An Assessment of Aquifer Storage Recovery

Using Ground Water Flow Models

by Christopher S. Lowry1 and Mary P. Anderson2

Abstract
Owing to increased demands on ground water accompanied by increased drawdowns, technologies that use
recharge options, such as aquifer storage recovery (ASR), are being used to optimize available water resources
and reduce adverse effects of pumping. In this paper, three representative ground water flow models were created
to assess the impact of hydrogeologic and operational parameters/factors on recovery efficiency of ASR systems.
Flow/particle tracking and solute transport models were used to track the movement of water during injection,
storage, and recovery. Results from particle tracking models consistently produced higher recovery efficiency than
the solute transport models for the parameters/properties examined because the particle tracking models neglected
mixing of the injected and ambient water. Mixing between injected and ambient water affected recovery effi-
ciency. Results from this study demonstrate the interactions between hydrogeologic and operational parameters
on predictions of recovery efficiency. These interactions are best simulated using coupled numerical ground water
flow and transport models that include the effects of mixing of injected water and ambient ground water.

Background reclaimed water, or even water from another aquifer.

Aquifer storage recovery (ASR) systems have been
in operation since the late 1960s in the United States
(Pyne 1994), mostly in coastal areas (e.g., Florida and
California). In ASR, water is stored in an aquifer much
like water is stored in a surface reservoir or tank for later
use. Water is diverted from a source when demand is low
and injected into an aquifer. Injected water then displaces
ambient ground water and creates what conceptually is
termed a ‘‘bubble’’ (Figure 1) or more recently a ‘‘bottle
brush’’ owing to aquifer heterogeneity (Vacher et al.
2006). The injected water is stored in the aquifer until
demand increases when water is recovered. A successful
ASR system requires a suitable aquifer in which to store
the water and a source of water such as a river, lake,
1Corresponding author: Department of Geology and Geo-
physics, University of Wisconsin—Madison, 1215 West Dayton
Street, Madison, WI 53706; lowry@geology.wisc.edu
2Department of Geology and Geophysics, University of
Wisconsin—Madison, 1215 West Dayton Street, Madison, WI 53706
Received October 2005; accepted April 2006.
Copyright ª 2006 The Author(s)
Journal compilation ª 2006 National Ground Water Association.
doi: 10.1111/j.1745-6584.2006.00237.x
Vol. 44, No. 5—GROUND WATER—September–October 2006 (pages 661–667)
In most sites in the United States, water used in ASR
systems is treated before injection inasmuch as injection
of untreated water may pose a risk of contamination.
The term recovery efficiency describes the percent-
age of water that can be recovered after injection. In this
research, hydrogeological controls on recovery efficiency
were evaluated using particle tracking models and solute
transport models, which include effects of both advection
and dispersion. When simulating purely advective trans-
port with a particle tracking code, recovery efficiency is
defined as the percentage of injected water that can be
recovered. When simulating advective and dispersive
transport with a solute transport model, recovery effi-
ciency is defined as the percentage of potable water
recovered based on the volume of injected water. Potable
water is defined based on the concentration of total dis-
solved solids (TDS) resulting from the amount of mixing
between the injected and ambient water. The ASR well is
pumped until the specified level of TDS is reached (usu-
ally 500 mg/L, the secondary drinking water standard).
Few studies on the hydraulic effects of ASR systems
are available in the open literature (Wanless 2004).
Modeling studies relate primarily to the injection of fresh
661

Figure 1. Schematic diagram showing injected water in an
ASR well forming a bubble within ambient ground water for
both the injection and recovery cycles.
water into a saline aquifer, for example, in Florida
(Merritt 1985, 1986; Quinones-Aponte and Wexler 1995;
Yobbi 1996) or in fresh water aquifers (Streetly 1998).
Other previous research includes simulation using physi-
cal laboratory models (Kimbler 1970; Kimbler et al.
