Direct-push hydrostratigraphic profiling: coupling electrical logging and slug tests.
ABSTRACT Spatial variations in hydraulic conductivity (K) can significantly affect the transport of contaminants in ground water. Conventional field methods, however, rarely provide a description of these variations at the level of detail necessary for reliable transport predictions and effective remediation designs. A direct-push (DP) method, hydrostratigraphic profiling, has been developed to characterize the spatial variability of both electrical conductivity (EC) and hydraulic conductivity in unconsolidated formations in a cost-effective manner. This method couples a dual-rod approach for performing slug tests in DP equipment with high-resolution EC logging. The method was evaluated at an extensively studied site in the Kansas River floodplain. A series of profiles was performed on a surface grid, resulting in a detailed depiction of the three-dimensional distribution of EC and K. Good agreement was found between K estimates obtained from this approach and those obtained using other methods. The results of the field evaluation indicate that DP hydrostratigraphic profiling is a promising method for obtaining detailed information about spatial variations in subsurface properties without the need for permanent wells.
Article: Cost-effective hydraulic tomography surveys for predicting flow and transport in heterogeneous aquifers.[show abstract] [hide abstract]
ABSTRACT: This study shows how a cost-effective hydraulic tomography survey (HTS) and the associated data estimator can be used to characterize flow and transport in heterogeneous aquifers. The HTS is an improved field hydraulic test that accounts for responses of hydraulic stresses caused by pumping or injection events at different locations of an aquifer. A sequential data assimilation method based on a cokriging algorithm is then used to map the aquifer hydraulic conductivity (K). This study uses a synthetic two-dimensional aquifer to assess the accuracy of predicted concentration breakthrough curves (BTCs) on the basis of the Kfields estimated by geometric mean, kriging, and HTS. Such Kfields represent different degrees of flow resolutions as compared with the synthetically generated one. Without intensive experimentsto calibrate accurate dispersivities at sites, the flow field based on the HTS Kfield can yield accurate predictions of BTC peaks and phases. On the basis of calculating mean absolute and square errors for estimated K fields, numerical assessments on the HTS operation strategy show that more pumping events will generally lead to more accurate estimations of Kfields, and the pump locations need to be installed in high Kzones to maximize the delivery of head information from pumps to measurement points. Additionally, the appropriate distances of installed wells are suggested to be less than one-third of the ln(K) correlation length in x direction.Environmental Science and Technology 06/2009; 43(10):3720-7. · 5.23 Impact Factor
Spatial variations in hydraulic conductivity (K) can
have a significant impact on the transport of contaminants
in ground water. An understanding of the three-dimen-
sional distribution of K at a site is often necessary for reli-
able transport predictions and the design of effective
remediation strategies. Conventional field methods, how-
ever, can rarely provide information on the spatial varia-
tions in K at the needed level of detail and accuracy in a
cost-effective manner. For example, pumping tests provide
a large-scale average of K (Butler and Liu 1993; Meier et
al. 1998), while methods based on empirical relationships
or laboratory analyses provide estimates with a high degree
of uncertainty (Butler and Bahr 1988; Farrar 1996; Lunne
et al. 1997). Single-well hydraulic tests (e.g., slug tests,
dipole-flow tests) can be an effective means of obtaining
high-resolution K data (Yeh et al. 1995; Zlotnik and Zur-
buchen 1998), but the cost of well installation is often sub-
stantial. As a result, there is rarely sufficient information at
a site to assess the influence of spatial variations in K on
contaminant transport. Thus, there is a clear need to
develop site characterization methods that enable informa-
tion about the three-dimensional distribution of K to be
determined in a time- and cost-efficient manner (National
Research Council 1994). The development of one such
method is described in this paper.
Over the last two decades, direct-push (DP) technol-
ogy has become a widely used alternative to conventional
drilling-based methods for site characterization investiga-
tions in unconsolidated formations. This technology
involves advancing a small-diameter rod string, with a sen-
sor/tool at its lower end, using hydraulic rams and the
weight of a large truck (cone penetrometer technology) or
a combination of hydraulic rams and a high-frequency
hammer (Geoprobe Systems 1998). DP technology is cur-
rently used in environmental investigations for a variety of
applications including soil coring, ground water and soil
Spatial variations in hydraulic conductivity (K) can significantly affect the transport of contaminants in ground
water. Conventional field methods, however, rarely provide a description of these variations at the level of detail nec-
essary for reliable transport predictions and effective remediation designs. A direct-push (DP) method, hydrostrati-
graphic profiling, has been developed to characterize the spatial variability of both electrical conductivity (EC) and
hydraulic conductivity in unconsolidated formations in a cost-effective manner. This method couples a dual-rod
approach for performing slug tests in DP equipment with high-resolution EC logging. The method was evaluated at
an extensively studied site in the Kansas River floodplain. A series of profiles was performed on a surface grid, result-
ing in a detailed depiction of the three-dimensional distribution of EC and K. Good agreement was found between K
estimates obtained from this approach and those obtained using other methods. The results of the field evaluation
indicate that DP hydrostratigraphic profiling is a promising method for obtaining detailed information about spatial
variations in subsurface properties without the need for permanent wells.
