You are on page 1of 20

Journal of Hydrology 281 (2003) 322

www.elsevier.com/locate/jhydrol

Use of time subsidence data during pumping to characterize


specific storage and hydraulic conductivity of semi-confining units
T.J. Burbey*
Department of Geological Sciences, Virginia Tech, 3053 Derring Hall, Blacksburg, VA 24061, USA
Received 17 April 2002; accepted 1 May 2003

Abstract
A new graphical technique is developed that takes advantage of time subsidence data collected from either traditional
extensometer installations or from newer technologies such as fixed-station global positioning systems or interferometric
synthetic aperture radar imagery, to accurately estimate storage properties of the aquifer and vertical hydraulic conductivity of
semi-confining units. Semi-log plots of time compaction data are highly diagnostic with the straight-line portion of the plot
reflecting the specific storage of the semi-confining unit. Calculation of compaction during one-log cycle of time from these
plots can be used in a simple analytical expression based on the Cooper Jacob technique to accurately calculate specific storage
of the semi-confining units. In addition, these semi-log plots can be used to identify when the pressure transient has migrated
through the confining layer into the unpumped aquifer, precluding the need for additional piezometers within the unpumped
aquifer or within the semi-confining units as is necessary in the Neuman and Witherspoon method. Numerical simulations are
used to evaluate the accuracy of the new technique. The technique was applied to time drawdown and time compaction data
collected near Franklin Virginia, within the Potomac aquifers of the Coastal Plain, and shows that the method can be easily
applied to estimate the inelastic skeletal specific storage of this aquifer system.
q 2003 Elsevier B.V. All rights reserved.
Keywords: Subsidence; Aquifer compaction; Well hydraulics

1. Introduction
Aquifer tests are the most fundamental means by
which groundwater practitioners can quantitatively
estimate the transmissivity and storage properties of
the aquifer and semi-confining unit. The pioneering
work of Meinzer (1928), Theis (1935), Jacob (1940),
and later Hantush (1956, 1959, 1960, 1967) for the
analysis of confined and leaky aquifers still remains
* Tel.: 1-5402316696; fax: 1-5402313386.
E-mail address: tjburbey@vt.edu (T.J. Burbey).

the benchmark by which other techniques and


modifications have been based. Extensions and
modifications to these earlier works are too numerous
to list here. However, several key works have been
pivotal in describing the nature of drawdown during
pumping and are still widely used today to characterize aquifer properties under a wide variety of settings
(e.g. Cooper and Jacob, 1946; Neuman and Witherspoon, 1968; 1969a c, 1972; Walton, 1970;
Moench, 1985; Sridharan et al., 1987). Many of
these techniques for estimating the storage and
hydraulic conductivity properties of the aquifer

0022-1694/03/$ - see front matter q 2003 Elsevier B.V. All rights reserved.
doi:10.1016/S0022-1694(03)00197-5

T.J. Burbey / Journal of Hydrology 281 (2003) 322

system have been incorporated into software


packages, developed for rapid analysis of aquifer
test data.
Each of these techniques makes a number of
assumptions in order to obtain an analytic solution or
an approximation to the exact solution that allows the
method to be used to estimate the transmissivity and
storativity of the aquifer. In addition, several of the
methods allow for estimation of the confining unit
diffusivity (e.g. Neuman and Witherspoon, 1972). The
field data collected during the aquifer test and
required to estimate these parameters generally
involves only the collection of hydraulic head and
time data for various hydrologic units. Semi-confining
units are known to contribute large quantities of
water to the pumped aquifer (Poland and Davis, 1969)
as a result of the large compressibilities of clay-rich
units. However, in order to evaluate the specific
storage and vertical hydraulic conductivity of the
semi-confining unit, head data (or time-lag data)
within the clay unit are required along with laboratory
data on the compressibility of the clay (Neuman and
Witherspoon, 1972). Because only head data are
collected during aquifer testing, further refinement
and characterization of the semi-confining units
cannot be made without piezometers located in each
hydrogeologic unit.
The implementation of extensometer data (Pope,
2002; Harmon, 2002) and the recent advent of radar
techniques such as interferometric synthetic aperture
radar (InSAR) and high precision global positioning
systems (GPS) in hydrogeologic applications (Zebker
et al., 1994; Ikehara, 1994; Thom et al., 1995; Zebker
et al., 1997; Fielding et al., 1998; Galloway et al.,
1998; Amelung et al., 1999; Davies and Blewitt,
2000; Hoffmann et al., 2001) have allowed for the realtime determination of subsidence or total compaction
occurring as the result of pumping. The application of
these techniques during an aquifer test of moderate
length of 10 days or longer, could provide valuable
data regarding the compressible units in the aquifer
system that would allow further parameterization of
the fine-grained units that cannot be made with
hydrograph information alone.
This paper presents a straightforward analytical
and graphical methodology based on the Cooper
Jacob method (Cooper and Jacob, 1946) for estimating the storage coefficient of confined aquifers and

the specific storage and vertical hydraulic conductivity of the semi-confining units in leaky aquifer
systems under transient pumping conditions. Numerical simulations are used to evaluate the validity and
accuracy of the methodology described below. The
technique has advantages over traditional straight-line
methods in that it is not necessary to know the exact
time of the start of pumping.

