulfur hexafluoride concentrations in the unsaturated zone monitoring borehole, UZB-2, at various depths and through time subsequent to
tracer injection at the shallow (14.9 m) and deep (47.9 m) ports in UZB-1.
Sampling depth in UZB-2†
Time since injection
5.5 m
d
———————————————————————————————SF6 , nL L−1 ———————————————————————————————
11.9 m
18.0 m
24.1 m
34.1 m
47.9 m
57.6 m
94.2 m
0 (background) ‡
3.4
5.7
7.0
5.4
8.6
11.1
17.9
13.7
1–7 (mean)
2.8
7.5
11.9
10.8
11.0
17.4
20.9
20.2
8
3.2
58.0
11.6
4.4
7.7
14.5
16.7
18.4
9
7.1
101.0
11.6
5.4
10.7
14.7
11.2
15.1
10
<1.0
125.4
17.3
19.0
42.9
26.9
113.1
33.7
12
4.7
254.8
23.6
5.4
8.2
60.1
36.7
6.1
13
0.4
280.6
63.4
7.4
14.7
78.5
22.1
32.7
14
20.2
223.8
78.3
24.3
31.0
86.6
44.5
50.7
16
20.4
300.5
145.8
35.3
28.2
90.0
45.5
32.4
18
8.7
381.4
160.6
13.0
14.1
112.1
32.5
27.4
20
§
604.4
165.9
6.0
16.1
253.8
11.4
§
21
5.9
571.5
192.3
4.8
14.6
302.9
11.6
14.9
26
14.9
678.5
337.5
5.3
15.3
433.7
21.8
8.9
29
8.5
815.8
345.4
7.7
4.3
489.2
41.1
6.2
35
29.4
857.4
467.6
16.6
49.4
817.9
45.2
6.9
42
40.9
926.7
468.6
42.4
64.0
996.8
82.9
22.1
56
100.5
893.7
635.2
59.7
52.6
1177.4
199.0
28.6
70
170.0
709.0
761.1
128.0
116.6
1312.1
239.1
65.4
89
196.2
498.8
722.9
124.0
75.9
1125.0
248.9
20.3
132
188.0
388.6
437.9
184.4
128.3
828.6
170.1
52.4
183
154.7
281.5
235.1
137.5
216.5
543.8
130.5
32.1
258
107.0
149.3
172.5
123.8
132.2
294.9
115.5
35.3
357
88.5
144.4
156.5
119.2
188.0
187.4
99.5
30.2
† Samples were also collected at depths of 106.4, and 108.8 m, but are not shown here, since tracer test influence was not detected in concentrations from these samples.
‡ Samples collected before tracer injection.
§ Sample not collected.
Fig. 3. Sulfur hexafluoride
concentrations
through
time in gas samples collected
in the UZB-2 monitoring
borehole. Time (t) represents
time since tracer injection at
shallow (14.9 m) and deep
(47.9 m) ports in nearby
UZB-1 borehole. Arrows
indicate time advancing
trend in shallow and deep
plume concentration.
Vadose Zone Journal
p. 6 of 10
Table 4. Time-dependent sulfur hexafluoride concentrations in the
unsaturated zone collected at shallow (14.9 m) and deep (47.9 m) injection ports in borehole UZB-1 on selected days after tracer injection.
Sampling depth in UZB-1
Time since injection
14.9 m
47.9 m
d
——————————SF6 , nL L−1 ——————————
56
18,542
89
2,102
7,197
98
1,419
5,422
132
795
3,288
183
369
1,147
258
241
562
357
174
471
54,373
oscillations induced by atmospheric pressure variations (Kuang
et al., 2013).