Direct-Push Hydrostratigraphic Profiling:
Coupling Electrical Logging and Slug Tests
by Stephen M. Sellwood1, John M. Healey2, Steffen Birk3, and James J. Butler Jr.4
1Formerly with Kansas Geological Survey, 1930 Constant Ave.,
Campus West, University of Kansas, Lawrence, KS 66047; currently
with BT2Inc., 2830 Dairy Dr., Madison, WI 53718
2Kansas Geological Survey, 1930 Constant Ave., Campus
West, University of Kansas, Lawrence, KS 66047
3Formerly with Kansas Geological Survey, 1930 Constant Ave.,
Campus West, University of Kansas, Lawrence, KS 66047; currently
with the Center for Applied Geoscience, University of Tübingen,
Sigwartstr. 10, D–72076 Tübingen, Germany
4Corresponding author: Kansas Geological Survey, 1930 Con-
stant Ave., Campus West, University of Kansas, Lawrence, KS
66047; (785) 864–2116; fax 785-864-5317; firstname.lastname@example.org
Received March 2003, accepted March 2004.
Published in 2004 by the National Ground Water Association.
Vol. 43, No. 1— GROUND WATER— January–February 2005 (pages 19–29)
gas sampling, and electrical-conductivity (EC) and cone-
penetrometer logging (Jacobs et al. 2000; Kram et al.
2000). Several efforts have been made to estimate K with
DP technology through various modifications of conven-
tional hydraulic tests (Hinsby et al. 1992; Cho et al. 2000;
Butler et al. 2002; Butler 2002; McCall et al. 2002). Butler
et al. (2002) and McCall et al. (2002) have shown that slug
testing in DP equipment is an effective means of obtaining
detailed and reliable information about K in permeable for-
mations. Their work serves as the basis for the method
Butler et al. (2002) describe a method for performing
slug tests in DP equipment in which a screen is driven
within protective steel casing to the target depth and then
exposed to the formation for the performance of slug tests.
Although the resulting K estimates compare favorably with
estimates obtained from slug tests and other approaches at
conventional monitoring wells (Butler et al. 2002; Butler
2002), the screen cannot be reshielded downhole. Thus, the
equipment has to be brought to the surface and redriven to
reach multiple levels, making the procedure rather ineffi-
cient for characterizing spatial variations in K. McCall et al.
(2002) significantly increased the efficiency of the proce-
dure by using a pair of nested DP rod strings that are driven
simultaneously into the subsurface. A solid drive point is
attached to the lower end of the inner rod string and rests
inside the cutting shoe of the outer rods (Butler et al. 2000).
The rod strings are advanced together until a test interval is
reached. At that point, the inside rods are removed, leaving
the outer rods with an open hole through the cutting shoe at
the bottom. A screen of a user-specified length is lowered
to the bottom of the rod string and then held in place while
the outer rods are pulled up. This upward movement of the
rods exposes the screen to the formation and creates a tem-
porary screened interval that can be used for slug testing,
water sampling, and head measurements. After all tasks
have been completed at a level, the screen is removed, the
inner rods and drive point are reinserted, and the nested rod
string is driven down to the next level. McCall et al. (2002)
have demonstrated that this approach, hydraulic profiling,
can be used to obtain vertical profiles of K at a level of
detail and accuracy that has not previously been possible in
the absence of permanent wells.
Despite the success of the initial implementation of the
hydraulic profiling approach, its effectiveness is hindered
by three important limitations. First, the inner rod string
must be removed and then reinserted at each test level,
adding significant time to the profiling process. Second,
water must be added to the rod string before and during the
removal of the inner rods to prevent sediments from enter-
ing and clogging the outer rods; this addition of water can
affect water chemistry. Finally, selection of test intervals is
based on information from nearby wells or borings, not on
information at the actual profile location, so potentially
important intervals can be overlooked.
The objective of the work reported here was to modify
the method of McCall et al. (2002) to eliminate the three
stated limitations. The resulting modified approach couples
DP EC logging with hydraulic profiling, so that informa-
tion about the electrical and hydraulic properties of the for-
mation can be obtained in the same probe hole. This
coupling creates a method for characterizing site hydros-
tratigraphy, which for this work is defined as the spatial
arrangement and hydraulic properties of units with differ-
ent proportions of clay, at a level of detail that has not pre-
viously been possible. The purpose of this paper is to
describe this new method, hydrostratigraphic profiling, and
an initial field evaluation of its potential.
This work took place at the Geohydrologic Experi-
mental and Monitoring Site (GEMS), a research site of the
Kansas Geological Survey (KGS) that lies within the flood-
plain of the Kansas River just north of Lawrence, Kansas
(Figure 1). The shallow subsurface at the site consists of
~11.5 m of mostly silt and clay overlying ~10.7 m of coarse
sand and gravel resting on bedrock. Much previous work
has taken place within the sand and gravel interval at
GEMS, including well- and DP-based slug tests, pumping
tests, EC logging, dipole-flow tests, and an induced-gradi-
ent tracer test (Butler et al. 1998; Butler et al. 1999; Butler
et al. 2002; Bohling 1999; Schulmeister et al. 2003). This
previous work enables the hydrostratigraphic profiling
method described here to be evaluated in a controlled field
In this project, a series of DP profiles was performed on
a surface grid in an area of GEMS where a great deal of pre-
vious research on conventional and DP hydraulic tests had
been done. Nine profiles (HP 1 through HP 9) were com-
pleted with ~6 m spacing between the grid points (Figure 1).
At each grid point, an EC log was completed at, or near,
bedrock, and K estimates were obtained from slug tests at
seven to nine levels within the sand and gravel interval.