2. Methodology and results


2.1. Confined aquifers
Typical aquifer tests employ the use of timedependent drawdown data for making estimates of
aquifer storativity and transmissivity. Semi-log
straight-line methods such as the Cooper Jacob
time drawdown method are commonly used because
they allow one to readily estimate these parameters
provided that the aquifer test is sufficiently long and
the assumptions inherent in the technique are not
grossly violated. Furthermore, straight-line methods
provide visual indicators of the presence of boundaries or significant leakage from adjacent semiconfining units because the field data depart from
the perfectly confined straight-line semi-log plot.
Estimation of aquifer transmissivities tends to be quite
accurate using this approach even for leaky aquifers as
long as r; the radial distance from the pumping well to
the observation well, is sufficiently small. The
accuracy of this technique for estimating transmissivity is attributed to the fact that transmissivity is highly
sensitive to drawdown during pumping. The Cooper
Jacob equation can be written as
s



2:303Q
2:25Tt
log
4pT
r2 S

where s is the drawdown, Q is the constant pumping


rate, T is the aquifer transmissivity, and S is the
aquifer storativity. Fig. 1 shows the semi-log plot of
the numerically produced time drawdown data at an
observation well 50 m from a constantly pumped well
in a perfectly confined 100 m thick aquifer (a Theistype aquifer) with a hydraulic conductivity of 5 m/d.
Plots are shown for storage coefficients spanning three
orders of magnitude. The slope of each plot in Fig. 1

T.J. Burbey / Journal of Hydrology 281 (2003) 322

Fig. 1. Simulated drawdown as a function of time during pumping


for three different values of storage derived from a perfectly
confined 100 m-thick homogeneous and isotropic aquifer.

can be quantified from Eq. (1) as


Ds
2:303Q

logt2 =t1
4pT

large errors, particularly in settings in which the


storage coefficient tends to be large. It is not
uncommon for estimated storage errors to be
upwards of one order of magnitude using this
approach. In addition, this approach requires one to
know the exact time since the start of pumping.
Hydrogeologists often correlate the insensitivity of
the storage with the lack of importance for a good
estimate. This can be troublesome when trying to
preserve groundwater supplies through proper
management.
A more diagnostic method for accurately estimating storage is to plot total compaction (subsidence
expressed at the land surface) with time during the
aquifer test. This approach is preferred because
compaction is directly related to the compressibility
and is therefore directly related to the storage of the
aquifer. Although this method does not account for
the release of water resulting from the expansion of
water, in compressible systems where this technique
would be applied, the matrix compressibility is
typically several orders of magnitude greater than
water compressibility. Therefore, the contribution
from water expansion is considered negligible. Fig. 2
shows the semi-log straight-line plots of the time
compaction record for the same confined aquifer

where Ds is the change in drawdown over the time


interval t2 2 t1 : The slopes are constant because the
transmissivity remains the same for all three storage
values simulated.
The storage coefficient of the aquifer is estimated from these data plots by extending the
straight-line portion of the plot to the zero drawdown axis. The estimated time, t0 ; at which the
drawdown is zero is used in Eq. (1) to yield


2:25Tt0
0 log
3
r2 S
or, by solving for S
S

2:25Tt0
r2

This is the graphical approach for estimating


storativity described by Cooper and Jacob (1946).
Unfortunately, storage properties tend to be only
moderately sensitive to drawdown in many settings
(Neuman, 1979; Anderson and Woessner, 1992).
Furthermore, estimating t0 can potentially lead to

Fig. 2. Simulated subsidence as a function of time during pumping


for three different values of storage derived from a perfectly
confined 100 m-thick homogeneous and isotropic aquifer.