Modeling Results
Fig. 4. (a) Breakthrough curves for SF6 in UZB-2 at unsaturated zone
monitoring depths and (b) temporal declines in SF6 at injection ports
in UZB-1.
than the deep plume. The shallow plume arrives at the UZB-2
11.9-m depth before the deeper plume arrives at the 47.9-m depth,
despite the vertical offset in the shallow plume injection monitoring depths. The concentration peak of the shallow plume also
occurs earlier in time (42 d) than the peak of the deep plume (70
d). The magnitudes of the UZB-2 concentration peak and the
UZB-1 concentration decline are both greater for the deep plume
than those for the shallow plume. The more rapid arrival and lower
ultimate peak concentrations in the shallow plume may result from
variations in porosity and tortuosity in the UZ sediments, but may
also indicate greater dispersion from effects of minor advective
The radial transport results for the SF6 tracer simulations for four
selected times since tracer injection at UZB-1 (the origin) are
shown in Fig. 5. Results shown here correspond to the simulation
that utilized the MQI formulation to calculate Deff at each node;
however, results from the MQII are visually identical to MQI
results when displayed at the same scale. Despite the modeled layered heterogeneity in the system, the diffusive plumes display little
difference from what would be expected in a homogeneous system.
This is due to the relatively large magnitude of diffusion coefficients
for gases like SF6 (0.77 m2 d−1, in air). For comparison, the diffusion coefficient of SF6 in pure water at 25°C is on the order of 1.04
´ 10−5 m2 d−1 (King and Saltzman, 1995), and differential diffusion resulting from layered heterogeneity for a saturated system
would be more pronounced.
In accordance with this study’s main objective of broadly evaluating diffusive transport in a deep desert UZ, results of the SF6
transport modeling, without calibration or other adjustments, were
compared with tracer test data. Figure 6 illustrates the comparison
Fig. 5. Numerical modeling results
for two-dimensional radial transport
simulations of SF6 unsaturated zone
tracer test. The 0-m distance at the
14.9- and 47.9-m depths represents
UZB-1 tracer-injection locations.
Vadose Zone Journal
p. 7 of 10
between observations and simulated results for the four UZB-2
monitoring ports closest in depth to the UZB-1 injection points,
and similar correspondence was observed for the other monitoring depths. In general, the comparisons between observed data
and model results are reasonable, suggesting that gas transport in
the deep UZ at the ADRS is controlled by Fickian diffusion, and
standard formulations for representing diffusivity are good first
approximations. However, notable discrepancies between expectations (model results) and observations exist. The most prominent
discrepancy is the models’ underprediction of peak tracer concentrations. For all UZB-2 monitoring ports, only the model results
for the 5.5-m (not shown) and 57.6-m (Fig. 6d) locations closely
reproduce observed peak concentrations.
With respect to the timing and concentration associated with the
advancing plume fronts, the MQII model reproduces the data
better than the MQI model. The MQII model also shows more
favorable comparison than the MQI model with the breakthrough
curve falling limb concentrations, although declining concentrations for the shallow plume (Fig. 6a,b) are better matched than
those for the deep plume (Fig. 6c,d). The MQI model tends to
overpredict concentrations related to the shallow and deep plume
breakthrough curve falling limbs as compared to observations.
Comparison of the commonly used MQI model results with the
observations indicates slightly faster than expected migration of
SF6 . The slightly faster transport relative to the diffusion-based
model predictions may result from long-range advection, advective
oscillations, or errors in the estimates of Deff. Long-range advection can be ruled out based on results of Maples et al. (2013), who
showed minimal movement of the tritium plume’s center of mass
in a study between 2001 and 2011. Because the direction of contaminant movement from the buried waste was perpendicular to
the line between UZB-1 and UZB-2, any advective transport in
this direction would tend to create the appearance of slower transport and smaller peaks. Testing the potential effects of advective
oscillations and structure of diffusivity on SF6 migration are topics
of a subsequent study.
Fig. 6. Comparison of observed breakthrough curves for selected locations in UZB-2 and modeled responses using Millington–­Quirk I (MQI) and
Millington–Quirk II (MQII) formulations for diffusivity calculations.