As stated previously, the hydrostratigraphic profiling
approach presented here is a modification of the hydraulic
profiling method of McCall et al. (2002). The major modi-
fications to the McCall et al. method are (1) the incorpora-
tion of DP EC logging into the procedure to allow selection
of test intervals based on information at the profiling loca-
tion, and (2) the performance of the hydraulic profile from
the bottom up, i.e., as the rods are retracted as opposed to
as the rods are advanced, to minimize the need for the addi-
tion of water and to streamline the procedure. The profiling
procedure that results from these modifications is schemat-
ically summarized in Figure 2.
The first phase of the hydrostratigraphic profiling pro-
cedure consists of EC logging, during which an EC probe
is advanced from the surface through the intervals of inter-
est. This is accomplished using a pair of nested rod strings
(inner rods with 0.025 m OD, outer rods with 0.038 m ID
and 0.054 m OD) with the EC probe connected to the inner
rod string. The lower end of the inner rod string consists of
a dipole EC probe (Geoprobe SC300, Geoprobe Systems,
Salina, Kansas) and solid plug attached to a 0.025 m OD
rod of 1.22 m length. This rod is placed within a 0.038 m
ID rod of 1.22 m length that has a cutting shoe attached to
its lower end. The EC probe and solid plug were machined
at the KGS so that they seat within the cutting shoe, while
S.M. Sellwood et al. GROUNDWATER43, no. 1: 19–2920
leaving 0.070 m of the probe extending beyond the lower
end of the shoe (Figure 3a) (Healey and Sellwood 2004).
This configuration allows measurement of the EC of rela-
tively undisturbed material.
The nesting of a 0.025 m OD rod inside a 0.038 m ID
rod leaves an annular space for the EC probe wire, which
connects the EC probe to acquisition and processing equip-
ment at the surface. A drive-cap assemblage was machined
at the KGS to drive both sets of rods without damaging the
probe wire. As the rods are advanced, inner and outer rods
are added at 1.22 m intervals. As outer rods are added, each
joint must be sealed with O-rings or tape to prevent leakage
during the slug tests. Note that the inner and outer rods are
prenested with the EC probe wire strung between to allow
rapid attachment of new rods during logging.
The nested rods are driven through the intervals of
interest, which at GEMS was to the top of bedrock at ~22 m
below land surface. A measurement of the EC of the sedi-
ments in the vicinity of the probe is taken every 0.015 m as
the probe is advanced (Christy et al. 1994), resulting in a
near-continuous profile of EC vs. depth. At GEMS, where
variations in pore fluid chemistry are small and the sedi-
ments are saturated below depths of 3 to 3.5 m, EC varia-
tions are primarily a function of variations in clay content
(Schulmeister et al. 2003). In this case, the EC logs can be
useful in identifying zones for hydraulic testing. At other
sites, EC variations may also be a function of variations in
pore fluid chemistry or fluid-filled porosity, so supplemen-
tary information, such as core and water samples, should
always be collected to determine the major controls on EC
at the site under investigation.
Once the nested rods have been driven below the inter-
vals of interest, the EC probe is removed along with the
inner rod string. Because the rod assemblage remains
empty of water during EC logging, there is a large
hydraulic gradient between the interior of the rod assem-
blage and the formation, similar to the situation discussed
by Butler et al. (2002). Water must be added to the interior
of the rods before and while the EC probe and inner rod
string are removed to neutralize the initial gradient and that
created during rod removal. If this is not done, water will
surge into the interior of the rods as the inner rods are
removed. This surging water can entrain sediments and
transport them into the interior of the rods (analogous to the
heaving sands discussed by Hackett  and others),
thereby clogging the rod string. This is the only time during
the profiling process at which the addition of water is
required, and the amount added can be easily monitored at
the surface. In addition, the water is usually added at some
distance below the intervals of interest. Thus, the added
water should have a minimal effect on the chemistry of
water samples collected from those intervals. A conserva-
tive tracer can be mixed with the added water if there is a
need to assess how later water samples are affected.
Once the EC probe has been removed with the inner
rod string, the outer rods are open to the formation through
the bottom of the cutting shoe. After determining the inter-
vals for slug testing, the outer rods are retracted until the
cutting shoe is just below the bottom of the lowest test
interval. The total depth inside the outer rods can then be
measured to assess if the formation has collapsed back to
fill the small-diameter hole left by rod retraction. In sand
and gravel formations such as at GEMS, the formation will
collapse back in a matter of seconds. In more cohesive
materials, however, the probe hole may remain open. In
that case, a small volume of grout can be pumped into the
hole through polyethylene tubing to prevent vertical flow in
the probe hole during the slug tests.
When it has been determined that the probe hole has
collapsed (or after grouting), a screen (0.027 m OD Sched-
ule 40 10 slot [0.25 mm] PVC in this work) is inserted into
the rod string and held in a fixed position at the bottom of
the cutting shoe with small-diameter metal rods (extension
rods) as the outer rods are retracted. The screen has a
threaded coupling at its top, which has a larger diameter
than the ID of the cutting shoe. As the outer rods are
retracted, the screen is progressively exposed until the cut-
ting shoe reaches the coupling (Figure 3b). For this study,
the total length of the screen was 0.35 m, which allowed
0.305 m of screen exposure when the screen was seated in
the cutting shoe. This length can be varied to accommo-
date the requirements of a particular investigation. Once
the screen is completely exposed, i.e., seated in the cutting
shoe and rising along with the outer rods, the extension
rods are removed and the outer rods are pulled up a short
distance to bring the screen opposite the exact interval of
interest. The top of the outer rods is then surveyed with
respect to a site datum. The length of the rod string is sub-
tracted from the surveyed elevation to determine the ele-
vation at the top of the test interval. In addition, the total
length of the rod string and screen is measured for each test
interval and compared to the expected length to ensure that
the screen is fully exposed and that it has not partially
filled with sediments.