T.J. Burbey / Journal of Hydrology 281 (2003) 322

using the same three storage coefficients as illustrated


in Fig. 1. Note that the slope is vastly different for
each of the three values of storage coefficient.
Calculation of storage can be obtained from this plot
using the slope in much the same way the transmissivity was estimated from Eq. (2) and Fig. 1. This
yields a potentially more accurate calculation of the
storage coefficient than the traditional use of time
drawdown plots. Furthermore, this method no longer
depends on the estimation of time. That is, the time of
the start of pumping is not required. This is particularly
beneficial in settings where compaction data has not
been measured until sometime after the start of
pumping. If accurate extensometer, GPS or InSAR
values can be obtained during some portion of the
aquifer test or period of pumping, then this technique
should be favored over the traditional techniques
where only time drawdown plots are used.
For one-dimensional consolidation the storativity
of the subsiding system can be expressed simply as
(Burbey, 2001)
S

Db
Dh

where Db is the change in thickness or measured


compaction measured within the aquifer system
during Dh; the change in head or total drawdown
during pumping. The storage expressed in Eq. (5)
assumes that the contribution of water expansion is
small. In Eq. (2) Dh Ds; that is, the change in head
is equivalent to the change in drawdown. If the change
in drawdown and change in compaction are measured
over one log cycle of time, the denominator in the left
side of Eq. (2) becomes equal to 1. Substituting Eq. (5)
into Eq. (2) and solving for S gives
S

Db4pT
5:46DbT

;
2:303Q
Q

Fig. 3. Simulated timesubsidence plot for a perfectly confined


aquifer with calculated compaction evaluated over one log-cycle of
time. This calculated compaction is used in Eq. (6) to determine the
storage coefficient of the compacting unit.

where Db is the straight-line measured compaction


occurring over one log cycle of time. In this
expression, the more diagnostic slope is used on the
time compaction plot as opposed to the time
intercept at s 0 on the time drawdown plot. For
the hypothetical aquifer described above (here it is
assumed that the aquifer represents the compressible
unit of interest), time subsidence data were produced
using the IBS3 package (Leake and Prudic, 1991) with
Modflow (McDonald and Harbaugh, 1988). These

simulated data are plotted in Fig. 3 along with the


calculated Db expressed over one log cycle of time.
Transmissivity was estimated using the classical
Theis method on the time drawdown data, with the
drawdown values calculated using Modflow. These
data are then incorporated into Eq. (6) to yield an
estimated S of 1.0 1024, which is identical to the
numerical value used in the simulation. Additional
analyses using numerical models resulted in predicted
storage coefficients that were always within 10
percent of the actual value.
One might argue that this technique is of limited
value because subsidence occurs from confining units
and not aquifers. However, in many settings in which
subsidence occurs, the aquifer system is used to
represent the aquifer containing compressible lenses
of clays and silty-clays. Hence, this technique would
produce an average value for the aquifer system. The
case study used for the application of this technique
represents such a system.
2.2. Leaky aquifers
The approach of incorporating time compaction
data can be extended beyond perfectly confined

T.J. Burbey / Journal of Hydrology 281 (2003) 322

aquifers to the more commonly encountered leaky


aquifer systems in which storage within the confining
unit is considered important. In these settings the
compressibilities of the confining unit are typically
one to several orders of magnitude greater than that of
the aquifer unit. Hence, the compaction data largely
reflect the compressibility of the semi-confining unit,
whereas the time drawdown data largely reflect the
aquifer properties where the head data are being
measured. Field data that depart from the Theis-type
curve for the aquifer being pumped are related to (1)
the distance from the observation well where the data
are being acquired to pumping well, (2) the thickness
and vertical hydraulic conductivity of the bounding
semi-confining units, and (3) the head in the
unpumped aquifer if one exists. Because we assume
here full penetration of the pumping well with a small
annulus and homogeneous and isotropic aquifer
conditions, deviations in the time drawdown
response due to a departure from these conditions
are not considered.
Two hydrogeologic conditions will be evaluated:
(1) Two aquifers separated by a single confining unit
where no drawdown occurs in the unpumped
aquifer that overlies the confining unit (Fig. 4).
This condition is valid where the overlying
aquifer is unconfined with a sufficiently large
storage such that downward leakage induced by
pumping of the lower aquifer would not
appreciably affect the water level of this
unconfined aquifer.

Fig. 5. Conceptual model of the multiple-confining unit aquifer


system used to evaluate an average specific storage and hydraulic
conductivity of the semi-confining units.

(2) Multiple aquifer and confining-unit system


(Fig. 5) in which one aquifer is pumped and
heads in the overlying aquifer can be affected by
pumping from within the pumped aquifer.
2.2.1. Condition 1: one semi-confining unit
and no drawdown in unpumped aquifer
The method for evaluating the first hydrogeologic
condition was first described analytically by Hantush
(1960). The solution for drawdown in the pumped
aquifer of this well known governing equation that
mathematically describes the leaky aquifer condition
presented in Fig. 4 has been estimated for both early
and late times and is expressed as
!
Q 1 e2y
bu1=2
sr; t
erfc
dy
7
4pT u y
yy 2 u1=2
where b is known as the Hantush leakage factor for
leaky aquifers, and u is the well function expressed as
u

Fig. 4. Conceptual model of the one-confining-unit aquifer system


used to evaluate specific storage and hydraulic conductivity of the
semi-confining unit.

r2 S
;
4Tt

and erfc is the complementary error function.