Vadose Zone Journal
p. 8 of 10
Implications for Contaminant Gas
Transport Prediction and Evaluation
Predicting migration of contaminants in the gas phase is especially critical in deep arid UZs where much of the United States’
low-level radioactive waste and chemical waste are interred. Gas
diffusion is a primary transport process enabling the spreading of
contaminants present in the gas phase through the UZ beyond
the spatial footprint of low-level radioactive and chemical waste
burial facilities. The utility of standard diffusion theory has not
been validated at the depth and scales (both spatial and temporal)
relevant to long-distance arid UZ transport. Results from the fieldscale ADRS gas tracer study suggest that that the use of empirical
lab-scale formulations, such as the Millington–Quirk expressions
for representing diffusivity, is a reasonable primary approach even
in deep heterogeneous structured UZs. This assessment confirms
the conclusions of an alternative method that tested this utility via
atmospheric gas tracer diffusion into deep UZs on the decadal timescale (Weeks et al., 1982). Overall, the less common MQII model
represents deep UZ diffusion in this large-scale field experiment
better than the widely used MQI model. This result in is agreement
with earlier findings of Jin and Jury (1996), who compared laboratory measured relative diffusion coefficients with those derived
from commonly used empirical models and found the MQII
model to excel compared to the MQI and other models investigated. In terms of this study’s objectives, the difference between
formulations has a relatively minor impact.
Contrasting with previously documented anomalous (theoretically unexpected) long-distance (>100 m) lateral transport of
vapor-phase tritium through the deep UZ at the ADRS, the SF6
tracer experiment results were consistent with existing theory.
The SF6 tracer data were evaluated for excursions reflective of
processes that could explain enhanced tritium transport from a
LLRW disposal area, but no such excursions are identified. The
slight differences in apparent rates of migration of SF6 versus theoretical estimates are not sufficient to explain the long distance
tritium transport at the site. For example, Striegl et al. (1996) and
Mayers et al. (2005) increased diffusion coefficients by an order
of magnitude in forward model runs and still were not able to
match the extent of tritium transport during the lifetime of the
facility. The SF6 results confirming transport controlled by gas
diffusion at the ADRS are consistent with those of a recent 10-yr
field study (2001–2011) of tritium plume dynamics in the shallow
(upper 2 m) UZ at the research site (Maples et al., 2013). In that
study, limited advancement of the plume was consistent with
theoretical estimates, and it was concluded that the observed
widespread distribution of tritium (>400 m from trenches) primarily resulted from transport before 2001. Maples et al. (2013)
proposed that waste-source gas generation within LLRW burial
sites may have contributed to the pre-2001 transport of tritium
from the LLRW area into the adjacent UZ. These relative comparisons between the shallow and deep UZ results are incomplete
Vadose Zone Journal
and uncertain, but they provide support for the hypothesis that
the documented anomalies in long-distance transport at the
ADRS are the result of early time perturbations rather than
ongoing processes and/or time-independent transport properties.
66Conclusions
The in situ experimental gas tracer test described here represents
transport over greater temporal and spatial scales than those
addressed by typical laboratory-scale soil column studies and
other in situ UZ tracer tests reported in the literature to evaluate
soil transport parameters. In general, “tried and true” empirical
relationships for estimating effective diffusion coefficients that
were originally developed at the lab scale adequately describe
gas transport measured in this study’s field-scale experiment in
the deep UZ. However, minor discrepancies were noted. Future
work should address these discrepancies by introducing multiple
approaches of conceptualizing the spatial variability of diffusivity
and evaluating the effect of natural forcings in the UZ.
Under natural conditions, the migration of volatile and semivolatile chemicals within arid UZs is often assumed to be dominated
by gas-phase diffusion. For conditions associated with buried waste
sites, this assumption may be called into question. Waste burial can
impose large gradients in temperature and pressure on the ambient
regime, producing advective and thermal transport that is difficult
to predict or hindcast with insufficient data. The results from the
SF6 tracer test at the ADRS offer confirmation of transport rates
similar to those of gas diffusion under current conditions and at
a distance of ?160 m from the nearest buried waste trench. The
gas transport behavior observed during the year-long tracer test
performed 13 yr after LLRW facility closure suggests that the
natural conditions and UZ soil properties at the ADRS are likely
not responsible for the anomalous vapor-phase transport observed
soon after facility closure. By process of elimination, the results
further suggest that early-time conditions or phase disequilibrium
may play critical roles in contaminant migration through the deep
UZ near buried waste facilities. Careful early-time monitoring is
needed to capture perturbations and reconcile excursions from
predicted diffusive transport.