21S.M. Sellwood et al. GROUNDWATER43, no. 1: 19–29
Figure 1. Site location map for GEMS with inset showing
profiling locations (only wells or profiling locations referred
to in text are shown in inset).
Once the screen has been positioned at the desired
depth, the test interval is developed. Most intervals were
developed in this study by pumping with a centrifugal suc-
tion pump connected to a plastic hose with a foot valve at its
lower end. The hose and foot valve assembly was used to
periodically surge the screened interval to mobilize fine
material that was then removed by pumping. Pumping rates
varied from 0.06 to 0.3 L/s and pumping continued until
water remained clear after surging. Zones that could not be
pumped with the centrifugal pump were hand bailed or man-
ually pumped with the hose and foot valve assembly. Devel-
opment generally took between 20 and 40 min per interval.
When all activities planned for a given level have been
completed, the screen is retrieved with the extension rods
and the outer rods are pulled up until the cutting shoe is
again just below the next test interval. This process is
repeated until all of the desired intervals have been tested.
The screen is removed between test intervals as a precau-
tion to prevent smearing that can be caused by dragging the
screen through clay layers.
Nine hydrostratigraphic profiles were completed in
this fashion, resulting in an EC profile and a K profile at the
nine locations shown in Figure 1. Slug tests were per-
formed at an average of eight levels per profile. Although
the intervals for slug testing can be determined from the
high-resolution EC log, slug tests for this study were per-
formed at predetermined levels that corresponded to inter-
vals used in previous investigations. These levels were
approximately evenly spaced over the thickness of the sand
and gravel aquifer at GEMS.
Slug Test Procedures
Slug tests were performed at each test interval to obtain
a K estimate for that interval. Following the guidelines set
forth in Butler et al. (1996) and Butler et al. (2003), a series
of six to 10 slug tests was performed at each interval using
a range of initial displacements (H0). As has been discussed
in previous work, comparison of repeat slug tests from the
same interval allows the viability of the theory underlying
the analysis models to be assessed and aids in the selection
of the most appropriate model for test analysis (Butler 1998;
McElwee and Zenner 1998; Butler et al. 2003).
S.M. Sellwood et al. GROUNDWATER43, no. 1: 19–29 22
Figure 2. Schematic diagram showing the hydrostratigraphic profiling procedure.
23 S.M. Sellwood et al. GROUNDWATER43, no. 1: 19–29
Figure 3. (a) EC probe at the lower end of the nested DP rod string (configuration as during logging). (b) Fully exposed screen
extending from the DP rod string.
All tests were initiated pneumatically as described in
Butler (1998) and Butler et al. (2002). Most tests were per-
formed in a rising-head mode using pressurized nitrogen gas
to create the initial head displacement. For a number of inter-
vals, tests were also performed in a falling-head mode using
a vacuum pump to create the initial displacement (Hinsby et
al. 1992). Comparison of falling- and rising-head tests per-
formed in the same interval is a further means of assessing
the viability of the theory underlying the analysis methods. A
reproducible dependence on the direction of slug-induced
water flow can indicate that further development is needed
(Butler 1998; Butler and Healey 1998).
Changes in head during a slug test were measured using
a pressure transducer (In-Situ PXD–261 0–20 psig trans-
ducer, Fort Collins, Colorado) connected to a datalogger
(Campbell Scientific CR23X, acquisition rate of 5 Hz,
Logan, Utah), that allowed real-time monitoring of test data
from a laptop computer). The pressure transducer was
placed in the water column within 0.5 m of the static water
level to minimize acceleration effects (McElwee 2001; Zur-
buchen et al. 2002; Butler et al. 2003). For the rising-head
tests, pressure in the air column within the rods was also
monitored prior to test initiation using a second pressure
transducer (In-Situ PXD–261). For falling-head tests, pres-
sure in the air column within the rods was monitored prior
to test initiation using a vacuum gauge.
Test data were converted into the form of normalized
displacement—H(t)/H0, where H(t) is the deviation from sta-
tic—vs. time since the start of the slug test and then analyzed
using one of two models. The majority of tests performed for
this project were analyzed using a spreadsheet procedure for
slug tests in highly permeable aquifers described in Butler et
al. (2003). This procedure, which implements a high K
extension of the Hvorslev model (Butler 1998), uses a
spreadsheet to process and plot test data, and then manually
fits the normalized data plots to type curves. A recent imple-
mentation of this high K extension of the Hvorslev model in
AQTESOLV (HydroSOLVE Inc. 2001) (designated as But-
ler  model) was used at the end of the project to check
the manual fits. In all cases, the K estimates obtained from
the manual and automatic fits were in good agreement
(within a few percent).
The model described previously is based on the
assumption that slug test data are independent of the mag-
nitude of H0. That assumption, however, may not always be
appropriate in highly permeable intervals, so a more
involved approach may be necessary. McElwee and Zenner
(1998) and McElwee (2001) describe a nonlinear variant of
the high K extension of the Hvorslev model that can be
used for the analysis of slug tests that demonstrate a depen-
dence on H0. The implementation of that model in the
SPBatch program of Bohling (1998) was used here to ana-
lyze tests from the levels at which a reproducible depen-
dence on H0was observed.
Regardless of which model was employed to analyze
the test data, the same parameters were used for the OD and
length of the screen (0.027 and 0.305 m, respectively), and
the ID of the rods (0.038 m). Isotropy with respect to K was
assumed in all cases for lack of better information. The
anisotropy ratio cannot be estimated from a slug test (But-
ler 1998), but if that ratio is known through other means, it
can be assumed as a fixed quantity for the analysis.