According to Hantush (1960) the validity of Eq. (7)

T.J. Burbey / Journal of Hydrology 281 (2003) 322

is limited to the range of time given by the


expression
t#

0:1b0 S0
K0

where the primes on b; K and S indicate the


thickness, vertical hydraulic conductivity and storativity of the semi-confining unit, respectively.
Neuman and Witherspoon (1969c) determined that
Eq. (8) is far too conservative and that its validity
can be expanded to a much larger range of time
than suggested by Eq. (8). They concluded that
Eq. (7) is valid while the time drawdown curve is
steep. That is, the solution expression is valid if it is
not affected by the constant head condition imposed
by the overlying aquifer. We will show that this
condition can be evaluated graphically using a
time subsidence plot obtained from the aquifer
test. In Eq. (7) the dimensionless leakage factor is
evaluated from type curves and is expressed
mathematically as

 0  0 1=2
r
K
S
:
4 Kbb0 S

To evaluate the usefulness of the methodology


described in the previous section on confined
aquifers for the leaky-aquifer condition, a simulation
was performed in which 10 model layers were used
to simulate the semi-confining unit and one model
layer was used for each the unconfined and confined
aquifers. The 10 layers were implemented so that a
more accurate and realistic head distribution through
time could be simulated through the semi-confining
unit as the underlying aquifer was pumped (Fig. 4).
Modflow (McDonald and Harbaugh, 1988) was used
to simulate hydraulic heads and the IBS3 package
(Leake and Prudic, 1991) was used to simulate
compaction and total subsidence. Table 1 and Fig. 4
provide the parameters and aquifer system geometry
used in the simulation. The lower confined aquifer
was pumped at a rate of 2000 m3/d for 30 days.
Three assumptions are made regarding the hydrogeologic conditions that lead to the simulation
results discussed below:
(1) Horizontal strain is negligible. Burbey (2001)
indicates that horizontal strain may not be

Table 1
Simulated and estimated aquifer and confining unit parameters used
for the first hydrogeologic condition with one confining unit and no
drawdown occurs in the unpumped aquifer
Aquifer and confining
unit parameters

Simulation
value

Estimated value
from new method

T (m2/d)
S (dimensionless)
K 0v (m/d)
S0s (1/m)

125.0
5.0 1025
0.005
8.0 1025

132.4
5.2 1025
0.003
5.6 1025

negligible in many settings in which aquifers


are pumped from unconsolidated sediments.
However, we are making evaluations near the
pumped well where horizontal deformation is
limited as a result of the zero-radial strain
condition imposed by the well casing. In this
analysis we assume that all strain is vertical.
The IBS3 package does not account for
horizontal strain.
(2) The hydraulic conductivity of the confining
unit is at least two orders of magnitude less
than that of the aquifers. This essentially leads
to the condition of vertical flow through the
confining unit. Neuman and Witherspoon
(1969a c) indicate through finite-element
modeling that when this condition occurs the
error introduced by this assumption is less
than 5%.
(3) The compressibility of the confining unit is
more than one order of magnitude greater than
the compressibility of the aquifers. This is
generally true for confining units rich in clay
and undergoing nonrecoverable or virgin
compression. If preconsolidation heads have
been lowered significantly within the confining
unit prior to the aquifer test such that
permanent deformation of the clays has
already occurred, then the deformation tends
to be elastic and the active compressibilities of
the confining unit are closer to that of the
aquifer. In this scenario the approach would be
to obtain an aquifer system (combined aquifer
and semi-confining unit average value) specific
storage through use of the previously described
method for confined aquifers.