1975). Controls on recovery efficiency of fresh water in
saline aquifers generally are classified into three catego-
ries: density-dependent stratification, fresh/saline inter-
face dispersion (mixing), and ambient ground water flow
displacement (Merritt 1986; Kimbler 1970; Moulder
1970). Density-dependent (buoyancy) stratification cau-
ses the zone between fresh and saline water to tilt, creat-
ing a larger interface as a result of density stratification
(Kimbler 1970). The present research does not investigate
density-dependent stratification because both the injected
water and the simulated ground water are fresh water.
Previous studies reported in the literature show that dis-
persion (mixing) between injected water and ambient
ground water decreases recovery efficiency. A few
researchers have investigated ways to limit the loss in
recovery efficiency by artificially controlling head
gradients (Molz and Bell 1977; Whitehead et al. 1978) or
optimizing injection/recovery schedules in multiwell
systems (Merritt 1986).
Although not addressed in this paper, geochemical
processes are important factors in the success of an ASR
system. Geochemical studies of ASR systems include
investigations of disinfection by-products produced by
reactions between the injected chlorinated water and
naturally occurring organic compounds (e.g., Thomas
et al. 2000; Fram et al. 2003). Of primary interest are the
mechanism and time period at which these disinfection
by-products degrade in the aquifer. These studies suggest
that the degradation of disinfection by-products is due to
dilution of the injected water or some combination of
dilution and biotransformation processes. Other studies
have modeled changes in fluoride concentrations dur-
ing injection (Gaus et al. 2002). Most of these studies
have investigated short-term chemical effects, with the
exception of Herczeg et al. (2004), whose 5-year study
662 C.S. Lowry, M.P. Anderson GROUND WATER 44, no. 5: 661–667
reported the effects of injection of storm water into
a brackish carbonate aquifer.
The objective of the current research was to investi-
gate the hydraulic controlling factors on ASR as they
relate to recovery efficiency in fresh water aquifers.
Hydraulic factors were classified into physical and opera-
tional factors. Physical factors are a function of the
ground water flow system and cannot be changed by the
operator. These factors include regional hydraulic gradi-
ent, effective porosity, dispersivity, and the presence of
preferential flow zones in the aquifer. Operational param-
eters investigated include storage period, volume of in-
jected water, and injection/recovery rates, which can be
changed at the wellhead by the operator.
Three representative hydrogeologic settings were in-
vestigated (Figure 2), with the primary focus on a confined
sandstone aquifer representative of the hydrogeological
conditions in southeastern Wisconsin, specifically in the
vicinity of Waukesha, Wisconsin (Figure 2). Other hydro-
geological settings included a glacial drift aquifer system
and an unconfined dolomite aquifer, both of which repre-
sent typical hydrogeological settings in Wisconsin.
Methodology
ASR systems at the three representative sites were
modeled using the finite-difference code MODFLOW
(McDonald and Harbaugh 1988) linked to the particle
tracking code MODPATH (Pollock 1994) and solute
transport code MT3DMS (Zheng and Wang 1999). All
three codes were run using the graphical user interface
Groundwater Vistas (Rumbaugh and Rumbaugh 2004).
MODFLOW simulates the flow field in response to
injection and recovery of water. MODPATH tracks the
movement of imaginary particles by advection at the
average linear velocity calculated from heads generated
by MODFLOW. Particles were added as water was
injected into the aquifer; each particle represents a given
volume of injected water. The particles were tracked
through the duration of the storage and recovery cycles in
Figure 2. Locations of representative hydrogeologic settings
in Wisconsin.
order to quantify the volume of injected water recovered
from the aquifer.
The transport code MT3DMS calculates concentra-
tions of water in the aquifer considering both advection
and dispersion. Inclusion of dispersion to allow for mixing
of injected water with ambient ground water is a better rep-
resentation of the field system. MT3DMS assumes that the
aquifer is an equivalent porous medium with connected
pore space or a well-connected network of fractures. The
code also assumes miscible flow and no density effects.