Results of Field Assessment
Nine EC logs were obtained as part of the hydrostrati-
graphic profiles performed in this work. Figure 4 is an
example EC log from this group that displays the major fea-
tures seen in all the logs. As shown in Figure 4, the shallow
subsurface at GEMS consists of an upper silt and clay unit,
a silt unit with occasional thin sand lenses (at 7.4 and 8.4 m
in Figure 4), a clay unit (vertical head difference of > 0.9 m
occurs across this unit), and a thick sand and gravel inter-
val overlying bedrock. Although the elevation of the
boundary between the clay and the underlying sand and
gravel varies somewhat across the site, the boundary is
always an abrupt transition (Figure 4). The EC spikes in the
sand and gravel between 18 and 19 m, and ~20.5 m are dis-
continuous lenses of a higher clay content that are observed
at these depths in numerous EC logs from GEMS.
A high-resolution view of both the vertical and lateral
variations in EC at GEMS can be obtained using the nine
EC logs and the Tecplot data visualization software (Amtec
Engineering 2001). Figures 5a and 5b display the resulting
three-dimensional EC images viewed from the southeast
and northwest of the surface grid, respectively. These fig-
ures clearly show the dramatic contrast between the upper
zone of high EC and the sand and gravel interval of low EC.
These two images also indicate that the sand lenses in the
silt zone between 7 and 10 m in depth may be intercon-
nected across the area. These images also provide some sup-
port for the previous interpretation of discontinuous lenses
of higher clay content in the sand and gravel interval.
S.M. Sellwood et al. GROUNDWATER43, no. 1: 19–29 24
Figure 4. EC vs. depth profile for HP7 and the generalized
GEMS stratigraphy (inverted triangle marks the approxi-
mate position of water table).
(Authors’ note: More detailed figures in color are provided
in Sellwood et al. .)
Figures 5a and 5b demonstrate the potential of high-res-
olution EC logging for identifying thin units and assessing
their lateral continuity. Such features can play an important
role in ground water flow and transport at a site, but will be
difficult to detect using conventional approaches. Note that
the inverse-distance interpolation scheme, coupled with the
distance between profile locations, created artifacts in these
and later figures. The issue of the most appropriate interpo-
lation scheme for such data is the subject of ongoing
research and beyond the scope of this paper.
Slug Test Results
Rising-head slug tests were performed at a total of 70
levels over the nine profiles. Six to 10 tests were performed
at each level, all of which were initially analyzed with the
high K Hvorslev model (Butler 1998; Butler et al. 2003).
Results from tests at the same level initiated with different
H0were compared to determine if there was a reproducible
dependence on H0. At eight of the 70 levels, a significant
dependence on H0was observed (K estimates varied by
more than 10% between tests initiated with H0that differed
by a factor of two). These tests were reanalyzed with the
nonlinear variant of the high K Hvorslev model (McElwee
and Zenner 1998; McElwee 2001). At 21 levels, falling-
head slug tests were performed in addition to the rising-
head tests. Only tests at one level (level 6 in HP9)
demonstrated a significant directional dependence, which
was attributed to insufficient development. The lack of
directional dependence at the other 20 levels indicates that
the development procedures used in this work were effec-
tive. Table 1 summarizes the results of the program of slug
tests. Further details can be found in Sellwood (2001).
The viability of the slug test estimates was assessed by
comparison with K estimates obtained using the hydraulic
profiling method of McCall et al. (2002) and with the aver-
age K obtained from pumping tests. Since the McCall et al.
(2002) method has previously been shown to produce reli-
able K estimates through comparison with multilevel slug
tests and dipole-flow tests performed in conventional mon-
itoring wells (McCall et al. 2002), K estimates obtained
with the McCall et al. (2002) method were used as stan-
dards for comparison with the hydrostratigraphic profiling
approach. Hydrostratigraphic profiles HP1 and HP8 were
completed in the vicinity of the DP808 profile (separation
of 2.17 and 4.48 m, respectively) (Figure 1) described in
McCall et al. (2002). Test intervals for HP1 and HP8 were
selected to be within a vertical distance of 0.03 m of the
intervals tested at DP808. Figure 6 is a plot of K vs. depth
comparing the profiles of HP1, HP8, and DP808. Because
the distance between the profiles varied from 2 to 6 m,
some differences were expected due to lateral heterogene-
ity. Given that possibility, the agreement between the K
profiles shown in Figure 6 is very good.
A number of pumping tests have been performed in
this portion of the well network at GEMS (Butler et al.
2002). The average K for the sand and gravel interval as
determined by these pumping tests is 116 m/d (Figure 6).
The unweighted vertical averages of the K estimates from
profiles HP1 and HP8 are 112 and 119 m/d, respectively.
Given the good agreement with the profile obtained using
the McCall et al. (2002) method and the average K from the
pumping tests, the K estimates from the hydrostratigraphic
profiling procedure appear to be reasonable representations
of the K of the formation in the vicinity of the test intervals.