T.J. Burbey / Journal of Hydrology 281 (2003) 322

Fig. 6a shows the numerically derived time


drawdown plot for this simulation at an observation
well 8.6 m from the pumping well. The results
indicate that after 30 days of pumping the pumped
aquifer has reached equilibrium with the overlying
unconfined aquifer as the time rate of drawdown has

Fig. 6. (a) Timedrawdown curve at an observation well 8.6 m


from the pumping well for the conceptual model depicted in Fig.
4; (b) Hantush type-curve match for this timedrawdown plot
showing leakage factor, transmissivity and storage coefficient for
the type-curve match.

asymptotically approached zero. The Hantush type


curve displayed in Fig. 6b is obtained from Eq. (7)
with a best fit b of 0.01. The value of b is small
(representing a curve that closely approaches the
Theis curve) because the extent to which leakage
affects drawdown in pumped aquifers is a function of
the radial distance from the pumping well (Neuman
and Witherspoon, 1972). The Hantush method is used
in favor of the Theis method here because an estimate
of b is desired for the technique used here and also
provides a better estimate of T and S for the aquifer.
The curve match yields a transmissivity and storativity for the aquifer that are nearly identical to the
actual values used in the simulation (Table 1). The b
value can be used in Eq. (9) but this still leaves the
product K 0 S0 that cannot be further reduced with the
traditional time drawdown data plot alone. Neuman
and Witherspoon (1972) indicate that the only way to
evaluate the properties of the semi-confining unit is to
have an observation wells located in this clay unit and
to either measure the head relative to that of the
pumped aquifer, or to evaluate the lag time for the
pressure transient to propagate from the aquifer to the
opening of the piezometer in the confining unit.
Time compaction data collected during the aquifer test can be used to provide accurate estimates of S0
and K 0 : This data set provides important information
that precludes the need for piezometers in the semiconfining unit, or laboratory estimates of compressibility, information that was necessary when only head
data were available during aquifer testing. As GPS
and InSAR techniques become further refined submillimeter resolution will be possible. Currently,
properly designed extensometers can provide near
micron resolution and fixed-station GPS using chokering technology can provide near millimeter vertical
resolution (Blewitt, NBMG, oral commun., 2002).
The magnitude of deformation that can be expected in
these limited duration aquifer tests is typically of the
order of 1 cm.
Fig. 7a is the semi-log time compaction plot for
this aquifer test. The curve is extremely diagnostic for
indicating the source of compaction and the units
responsible for the release of water from storage. The
curve can be divided into three segments. Early-time
data (labeled as segment 1 on the curve) do not fall on
a straight line using semi-log graph paper because the
initial release of water and subsequent compaction

10

T.J. Burbey / Journal of Hydrology 281 (2003) 322

Fig. 7. (a) Timesubsidence curve at an observation well 8.6 m


from the pumping well for the conceptual model depicted in Fig.
4; (b) calculated compaction over one log cycle of time used in
Eq. (6) to estimate the storage and specific storage of the semiconfining unit.

Fig. 8. Histograms showing the distribution of compaction by layer


(a) after 0.3 days of pumping and (b) after 40 days of pumping.
Layer 1 is the overlying unconfined aquifer, layers 211 represent
the confining unit, and layer 12 is the pumped confined aquifer.

occurs from the less compressible pumped aquifer.


Fig. 8a is a histogram of the simulated compaction for
each model layer. Note that during this early time
virtually all the water released from storage occurs
from within the pumped aquifer. Water does not begin

to be released from the semi-confining unit until the


inflection point on the curve is reached at a time in
excess of 0.3 days. From this time until approximately
4 days into the test (segment 2), the subsidence data
fall on a straight line using semi-log paper. During this

T.J. Burbey / Journal of Hydrology 281 (2003) 322

period all the measured compaction originates from


the semi-confining unit as significant quantities of
water are being released from this unit through the
dissipation of excess pore pressure. At the next
inflection point (at approximately 6 days) the pressure
transient has made its way to the overlying unconfined
aquifer as the subsidence asymptotically approaches a
terminal value (segment 3). The time between
inflection points should be close to the time constant
of the semi-confining unit for a singly draining
continuous layer. This time constant derived by
Riley (1969) is expressed as

S0s b2
K 0v

10

and results in the time required for 93% of the pore


water to dissipate from the unit. The time constant
from Eq. (10) is very close to the time between the
two inflection points on time subsidence plot shown
in Fig. 7b. Fig. 8b is the histogram indicating the final
compaction, by model layer, at the end of the aquifer
test. The total compaction from the semi-confining
layer represents 92.5 % of the total subsidence
measured at the land surface. It should be noted that
the distance from the pumped well at which
observations are made does slightly affect the slope
of the time subsidence curve at distances greater than
about 2.5 times the aquifer thickness. At distances
equaling four times the aquifer thickness the decrease
in slope results in about 10 percent less estimated
compaction over one log cycle of time. This would
translate into a smaller value of estimated specific
storage for the confining unit. However, because the
length of the aquifer test results in small quantities of
total compaction, one would in practice want to make
the observations near the pumping well where the
subsidence is greatest, thus alleviating this source of
error.
The next step is similar to the method described by
Cooper and Jacob (1946) and discussed previously for
perfectly confined aquifers where we use a straightline approximation through segment 2, which should
nearly always approximate a straight line on semi-log
paper (Fig. 7b). In other words, the dissipation of pore
pressure from the semi-confining unit and its related
compaction, as expressed through the poroelastic
stress strain constitutive relation, responds like