For most simulations, the governing equation was solved
using the total variation diminishing (TVD) explicit finite-
difference solution technique (Zheng and Wang 1999) to
minimize numerical (artificial) dispersion. For simulations
of the dolomite aquifer, the implicit finite-difference solu-
tion with upstream weighting was used.
The concentration of the injected water was arbitrarily
set to equal 0 units, and the concentration of the ambient
ground water was set to 1000 units. The definition of re-
covery efficiency (i.e., percentage of potable water re-
covered based on the volume of injected water) for the
advection-dispersion model is based on a specified concen-
tration of TDS within the well. Concentration limits of
25% and 50% were specified. The 25% concentration limit
indicates that three-quarters of the water within the well
has a concentration of 0 (representing the injected water)
(Figure 3). The 50% concentration limit is reached when
half of the water in the well is the injected water and half is
ambient ground water. Recovery ends when the concentra-
tion limit has been reached within the production well.
The ASR model of the sandstone aquifer contains 13
layers that represent the unconfined system (layers 1 to
3), the regional confining unit (layers 4 and 5), and the
confined sandstone aquifer (layers 6 to 13) (Figure 4).
Layer thickness and variations in elevation of each model
layer were taken from the Southeastern Wisconsin
regional ground water model (Feinstein et al. 2003). The
ASR well was open to layers 7 to 13, which encompass
Figure 3. Illustration of changes in the concentration at the
ASR well during injection, storage, and recovery.
C.S. Lowry, M.P. Anderson GROUND WATER 44, no. 5: 661–667
a thickness of 558 m (Figure 4). The discretization
around the ASR well was 3 by 3 m. Cell widths
increased, using an expansion factor of 1.5, from 3 m
around the well to 123 m at the model boundary. Lateral
boundary conditions in layers 6 to 13 were set as
constant-head boundaries on the eastern and western sides
of the model based on the potentiometric surface map of
the deep bedrock aquifer (Southeastern Wisconsin
Regional Planning Commission and Wisconsin Geo-
logical and Natural History Survey 2002). Elsewhere, the
boundary conditions were based on results from the
regional model. Hydraulic conductivity and effective
porosity were taken from the regional model of Feinstein
et al. (2003). The base ASR cycle consisted of 39 d of
injection at 3.8 million L/d (1 million gal/d) followed by
90 d of storage and 39 d of recovery at 3.8 million L/d.
Both particle tracking and solute transport simulations
were run. The sensitivity of the model to the full range of
physical and operational parameters/factors, namely hy-
draulic gradient, effective porosity, dispersion, zones of
preferential flow, storage period, volume of injected water,
and injection/recovery rates, was tested.
The zone of injection for the sandstone model (558 m)
is thicker than in previously modeled systems for which
the zone of injections ranged from ~50 to 200 m (Merritt
1985; Quinones-Aponte and Wexler 1994; Yobbi 1996;
Streetly 1998). The thick injection zone creates a long,
narrow bubble rather than the cylindrical bubble usually
pictured (e.g., Figure 1). This bubble has a larger surface
area as compared to a cylindrical bubble with the same
volume of water. The relation between aquifer thickness
and recovery efficiency was not evaluated in this study.
The hydrogeologic setting for the glacial drift aquifer
(Figure 2) was based on an aquifer in the Troy Valley of
Southeastern Wisconsin (Conlon 1991). The model con-
sisted of three horizontal layers, including an upper
unconfined sand aquifer, and a clay confining unit under-
lain by a sandy confined aquifer. Constant-head values
were set at the boundaries to simulate horizontal flow
under nonpumping/injection conditions. The ASR well
was placed in the center of the model and open to layer 3,
the confined aquifer, which is 18 m thick. The hydraulic
gradient across the system was 0.001 m/m with
a pumping rate of 3.8 million L/d (1 million gal/d) and
a storage period of 90 d. Uniform dispersivities of 3, 0.3,
and 0.03 m in the longitudinal, transverse horizontal, and
transverse vertical directions were assumed, respectively.