A view of both the lateral and vertical variations in K
within the sand and gravel interval can be obtained using
25S.M. Sellwood et al. GROUNDWATER43, no. 1: 19–29
Figure 5. Three-dimensional images of EC at GEMS. (a) View from the southeast of the profiling grid. (b) View from the north-
west of the profiling grid (approximate surface locations of profiles are marked; origin of grid coordinate system at HP8; inter-
polation performed with inverse-distance weighting [exponent = –3.5] algorithm in Tecplot [Amtec Engineering 2001]; shapes
of many of the smaller-scale features are artifacts of the grid spacing and the interpolation algorithm).
the nine K profiles and the data visualization software. Fig-
ures 7a and 7b display the resulting three-dimensional K
images viewed from the southeast and northwest of the sur-
face grid, respectively. These figures reveal that there is a
general trend from lower K material at the top of the inter-
val to higher K material at the bottom. However, lenses of
differing K are superimposed on this trend throughout the
interval. The order of magnitude variation in K shown here
could have a significant impact on contaminant transport
and the design of remediation schemes. (More detailed fig-
ures in color are provided in Sellwood et al. .)
An underlying principle of the hydrostratigraphic pro-
filing method is that EC variations can be used to distin-
guish between low K clay-rich materials and high K
materials with little to no clay. When the focus of an inves-
tigation is on differentiating between major lithologic units,
such as shown in Figure 4, there is a strong inverse corre-
lation between EC and K, as the large contrast in clay con-
tent is the primary control on the EC and K variations.
However, when the focus is on K variations within an
aquifer, the relationship between EC and K may be less
straightforward. Figure 8a depicts the three K profiles of
Figure 6 along with the EC log at HP8. In this case, there
appears to be little correlation between K and EC. Thus, the
variations in K in the vicinity of HP8 do not appear to be
caused by variations in clay content. In contrast, a much
stronger correlation between EC and K is observed when
an EC log immediately to the west of well Gems4S is used
(Figure 8b). In that case, low K values are associated with
peaks in the EC log at 16.3 and 19.1 m, indicating that
increases in clay content may be an important control on K
in those intervals at some locations. However, in other
intervals (e.g., below 19.5 m), increases in EC coincide
with increases in K, such as might be expected if porosity
variations are the primary control on EC. Since changes in
EC within a saturated sand and gravel interval can be
caused by variations in clay content, porosity, and fluid
chemistry, which can produce opposing or no effect on K,
it is difficult to relate those EC changes to variations in K
(Schulmeister et al. 2003).
Discussion and Conclusions
Obtaining estimates of K at a sufficient spatial density
for reliable transport predictions and effective remediation
S.M. Sellwood et al. GROUNDWATER43, no. 1: 19–29 26
Average K Values (m/d) for Each Test Interval at GEMS
12.6 to 12.9
13.8 to 14.1
15.1 to 15.4
16.3 to 16.6
17.5 to 17.8
18.7 to 19.0
19.9 to 20.2
21.2 to 21.5
21.7 to 22.0
22.1 to 22.4
aDatum is top of casing at well Gems4S
bLevel at which nonlinear variant of high K extension of Hvorslev model was used for analysis
In most cases, reported values are arithmetic averages of K values from individual tests; when a nonlinear variant of high K Hvorslev model was used, a single K value
was obtained for each level; K values missing from levels 1 through 8 are due to unsuccessful development of relatively low K intervals, slug tests were not attempted at
levels 9 and 10 except for HP4 level 9 and HP8 level 10; Sellwood (2001) provides further details.
Figure 6. Plot of K vs. depth for profiles HP1, HP8, and
DP808, and the average K determined from pumping tests at
wells Gems4N and Gems4S (datum is top of casing at well
Gems4S; vertical dimension of plotted symbols is equal to the
screen length; lateral separation between HP1 and HP8 is
designs is typically a time- and resource-intensive task. As a
result, there is rarely sufficient information at a site to assess
the influence of spatial variations in subsurface properties on
contaminant transport. In order to address this problem, a DP
method has been developed for the characterization of spatial
variations in both EC and K in unconsolidated formations.
This approach, which couples an existing method for per-
forming slug tests in DP equipment with high-resolution EC
logging, is more efficient than previous methods. The pur-
pose of this paper was to describe the development and ini-
tial field assessment of this approach.
The hydrostratigraphic profiling procedure described
here allows variations in both EC and K to be determined
at a resolution and efficiency that has rarely been possible.
A high-resolution log of EC is obtained as a pair of nested
DP rod strings is advanced into an unconsolidated forma-
tion. After logging through the intervals of interest, the
inner rod string is removed and slug tests are performed as
the remaining rod string is retracted. The intervals for slug
testing can be selected on the basis of the EC logs. An
assessment of the approach in a controlled field setting
demonstrated its potential for subsurface characterization.
27S.M. Sellwood et al. GROUNDWATER43, no. 1: 19–29
Figure 7. Three-dimensional images of K within the sand
and gravel interval at GEMS. (a) View from the southeast of
the profiling grid. (b) View from the northwest of the pro-
filing grid (approximate surface locations of profiles are
marked; origin of grid coordinate system at HP8; interpola-
tion performed with inverse-distance weighting [exponent =
–3.5] algorithm in Tecplot [Amtec Engineering 2001]; shapes
of many of the smaller-scale features are artifacts of the grid
spacing and the interpolation algorithm).
Figure 8. Comparison of profiles of K with profiles of EC. (a)
EC profile at HP8. (b) EC profile to the immediate west of
Gems4S (K profiles from Figure 6; datum is top of casing at
well Gems4S; EC log from HP1 not plotted due to mechani-
cal failure during logging in the sand and gravel interval).