11

a confined aquifer under a pumping stress. Jorgensen


(1980) presents an expression for compaction within
the semi-confining unit that is logarithmically
proportional to the effective stress at the boundary
of the unit, assuming no significant change in porosity
(void ratio) during compression. Using the straightline approach through segment 2, the compaction
occurring over one log cycle of time is 3.9 mm. The
Hantush method was used to estimate T and S of the
aquifer because the head changes used to estimate
these parameters were obtained from the aquifer with
modification resulting from leakage from the overlying semi-confining unit. It is proposed that Eq. (6)
can be used to calculate the storativity of the semiconfining unit because the compaction, which is used
to estimate storativity, is occurring largely within this
more highly compressible unit (Fig. 8b). Once the
confining-unit storativity is known (and specific
storage if b0 is known) it can be used to solve for the
vertical hydraulic conductivity by rearranging Eq. (9)
K0

16b2 Kbb0 S
:
r 2 S0

11

Table 1 lists the final estimated values and shows how


they compare with the simulated values. Close
agreement occurs for each parameter and only one
piezometer was required. It is assumed here that the
aquifer and confining unit thicknesses are known
through drillers logs or other available data.
It should be noted that it does not matter if the
pressure transient has propagated through the confining unit (segment 3 effectively would be absent). As
long as a straight-line can be drawn through the data
representing compression of the semi-confining unit
(segment 2); then accurate values of specific storage
and hydraulic conductivity can be obtained with this
technique.
2.2.2. Condition 2: multiple semi-confining units
with drawdown in the unpumped aquifer
The second hydrogeologic condition to be evaluated involves the more general case of multiple
aquifers and intervening semi-confining units. The
assumptions presented in the first condition are
applied here except that the overlying unconfined
aquifer has a variable head. That is, drawdown can
occur in this overlying aquifer unit. Fig. 5 represents
the conditions used for the numerical simulation to

12

T.J. Burbey / Journal of Hydrology 281 (2003) 322

obtain data sets for drawdown and subsidence as a


function of time. The values for the semi-confining
units and aquifers are the same as those listed in
Table 1. That is, each confining unit and each aquifer
have identical hydraulic properties. This distinction is
not required, but as we shall see, without this
assumption explicit values of specific storage and
hydraulic conductivity for each semi-confining unit
can not be made without head date for each unpumped
aquifer unit.
Neuman and Witherspoon (1969a c, 1972)
present a complex general solution for the condition
shown in Fig. 5 such that each aquifer depends on five
dimensionless parameters, one of which is Hantushs
leakage factor expressed in Eq. (9). A tractable
solution is obtained by making predictions at early
times or alternatively late times and thus reducing the
number of parameters, or by setting dimensionless
parameters equal and thus effectively making hydraulic properties within individual units equal. To
explicitly calculate an estimate of the specific storage
of the semi-confining units a piezometer must be
installed within these low permeable units at a known
vertical distance above or below the pumped aquifer.
In other words, to characterize the semi-confining
units, hydraulic head data or transient pressure data
within the low permeable units are required. In many
field applications, however, an observation point
within the semi-confining unit is not available and
in practice is avoided.
Fortunately, the methodology described earlier can
be used in this more complex setting without the more
data intensive, and mathematically complex, solution
of Neuman and Witherspoon (1969a c). As before
the key is to plot the semi-log time subsidence
data, which is diagnostic for evaluating (1) when
compaction is largely occurring within the semiconfining units, and (2) when significant drawdown
begins within the unpumped aquifer such that the
drawdown curves within the pumped aquifer would be
affected by the contribution from the overlying
aquifer. Fig. 9 represents the 30-day time subsidence
plot for the aquifer test described by Fig. 5, with
aquifer properties listed in Table 1. The straight-line
drawn through the linear portion of the curve
represents, as before, the compression within the
semi-confining units, whose slope is representative of
the specific storage of the most compressible units.

Fig. 9. Time subsidence curve at an observation well 8.6 m from


the pumping well for the conceptual model depicted in Fig. 5. The
calculated compaction is measured over one log cycle of time and
used in Eqs. (6) and (13) to estimate the average specific storage of
the semi-confining units.