The sensitivity of the model to regional hydraulic gradi-
ent and volume of injected water was tested.
The dolomite aquifer is representative of the area
around Sturgeon Bay in Northeastern Wisconsin (Figure 2)
where the Silurian Dolomite is an important aquifer.
Hydrogeological parameters were taken from Rayne et al.
(2001). Following Rayne et al. (2001), the three-layer
model accounts for vertical fractures by enhanced vertical
hydraulic conductivity of the system. At the scale of the
ASR system, it was assumed that the aquifer could be
simulated as an equivalent porous medium representing
a network of connected fractures. Rapid transport of
injected water through fractures was simulated by using
an effective porosity of 0.001 inasmuch as ground water
663
Figure 4. Hydrogeological units and model layers for the sandstone aquifer, glacial drift aquifer, and dolomite aquifer models.
velocity is inversely related to effective porosity. Effec-
tive porosity was chosen from a range of values (0.0003
to 0.002) given by Rayne et al. (2001). The ASR well was
open to all three layers of the unconfined system repre-
senting a total thickness of 137 m. The hydraulic gradient
across the system was 0.001 m/m. The period of injection
was 20 d at a rate of 0.9 million L/d (0.25 million gal/d)
for a total volume of 18.9 million L (5 millions gallons),
storage period of 90 d, and recovery period based on the
limit of potable water recovered. Injection and recovery
rates were limited by potential mounding of water during
injection and dewatering during recovery. Dispersivities
were 3, 0.3, and 0.03 m in the longitudinal, transverse
horizontal, and transverse vertical directions, respectively.
The sensitivity of the model to regional hydraulic gradi-
ent and volume of injected water was tested.
The low effective porosity and resulting high ground
water velocities in the unconfined dolomite model caused
the TVD solution in MT3DMS to require a transport time
step on the order of 1025 d, causing long run times. Hence,
the implicit finite-difference solution with upstream weight-
ing was used as it did not require such small time steps.
Because the implicit solution can be affected by numerical
dispersion, the results were checked for possible numerical
dispersion by running the model with an effective porosity
Figure 5. Effect of hydraulic gradient on recovery efficiency. (A) the confined sandstone aquifer; (B) advection-dispersion
models of the glacial drift aquifer and the dolomite aquifer.
664 C.S. Lowry, M.P. Anderson GROUND WATER 44, no. 5: 661–667
of 0.01 for both TVD and implicit finite-difference sol-
utions. Both solutions produced similar concentrations at
the ASR well (Lowry 2004).
Results and Discussion
Simulations of ASR in all three hydrogeologic set-
tings produced similar results with respect to the relative
effect of each hydraulic factor on recovery efficiency as
summarized subsequently and discussed in detail by
Lowry (2004).
Mixing is an important factor governing recovery
efficiency. The advection-only models for the sandstone
aquifer gave a 30% higher recovery than the advection-
dispersion models (Figure 5A). Furthermore, recovery
efficiency decreased as regional hydraulic gradient was
increased (Figures 5A and 5B); a high regional gradient
allows water to flow away from the ASR well. Higher
hydraulic conductivity in the glacial drift and dolomite
models (Figure 5B) caused a significant decrease in re-
covery efficiency as compared to the sandstone advection-
dispersion model (Table 1). Hydraulic gradient becomes
an even more sensitive parameter when smaller volumes
of water are injected and in combination with high values
of hydraulic conductivity and low effective porosity
(Lowry 2004).