Acknowledgments
This study was supported by funding from the following USGS programs: Toxic Substances
Hydrology, National Research, and National Water-Quality Assessment. We gratefully acknowledge field and lab support provided by USGS colleagues, Ron Baker, Michael Johnson,
Leigh Justet, Justin Mayers, and Kim Wickland. We thank Ed Weeks (USGS) and two anonymous reviewers for their thoughtful comments that helped improve the manuscript.
References
Andraski, B.J. 1997. Soil-water movement under natural-site and wastesite conditions: A multiple-year field study in the Mojave Desert, Nevada. Water Resour. Res. 33:1901–1916. doi:10.1029/97WR01502
Andraski, B.J., and E.A. Jacobson. 2000. Testing a full-range soil-water retention function in modeling water potential and temperature. Water
Resour. Res. 36:3081–3089. doi:10.1029/2000WR900193
p. 9 of 10
Andraski, B.J., and D.A. Stonestrom. 1999. Overview of research on water,
gas, and radionuclide transport at the Amargosa Desert Research
Site, Nevada. In: D.W. Morganwalp and H.T. Buxton, editors, U.S. Geol.
Surv. Toxic Substances Hydrology Program–Proceedings of the Technical Meeting, Charleston, South Carolina, 8–12 Mar. 1999, Vol. 3 of 3.
Subsurface Contamination from Point Sources. Water-Resour. Invest.
Rep. 99-4018C. USGS, Reston, VA. p. 459–465.
Arthur, J.M., M.J. Johnson, C.J. Mayers, and B.J. Andraski. 2012. Micrometeorological, evapotranspiration, and soil-moisture data at the
Amargosa Desert Research Site in Nye County near Beatty, Nevada,
2006–2011. USGS Data Ser. 725. USGS, Reston, VA.
Baker, R.J., B.J. Andraski, D.A. Stonestrom, and W. Luo. 2012. Volatile organic compounds in the unsaturated zone from radioactive wastes. J.
Environ. Qual. 41:1324–1336. doi:10.2134/jeq2011.0480
Busenberg, E., and L.N. Plummer. 2000. Dating young ground water with
sulfur hexafluoride: Natural and anthropogenic sources of sulfur hexafluoride. Water Resour. Res. 36:3011–3030. doi:10.1029/2000WR900151
Costanza-Robinson, M.S., and M.L. Brusseau. 2006. Gas-phase dispersion
in porous media. In: C. Ho and S. Webb, editors, Gas transport in porous media. Springer, Dordrecht, the Netherlands. p. 121–132
Dirkes, R.L., R.W. Hanf, and T.M. Poston, editors. 1999. Hanford site environmental report for calendar year 1998. Rep. PNNL-12088. Pacific Northwest Natl. Lab, Richland, WA.
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. Water-Resour.
Invest. Rep. 92-4032. USGS, Reston, VA.
Jin, Y., and W.A. Jury. 1996. Characterizing the dependence of gas diffusion coefficient on soil properties. Soil Sci. Soc. Am. J. 60:66–71.
doi:10.2136/sssaj1996.03615995006000010012x
King, D.B., and E.S. Saltzman. 1995. Measurement of the diffusion coefficient of sulfur hexafluoride in water. J. Geophys. Res. 100:7083–7088.
doi:10.1029/94JC03313
Kreamer, D.K., E.P. Weeks, and G.M. Thompson. 1988. A field technique
to measure the tortuosity and sorption-affected porosity for gaseous
diffusion of materials in the unsaturated zone with experimental results
from near Barnwell, South Carolina. Water Resour. Res. 24:331–341.
doi:10.1029/WR024i003p00331
Kuang, X., J.J. Jiao, and H. Li. 2013. Review on airflow in unsaturated
zones induced by natural forcings. Water Resour. Res. 49:6137–6165.
doi:10.1002/wrcr.20416
Maiss, M., L.P. Steele, R.G. Francey, P.J. Fraser, R.L. Langenfelds, N.B.A.