Fig. 10a and b indicate the proportion of compaction


occurring at early time (primarily aquifer) and later
time (primarily aquitards) during the aquifer test for
the multiple aquifer system where the aquifer is
represented by model layer 12. Drawdown within the
unpumped aquifer is not significant based on the slope
of this plot. If drawdown was significant, the slope of
the time subsidence plot would undergo an abrupt
change indicating that the pressure gradient across the
confining unit had become smaller as the transient
pressure wave propagated to the unpumped aquifer.
The simulated drawdown within the unpumped
aquifer at a distance of 8.6 m from the pumped well
was only 0.25 m after 6 days and 1.0 m after 40 days;
only a small fraction of the drawdown in the pumped
aquifer that reached 12 m at this same horizontal
distance at the conclusion of the aquifer test. This
small amount of drawdown apparently is not
significant enough to lower the slope of the
compaction curve, which would occur if the head
gradient across this unit were reduced as a result of
significant contribution from the overlying aquifer.
Fig. 10b indicates that the compaction through

T.J. Burbey / Journal of Hydrology 281 (2003) 322

13

appreciable error. This assessment can be made with


the time subsidence plot and without an observation
well in the overlying aquifer.
The Hantush (1960) method is more tractable than
the Neuman and Witherspoon (1969a c) method and
requires only a single leakage factor expressed by
(
 
 00 00  00 1=2 )
r
K 0 =b0 S0 1=2
K =b S
; 12
b

4
T
S
T
S

Fig. 10. Histograms showing the distribution of compaction by layer


(a) after 0.3 days of pumping and (b) after 40 days of pumping.
Layer 1 is the overlying unconfined aquifer, layers 211 represent
the overlying confining unit, and layer 12 is the pumped confined
aquifer, and layers 1322 represents the lower confining unit.

the upper confining unit is greatly reduced at the top


of the unit, confirming this conclusion (see Eq. (6)).
As a result, it can be concluded that the Hantush
method that assumes no drawdown in the overlying
aquifer can be used with confidence without any

where the double primes refer to the second


underlying semi-confining unit shown in Fig. 5.
As before, the storage coefficient and transmissivity of the aquifer can be evaluated by matching the
drawdown curve of the pumped aquifer with the
Hantush type curve as was done previously. Fig. 11
represents the time drawdown plot for this aquifer
test with the matching Hantush type curve that yields
a dimensionless leakage factor (Eq. 12) of 0.028. The
curve match yields a hydraulic conductivity and
storage coefficient very close to the simulated values
used (Fig. 11 and Table 2).
The information necessary to determine the storage
properties of the semi-confining units is now available. The estimated aquifer transmissivity of
135.7 m2/d from the Hantush type curve and the
semi-log subsidence value (subsidence occurring over
one log cycle of time in Fig. 9) of 0.0084 m can now
be used in Eq. (6) to estimate the storage coefficient of
the semi-confining units. This yields a storage
coefficient of 3.11 1023. This estimated value
represents the average storage of the more highly
compressible semi-confining units. To obtain a
specific storage for each confining unit, a cumulative
thickness of all compressible units undergoing
compression is required. In this case each semiconfining unit was simulated to be 20 m thick. Thus,
dividing the estimated storage by 40 m yields a
specific storage value nearly identical to the simulated
value (Table 2). The general expression for estimating
the specific storage from Eq. (6) for multiple semiconfining units is


1
1
S0save S0ave

13
b1
b2
where the thickness of the bounding semi-confining
units b1 and b2 is required to estimate the average
specific storage.

14

T.J. Burbey / Journal of Hydrology 281 (2003) 322

Fig. 11. Timedrawdown curve at an observation well 8.6 m from the pumping well for the conceptual model depicted in Fig. 5. A Hantush
type-curve match for this time drawdown plot shows the leakage factor, transmissivity and storage coefficient for the type-curve match.

If the semi-confining units all reside above the


pumped aquifer and are separated by more highly
permeable aquifer units, then the storage value
obtained by Eq. (6) is dependent upon both the
thickness of the unit and the head change across each
semi-confining unit. It should be noted, however, that
if confining units above the unpumped aquifer are
compressing, then the head within the unpumped
aquifer is being affected by pumping within the
pumped aquifer. In this case the Hantush method is
not valid and the method described above by Neuman
and Witherspoon (1969a c) would be needed to
obtain a legitimate type curve for the pumped aquifer.
Nonetheless, Eq. (6) is still valid if piezometers are
located within each aquifer so that a head gradient
across each semi-confining unit can be measured, then
a specific storage can be estimated as well as an
estimate of the contribution of compaction from each
semi-confining unit if it is assumed that the compressibilities of each unit are the same.