 To simulate the effect of preferential flow, an in-
Table 1 crease in the hydraulic conductivity of layer 10 (Figure 4)
Model Properties in the sandstone aquifer model was assumed. During in-
jection, water moves preferentially from the ASR well
Hydraulic
through the layer of high hydraulic conductivity and is
Thickness Conductivity Effective
drawn back rapidly to the ASR well through the same
Model Layer (m) (m/d) Porosity
layer during recovery, causing an increase in recovery
Dolomite 1 46 1.00E100 0.001 efficiency (Figure 9). As hydraulic conductivity of the
aquifer 2 46 1.00E100 0.001 layer is increased, there is a point at which some portion
3 46 1.00E100 0.001 of injected water moves beyond the capture zone of
Glacial 1 24 3.05E100 0.15 the ASR well, causing a decrease in recovery efficiency
drift aquifer 2 18 3.05E203 0.15 (Figure 9). This same effect would occur in aquifers with
3 18 3.05E101 0.30
connected bedding plane fractures.
Sandstone 7 82 1.22E202 0.1
aquifer 8 45 7.32E201 0.25
9 53 3.66E201 0.25 Conclusions
10 61 3.66E201 0.25
Interrelated hydraulic and operational factors that
11 91 3.66E201 0.25
12 91 3.66E201 0.25
control recovery efficiency in ASR systems cause the pre-
13 134 7.32E201 0.25 diction of recovery efficiency to be a complex problem
best solved using coupled numerical ground water flow
and solute transport models that include the effects of
mixing between injected water and ambient ground
water. Particle tracking simulations seriously overpredict
Dispersivity and advective velocity directly affect the recovery rates as compared to solute transport models, by
dispersion or mixing between ambient and injected water. as much as 30% for the system studied. While mixing
Increased mixing causes a decrease in recovery effi- between injected and ambient water is an important pro-
ciency. Recovery efficiency in the sandstone aquifer cess affecting recovery efficiency, dispersivity, which
model was calculated for concentration limits of 50% and describes the mixing process, is one of the hardest param-
25%, while varying longitudinal dispersivity from 1.8 to eters to quantify. The use of particle tracking models to
9.1 m. Values of dispersivity are uncertain in most geo- simulate ASR systems should be limited to sites where
logical settings (Schulze-Makuch 2005). We assumed injected and ambient water are of the same chemistry or
horizontal transverse dispersivity is one-tenth of longi- when mixing is not an issue. (For the ASR simulations
tudinal dispersivity and vertical transverse dispersivity is considered here, the particle tracking and solute transport
one-tenth of horizontal transverse dispersivity. The rate of models required similar amounts of preparation and run-
change in recovery efficiency decreased as dispersivity ning time.) It is also noteworthy that results suggest that
was increased (Figure 6).
As effective porosity is inversely related to ground
water velocity, effective porosity directly affects recovery
efficiency. Combining low values of effective porosity
with a high hydraulic gradient or long storage periods
causes extremely low recovery efficiencies (<40%)
(Lowry 2004). Although effective porosity is a difficult
parameter to quantify, it varies over a much smaller range
than hydraulic gradient or dispersivity.
Recovery efficiency decreased in a nearly linear fash-
ion as storage period was increased (Figure 7). Recovery
efficiency consistently increased in each of the systems as
volume of injected water was increased, except for the
unconfined dolomite aquifer (Figure 8) where recovery
efficiency decreased after an initial increase. Recovery
efficiency approaches a maximum in the two confined
systems (sandstone aquifer and glacial drift aquifer) and
appears to begin to level off to a value dependent on the
combination of parameters specific to each setting. In the
dolomite aquifer, recovery efficiency initially increased Figure 6. Effect of dispersivity on recovery efficiency for the
with volume of injected water and then decreased as the sandstone aquifer model. The 50% limit refers to the point
large volume of injected water created high hydraulic gra- at which recovery has ended, and the concentration in the
dients near the ASR well, causing water to move rapidly well represents 50% injected water and 50% ambient water.
away from the well. The same phenomenon may occur in The 25% limit refers to the point at which recovery has
ended, and the concentration in the well represents 75% in-
the sandstone aquifer and glacial drift aquifer with larger jected water and 25% ambient water.
volumes of injected water.