Trivett, and I. Levin. 1996. Sulfur hexafluoride—A powerful new atmospheric tracer. Atmos. Environ. 30:1621–1629. doi:10.1016/13522310(95)00425-4
Maples, S.R., B.J. Andraski, D.A. Stonestrom, C.A. Cooper, G. Pohll, and
R.L. Michel. 2013. Tritium plume dynamics in the shallow unsaturated zone in an arid environment. Vadose Zone J. 12(4). doi:10.2136/
vzj2013.05.0080
Mayers, C.J., B.J. Andraski, C.A. Cooper, S.W. Wheatcraft, D.A. Stonestrom, and R.L. Michel. 2005. Modeling tritium transport through a
deep unsaturated zone in an arid environment. Vadose Zone J. 4:967–
976. doi:10.2136/vzj2004.0179
Millington, R.J. 1959. Gas diffusion in porous media. Science 130:100–102.
doi:10.1126/science.130.3367.100-a
Millington, R.J., and J.P. Quirk. 1960. Transport in porous media. p. 97–106.
In: F.A. Van Beren et al., editors, Trans. Int. Congr. Soil Sci. 7th, Madison,
WI. 14–24 Aug. 1960. Vol. 1. Elsevier, Amsterdam, the Netherlands.
Moldrup, P., T. Olesen, P. Schjønning, T. Yamaguchi, and D.E. Rolston.
2000. Predicting the gas diffusion coefficient in undisturbed soil from
soil water characteristics. Soil Sci. Soc. Am. J. 64:94–100. doi:10.2136/
sssaj2000.64194x
Nichols, W.D. 1987. Geohydrology of the unsaturated zone at the burial
site for low-level radioactive waste near Beatty, Nye County, Nevada.
Water-Supply Pap. 2312. USGS, Reston, VA.
Nicot, J.P., and P.C. Bennett. 1998. Shallow subsurface characterization
of gas transport in the Playa wetland. J. Environ. Eng. 124:1038–1046.
doi:10.1061/(ASCE)0733-9372(1998)124:11(1038)
Prudic, D.E., D.A. Stonestrom, and R.G. Striegl. 1997. Tritium, deuterium,
and oxygen-18 in water collected from unsaturated sediments near a
low-level radioactive-waste burial site south of Beatty. Water-Resour.
Invest. Rep. 97-4062. USGS, Carson City, NV.
Vadose Zone Journal
Prudic, D.E., and R.G. Striegl. 1995. Tritium and radioactive carbon (14C)
analyses of gas collected from unsaturated sediments next to a lowlevel radioactive-waste burial site south of Beatty, Nevada, April 1994
and July 1995. USGS Open-file Rep. 95-471. USGS, Carson City, NV.
Rattray, G. W., R.G. Striegl, and I.C. Yang. 1995. Adsorption of sulfur hexafluoride onto crushed tuffs from the Yucca Mountain area, Nye County, Nevada. Water-Resour. Invest. Rep. 95-4057. USGS, Denver, CO.
Smiles, D.E., W.R. Gardner, and R.K. Schulz. 1995. Diffusion of tritium in arid disposal sites. Water Resour. Res. 31(6):1483–1488.
doi:10.1029/94WR02013
Stonestrom, D.A., J.D. Abraham, B.J. Andraski, R.J., Baker, C.J., Mayers,
R.L., Michel, D.E. Prudic, R.G. Striegl, and M.A. Walvoord. 2004. Monitoring radionuclide contamination in the unsaturated zone—Lessons
learned at the Amargosa Desert Research Site, Nye County, Nevada:
Proceedings, Workshop on long-term performance monitoring of metals and radionuclides in the subsurface, Reston, VA, 20–22 Apr. 2004.