The two major limitations of this approach are (1) it
can only provide stratigraphic information when variations
in EC are primarily a function of variations in clay content,
and (2) the DP procedure produces compaction in the por-
tions of the formation in the immediate vicinity of the probe
hole, which can possibly result in EC and K values that are
not representative of the formation outside of this zone of
compaction. These limitations are considered further in the
An underlying assumption of the approach described
here is that variations in EC are primarily a function of vari-
ations in clay content. However, variations in water chem-
istry (Schulmeister et al. 2003) and, to a lesser extent,
porosity can also have a major impact on EC. Thus, as
emphasized earlier, the assumption that EC is primarily
controlled by variations in clay content must be checked at
each site through the collection of core and water samples.
Water samples can be readily obtained in the retraction
phase of the hydrostratigraphic profiling procedure.
Previous work has assessed the impact of the zone of
compaction created during DP activities. Schulmeister et
al. (2003) have compared DP EC logs with conventional
electrical logs (focussed induction) obtained at nearby
monitoring wells. The close correspondence between the
logs indicates that the zone of compaction has little impact
on EC values. Similarly, a number of recent studies (Butler
et al. 2002; Butler 2002; McCall et al. 2002) have com-
pared K estimates obtained from slug tests in DP installa-
tions with estimates obtained from hydraulic tests
performed in standard monitoring wells. Good agreement
between estimates from DP installations and those from
wells was obtained when sufficient attention was given to
The time required to complete a series of hydrostrati-
graphic profiles at a site is dependent on many factors. The
most significant of these are the depth to which the EC logs
are run, the number of intervals at which slug tests are per-
formed, and the K of the test intervals. In this study, it took
~2 d to complete a single profile (an EC log to 22 m and slug
tests at eight levels within the sand and gravel). If fewer test
levels are acceptable, the time for a profile can be reduced
significantly. For example, an EC log and slug tests at three
to four levels could be readily completed at GEMS in 1 d.
The nested-rod EC logging procedure used here takes con-
siderably more time than logging with a single rod string.
Expendable dipole probes are available at a relatively low
cost, so a single-rod modification of the approach described
here could be developed to further decrease the time of the
profiling procedure. In that case, a single rod string would
be advanced through the intervals of interest, after which the
dipole probe would be knocked out of the bottom of the
string with the extension rods. Considerably less water
would be added since there would be no need to continue to
add water once the probe has been knocked out. Using a sin-
gle-rod system, a profile consisting of an EC log to 20+ m
and slug tests of five to six levels in a sand and gravel inter-
val should be possible within 1 d.
This research was supported in part by the Kansas
Water Resources Institute under grant HQ96GR02671
Modif. 008 (subaward S01044; JJB PI) and by the Hydro-
logic Sciences Program of the National Science Foundation
under grant 9903103 (JJB PI). Stephen Sellwood and Stef-
fen Birk were the 2001 participants in the Applied Geohy-
drology Summer Research Assistantship program of the
Kansas Geological Survey. This work has benefited from
reviews provided by Solomon Isiorho, David Hyndman,
Wes McCall, and one anonymous reviewer.
Editor’s Note: The use of brand names in peer-reviewed
papers is for identification purposes only and does not con-
stitute endorsement by the authors, their employers, or the
National Ground Water Association.
Amtec Engineering. 2001. Tecplot User’s Manual, Version 9.0.
Bellevue, Washington: Amtech Engineering.
Bohling, G.C. 1998. Using the nonlinear Hvorslev function in
Spbatch. Unpublished report.
Bohling, G.C. 1999. Evaluation of an induced gradient tracer test.
Ph.D. dissertation, Department of Geology, University of
Butler, J.J., Jr. 1998. The Design, Performance, and Analysis of
Slug Tests. Boca Raton, Florida: Lewis Publishers.
Butler, J.J., Jr. 2002. A simple correction for slug tests in small-
diameter wells. Ground Water 40, no. 3: 303–307.
Butler, J.J., Jr., and J.M. Bahr. 1988. Summary of field and labo-
ratory testing methods for flow and nonreactive transport
parameters. In Consequences of Spatial Variability in
Aquifer Properties and Data Limitations for Groundwater
Modelling Practice, ed. A. Peck, S. Gorelick, G. de Marsily,
S. Foster, and V. Kovalesky, 56–77. Wallingford, United
Kingdom: IAHS Press.
Butler, J.J., Jr., and J.M. Healey. 1998. Relationship between
pumping test and slug test parameters: Scale effect or arti-
fact? Ground Water 36, no. 2: 305–313.
Butler, J.J., Jr., and W.Z. Liu. 1993. Pumping tests in nonuniform
aquifers: The radially asymmetric case. Water Resources
Research 29, no. 2: 259–269.
Butler, J.J., Jr., E.J. Garnett, and J.M. Healey. 2003. Analysis of
slug tests in formations of high hydraulic conductivity.
Ground Water 41, no. 5: 620–630.
Butler, J.J., Jr., C.D. McElwee, and W.Z. Liu. 1996. Improving
the reliability of parameter estimates obtained from slug
tests. Ground Water 34, no. 3: 480–490.
Butler, J.J., Jr., C.D. McElwee, and G.C. Bohling. 1999. Pump-
ing tests in networks of multilevel sampling wells: Motiva-
tion and methodology. Water Resources Research 35, no.
Butler, J.J., Jr., J.M. Healey, V.A. Zlotnik, and B.R. Zurbuchen.
1998. The dipole flow test for site characterization: Some prac-
tical considerations. Eos Transactions, American Geophysical
Union 79, no. 17: S153 (full report—Kansas Geological Sur-
vey Open-File Rept. 98–20—www.kgs.ku.edu/Hydro/
Butler, J.J., Jr., J.M. Healey, G.W. McCall, E.J. Garnett, and S.P.