C.S. Lowry, M.P. Anderson GROUND WATER 44, no. 5: 661–667 665
Figure 7. Effect of storage period on recovery efficiency for
the sandstone aquifer model using an advection-dispersion
model.
pilot tests using small volumes of injected water may be
misleading.
This research demonstrates that the effect of key
hydraulic factors, both physical and operational, on the
recovery efficiency of an ASR system can be evaluated
using ground water models. Through a better understand-
ing of the hydraulic factors controlling recovery efficiency
and the interacting effects of ASR operational factors,
water utilities can determine if ASR is potentially viable
given the local hydrogeology and water resources needs.
Figure 8. Comparison of the effect of volume of injected
water on recovery efficiency for the three systems with longi-
tudinal dispersivity of 3 m, gradient of 0.001 m/m, storage
period of 90 d, and a concentration limit of 50%. All simu-
lations include effects of advection and dispersion.
666 C.S. Lowry, M.P. Anderson GROUND WATER 44, no. 5: 661–667
Figure 9. Effect of an increase in hydraulic conductivity of
layer 10 on recovery efficiency for a concentration limit of
50% for the sandstone aquifer model.
Acknowledgments
Support for this study was provided by the Wisconsin
Groundwater Coordinating Council and the Department of
Geology and Geophysics at the University of Wiscon-
sin—Madison. John Jansen provided guidance for the
sandstone aquifer simulations. Helpful reviews were pro-
vided by two anonymous reviewers as well as Keith Half-
ord who also acted as editor in chief for the paper.
References
Conlon, T.D. 1991. A hydrogeologic investigation of a sand-
and-gravel aquifer using reflection seismology and ground-
water flow modeling. M.S. thesis, Department of Geology
and Geophysics, University of Wisconsin—Madison.
Feinstein, D.T., T.T. Eaton, D.J. Hart, J.T. Krohelski, and K.R.
Bradbury. 2003. Regional aquifer model for southeastern
Wisconsin. Report 1: Data collection, conceptual model
development, numerical model construction, and model
calibration. Administrative Report to the Southeastern
Wisconsin Regional Planning Commission: Waukesha,
Wisconsin.
Fram, M.S., B.A. Bergamashci, K.D. Goodwin, R. Jujii, and J.F.
Clark. 2003. Processes affecting the trihalomethane con-
centrations associated with the third injection, storage and
recovery test at Lancaster, Antelope Valley, California,
March 1998 through April 1999. USGS Water Resources
Investigations Report 03-4062.
Gaus, I., P. Shand, I.N. Gale, A.T. Williams, and J.C. Eastwood.
2002. Geochemical modeling of fluoride concentration
changes during aquifer storage and recovery (ASR) in the
Chalk aquifer in Wessex, England. Quarterly Journal of
Engineering Geology and Hydrogeology 35, no. 2: 203–
208.
Herczeg, A.L., K.J. Rattray, P.J. Dillon, P. Pavelic, and E.B.
Karen. 2004. Geochemical processes during five years of
aquifer storage recovery. Ground Water 42, no. 3: 438–445.
Kimbler, O.K. 1970. Fluid model studies of the storage of fresh-
water in saline aquifers. Water Resources Research 6, no. 5:
1522–1527.
Kimbler, O.K., R.G. Kazmann, and W.R. Whitehead. 1975.
Cyclic storage of fresh water in saline aquifers. Louisiana
Water Resources Research Institute Bulletin 10. Baton
Rouge, Louisiana: Louisiana State University.
Lowry, C.S. 2004. Assessment of aquifer storage recovery:
Defining hydraulic controls on recovery efficiency at three
representative sites in Wisconsin. M.S. thesis, Department
of Geology and Geophysics, University of Wisconsin—
Madison.
McDonald, M.G., and A.W. Harbaugh. 1988. A modular three-
dimensional finite-difference ground water flow model.
USGS Techniques of Water-Resources Investigation
Report, v. 06-A1. 586 p.
Merritt, M.L. 1986. Recovering fresh water stored in
saline limestone aquifers. Ground Water 24, no. 4: 516–
529.