USGS, Menlo Park, CA. http://web.archive.org/web/20120915051957/
http://www.cistems.fsu.edu/PDF/stonestrom.pdf
Striegl, R.G., D.E. Prudic, J.S. Duval, R.W. Healy, E.R. Landa, D.W. Pollock,
D.C. Thorstenson, and E.P. Weeks. 1996. Factors affecting tritium and
14carbon distributions in the unsaturated zone near the low-level radioactive-waste burial site south of Beatty, Nevada. USGS Open-file
Rep. 96-110. USGS, Reston, VA.
Taylor, E.M. 2010. Characterization of geologic deposits in the vicinity of
US Ecology, Amargosa Basin, southern Nevada. Sci. Invest. Rep. 20105134. USGS, Reston, VA.
Tumbusch, M.L., and D.E. Prudic. 2000. Temperatures and water potentials in shallow unsaturated alluvium next to a burial site for low-level
radioactive waste, Amargosa Desert, Nye County, Nevada, 1987-96.
Water-Resour. Invest. Rep. 99-4261. USGS, Reston, VA.
Thibodeaux, L.J. 1981. Estimating the air emissions of chemicals from hazardous waste landfills. J. Hazard. Mater. 4:235–244. doi:10.1016/03043894(81)80002-7
Tick, G.R., C.M. McColl, I. Yolcubal, and M.L. Brusseau. 2007. Gas-phase
diffusive tracer test for the in-situ measurement of tortuosity in the vadose zone. Water Air Soil Pollut. 184:355–362. doi:10.1007/s11270-0079403-3
Walvoord, M.A., D.A. Stonestrom, B.J. Andraski, and R.G. Striegl. 2004.
Constraining the inferred paleohydrologic evolution of a deep unsaturated zone in the Amargosa Desert. Vadose Zone J. 3:502–512.
doi:10.2113/3.2.502
Walvoord, M.A., and D.A. Stonestrom. 2004. Enhanced gas-phase transport in a deep unsaturated zone, Amargosa Desert (U.S.A). In: K.
Kovar, Z. Hrkal, and J. Bruthans, editors, Proceedings of the International Conference on Finite Element Models, MODFLOW, and More:
Solving Groundwater Problems. IAHS, Karlovy Vary, Czech Republic.
p. 227–230.
Walvoord, M.L., B.J. Andraski, D.P. Krabbenhoft, and R.G. Striegl. 2008.
Transport of elemental mercury in the unsaturated zone from a
waste disposal site in an arid region. Appl. Geochem. 23:572–583.
doi:10.1016/j.apgeochem.2007.12.014
Webb, S. 2006. Gas transport mechanisms. In: C. Ho and S. Webb, editors,
Gas transport in porous media. Springer, Dordrecht, The Netherlands.
p. 5–26.
Weeks, E.P., D.E. Earp, and G.M. Thompson. 1982. Use of atmospheric fluorocarbons F-11 and F-12 to determine the diffusion parameters of the
unsaturated zone in the southern high plains of Texas. Water Resour.
Res. 18:1365–1378. doi:10.1029/WR018i005p01365
Weeks, E.P., and P.B. McMahon. 2007. Nitrous oxide fluxes from cultivated areas and rangeland: U.S. High Plains. Vadose Zone J. 6:496–510.
doi:10.2136/vzj2006.0164
Werner, D., P. Grathwohl, and P. Höhener. 2004. Review of field methods
for the determination of the tortuosity and effective gas-phase diffusivity in the vadose zone. Vadose Zone J. 3:1240–1248. doi:10.2136/
vzj2004.1240
Werner, D., and P. Höhener. 2003. In situ method to measure effective
and sorption-affected gas-phase diffusion coefficients in soils. Environ.
Sci. Technol. 37:2502–2510. doi:10.1021/es020101s
Zyvoloski, G.A., B.A. Robinson, Z.V. Dash, and L.L. Trease. 1997. Summary of
the models and methods for the FEHM application—A finite-element
heat-and mass-transfer code. Rep. LA-13307-MS. Los Alamos Natl.
Lab, Los Alamos, NM.
p. 10 of 10