Loheide II. 2002. Hydraulic tests with direct-push equip-
ment. Ground Water 40, no. 1: 25–36.
Butler, J.J., Jr., A.A. Lanier, J.M. Healey, S.M. Sellwood, G.W.
McCall, and E.J. Garnett. 2000. Direct-push hydraulic pro-
filing in an unconsolidated alluvial aquifer. Kansas Geolog-
ical Survey Open-File Rept. 2000–62, www.kgs.ku.edu/
S.M. Sellwood et al. GROUNDWATER43, no. 1: 19–2928
Cho, J.S., J.T. Wilson, and F.P. Beck Jr. 2000. Measuring verti-
cal profiles of hydraulic conductivity with in-situ direct
push methods. Journal of Environmental Engineering 126,
no. 8: 775–777.
Christy, C.D., T.M. Christy, and V. Wittig. 1994. A percussion
probing tool for the direct sensing of soil conductivity. In
Proceedings of the Eighth Annual National Outdoor Action
Conference and Exposition on Aquifer Remediation,
Ground Water Monitoring, and Geophysical Methods, May
23–25, Minneapolis, Minnesota, 381–394. Dublin, Ohio:
National Ground Water Association.
Farrar, J.A. 1996. Research and standardization needs for direct
push technology applied to environmental site characteriza-
tion. In Sampling Environmental Media, ASTM Special
Technical Publication 1282, ed. J.H. Morgan, 93–107.
Philadelphia, Pennsylvania: American Society for Testing
Geoprobe Systems. 1998. Tools and Equipment Catalog. Salina,
Hackett, G. 1987. Drilling and constructing monitoring wells
with hollow-stem augers, Part 1. Drilling considerations.
Ground Water Monitoring Review 7, no. 4: 51–62.
Healey, J.M., and S.M. Sellwood. 2004. The KGS direct-push
hydrostratigraphic profiling tool. Kansas Geological Survey
Open-File Report 2004–44.
Hinsby, K., P.L. Bjerg, L.J. Andersen, B. Skov, and E.V.
Clausen. 1992. A mini slug test method for determination of
a local hydraulic conductivity of an unconfined sandy
aquifer. Journal of Hydrology 136, 87–106.
HydroSOLVE Inc. 2001. AQTESOLV for Windows User’s
Guide. Reston, Virginia: HydroSOLVE.
Jacobs, J.A., M. Kram, and S. Lieberman. 2000. Direct push
technology sampling methods. In Standard Handbook of
Environmental Science, Health, and Technology, ed. J.
Lehr, 11.151–11.163. New York: McGraw-Hill.
Kram, M., S. Lieberman, and J.A. Jacobs. 2000. Direct sensing
of soils and groundwater. In Standard Handbook of Envi-
ronmental Science, Health, and Technology, ed. Lehr, J.,
11.124–11.150. New York: McGraw-Hill.
Lunne, T., P.K. Robertson, and J.J.M. Powell. 1997. Cone Pene-
tration Testing in Geotechnical Practice. London: Blackie
Academic and Professional.
McCall, W., J.J. Butler Jr., J.M. Healey, A.A. Lanier, S.M. Sell-
wood, and E.J. Garnett. 2002. A dual-tube direct-push
method for vertical profiling of hydraulic conductivity in
unconsolidated formations. Environmental & Engineering
Geoscience 8, no. 2: 75–84.
McElwee, C.D. 2001. Application of a nonlinear slug test model.
Ground Water 39, no. 5: 737–744.
McElwee, C.D., and M.A. Zenner. 1998. A nonlinear model for
analysis of slug-test data. Water Resources Research 34, no.
Meier, P.M., J. Carrera, and X. Sanchez-Vila. 1998. An evalua-
tion of Jacob’s method for the interpretation of pumping
tests in heterogeneous formations. Water Resources
Research 34, no. 5: 1011–1025.
National Research Council. 1994. Alternatives for Ground Water
Cleanup. Washington, D.C.: National Academies Press.
Schulmeister, M.K., J.J. Butler Jr., J.M. Healey, L. Zheng, D.A.
Wysocki, and G.W. McCall. 2003. Direct-push electrical
conductivity logging for high-resolution hydrostratigraphic
characterization. Ground Water Monitoring & Remediation
23, no. 3: 52–62.
Sellwood, S.M. 2001. A direct-push method of hydrostrati-
graphic characterization. M.S. thesis, Department of Geol-
ogy, University of Kansas, Lawrence.
Sellwood, S.M., J.M. Healey, S. Birk, and J.J. Butler Jr. 2004.
Supplementary color figures for “Direct-push hydrostrati-
graphic profiling: Coupling electrical logging and slug
tests.” Kansas Geological Survey Open-File Rept. 2004–12;
Yeh, T.-C.J., J. Mas-Pla, T.M. Williams, and J.F. McCarthy.
1995. Observation and three-dimensional simulation of
chloride plumes in a sandy aquifer under forced-gradient
conditions. Water Resources Research 31, no. 9: 2141–
Zlotnik, V.A., and B.R. Zurbuchen. 1998. Dipole probe: Design
and field applications of a single-borehole device for mea-
surements of vertical variations in hydraulic conductivity.
Ground Water 36, no. 6: 884–893.
Zurbuchen, B.R., V.A. Zlotnik, and J.J. Butler Jr. 2002. Dynamic
interpretation of slug tests in highly permeable aquifers.
Water Resources Research
38, no. 3: 10.1029/
29S.M. Sellwood et al. GROUNDWATER43, no. 1: 19–29