Merritt, M.L. 1985. Subsurface storage of freshwater in south
Florida: A digital model analysis of recoverability. USGS
Water-Supply Paper 2261.
Molz, F.J., and L.C. Bell. 1977. Head gradient control in aqui-
fers used for fluid storage. Water Resources Research 13,
no. 4: 795–798.
Moulder, E.A. 1970. Freshwater bubbles: A possibility for using
saline aquifers to store water. Water Resources Research 6,
no. 5: 1528–1531.
Pollock, D.W. 1994. User’s Guide for MODPATH/MODPATH-
PLOT, version 3: A particle tracking post-processing
package for MODFLOW, the U.S. Geological survey finite-
difference ground-water flow model. USGS Open-File
Report 94-464. USGS: Reston, Virginia.
Pyne, R.D.G. 1994. Groundwater Recharge and Wells, 1st ed.
Boca Raton, Florida: Lewis Publishers.
Quinones-Aponte, V., and E.J. Wexler. 1995. Preliminary assess-
ment of injection, storage, and recovery of freshwater in the
lower hawthorn aquifer, Cape Coral, Florida. USGS Water
Resources Investigations Report 94-4121. USGS: Talla-
hassee, Florida.
Rayne, T.W., K.R. Bradbury, and M.A. Muldoon. 2001.
Delineation of capture zones for municipal wells in frac-
C.S. Lowry, M.P. Anderson GROUND WATER 44, no. 5: 661–667
tured dolomite, Sturgeon Bay, Wisconsin, USA. Hydro-
geology Journal 9, no. 5: 432–450.
Rumbaugh, J.O., and D.B. Rumbaugh. 2004. Groundwater Vis-
tas. Reinholds, Pennsylvania: Environmental Simulations
Inc. www.groundwatermodels.com.
Schulze-Makuch, D. 2005. Longitudinal dispersivity data and
implications for scaling behavior. Ground Water 43, no. 3:
443–456.
Southeastern Wisconsin Regional Planning Commission, and Wis-
consin Geological and Natural History Survey. 2002. Ground
water resources of southeastern Wisconsin. Waukesha, Wis-
consin: Southeastern Wisconsin Regional Planning Commis-
sion, Wisconsin Geological and Natural History Survey.
Streetly, M.J. 1998. The use of modelling to predict the behavior of
ASR systems. In Artificial Recharge of Ground Water, ed.
Peter et al., 263–267. Rotterdam, Netherlands: A.A. Balkema.
Thomas, J.M., W.A. McKay, E. Cole, J.E. Landmeyer, and P.M.
Bradly. 2000. The fate of haloaceticacid and trihalo-
methanes in an aquifer storage and recovery program, Las
Vegas, Nevada. Ground Water 38, no. 4: 605–614.
Vacher, H.L., W.C. Hutchings, and D.A. Budd. 2006. Metaphors
and models: The ASR bubble in the Floridan aquifer.
Ground Water 44, no. 2: 149–154.
Wanless, H.R. 2004. Keep it clean and do the research. Ground
Water 42, no. 2: 157.
Whitehead, W.R., R.G. Kazmann, and O.K. Kimbler. 1978.
Comment on ‘Head Gradient Control in Aquifers Used for
Fluid Storage’ by Fred J. Molz and Lansford C. Bell. Water
Resources Research 14, no. 2: 381–384
Yobbi, D.K. 1996. Simulation of subsurface storage and recov-
ery of treated effluent injected in a saline aquifer, St.
Petersburg, Florida. USGS Water-Resources investigations
report 95-4271. USGS: Tallahassee, Florida.
Zheng, C., and P.P. Wang. 1999. MT3DMS: A modular three-
dimensional multi-species model for simulation of advec-
tion, dispersion and chemical reactions of contaminants in
ground water systems: Documentation and user’s guide.
SERDP-99-1. Vicksburg, Mississippi: U.S. Army Engineer
Research and Development Center.
667