3. Case study: Franklin Virginia


Groundwater withdrawals from the lower confined
aquifers of the Virginia Coastal Plain have increased
dramatically over the past century, particularly near
the city of Frankin (Fig. 12), where withdrawals
associated with paper milling have resulted in a large
cone of depression. Water levels currently are more
Table 2
Simulated and estimated aquifer and semi-confining unit parameters
used for the second hydrogeologic condition with two confining
units and variable drawdown in the unpumped aquifer
Aquifer and confining
unit parameters

Simulation
value

Estimated value
from new method

T (m2/d)
S (dimensionless)
K 0v (m/d)
S0s (1/m)

125.0
5.0 1025
0.005
8.0 1025

135.7
5.1 1025
0.0038
7.8 1025

T.J. Burbey / Journal of Hydrology 281 (2003) 322

15

Fig. 12. Location of study area in the Virginia Coastal Plain (modified from Pope, 2002).

than 50 m lower than predevelopment conditions.


Subsidence in this region was first identified in the
1970s through a high-precision re-leveling survey
(Holdahl and Morrison, 1974). In an attempt to further
characterize subsidence trends, the US Geological
Survey installed an extensometer in close proximity to
the center of the cone of depression in Franklin.
Sixteen years of compaction data from 1979 to 1995

were collected before the extensometer was removed


from service.
A thorough evaluation of the relationship between
pumping and subsidence in this complex aquifer
system was conducted by Pope (2002), who was able
to accurately match historical leveling and extensometer data through simulation using the one-dimensional compaction model COMPAC developed by

16

T.J. Burbey / Journal of Hydrology 281 (2003) 322

Helm (1975). In addition, Pope evaluated the


contribution of compaction attributed to each aquifer
and confining unit within the Coastal Plain system that
consists of 10 aquifers and intervening confining
units. Fig. 13 shows the lithologic log and the
representative hydrogeologic and modeling units
and their thicknesses at the Franklin site. Results
from Pope (2002) suggest that much of the permanent

(inelastic) compaction occurred during the most


intensive pumping and subsequent water-level
declines between 1940 and 1966. The vast majority
of pumping occurred within the Potomoc group
(Fig. 13), which represents the lowest unit of the
sedimentary wedge that makes up the Coastal Plain
and represents the thickest most prolific aquifers
(Meng and Harsh, 1988; Hamilton and Larson, 1988).

Fig. 13. Lithology and hydrogeologic and modeled units at the Franklin site. Model unit numbers correspond to the unit numbers in Fig. 13b
(from Pope, 2002).

T.J. Burbey / Journal of Hydrology 281 (2003) 322

Additionally, prior to 1980, only the lower and middle


Potomac aquifers were pumped. Fig. 14a and b
represent the measured and reconstructed water-level
and compaction data for the Franklin site (Pope,
2002). The units shown in Fig. 14b correspond to the
units shown in Fig. 13. Pope determined the
cumulative compaction as a function of time
associated with the lower and middle Potomac
aquifers and lower Potomac confining unit. These
totals were used in this investigation.
Aquifer properties for the lower and middle
Potomac aquifers and the lower Potomac confining
unit separating these aquifers have been estimated
from hydraulic testing and numerical modeling. The
aquifer transmissivities reported in the literature range
from 400 to 1500 m2/d (Harsh and Lacnziak, 1990;
Hamilton and Larson, 1988). The vertical hydraulic
conductivity of the confining unit has been estimated
from laboratory analyses to be as low as
5.8 1027 m/d (Harsh and Lacnziak, 1990), while
values estimated from aquifer tests and modeling
range from 2.6 1024 to 1.3 1025 m/d (Harsh and
Lacnziak, 1990).
Storage properties of the Potomac aquifers have
varied widely. Numerical models and aquifer testing
indicate that the estimated storage coefficients for
these aquifers range from 1.6 1026 to 9.3 1023
(Hamilton and Larson, 1988). No storage value for the
confining unit has been estimated.
While the range of estimated transmissivities of
the aquifers differ by less than a factor of four, the
estimated range of storage coefficients differ by
close to four orders of magnitude, indicating the
need for techniques to better quantify the storage
properties of aquifer systems. Pope (2002) used the
subsidence data and the COMPAC model of Helm
(1975) to obtain an inelastic storage coefficient of
1.5 1025 for the lower two aquifers (which
includes the confining unit as part of a bulk
storage value). This estimate is probably more
accurate than previous estimates because it
accounts not only for the head response due to
pumping, but also takes into account the compaction record as well as the role of the storage within
the confining units, which previous numerical
models did not account for.
Fig. 14a and b were used to test the graphical
method presented here. The application of the method

17

developed here does not require an entire record from


the outset of pumping. Although the entire record was
plotted on semi-log paper, only the portion of the
record between 1941 and 1966 was evaluated for
transmissivity and storage. The reason for this is
twofold: Firstly, the pumping rates were more
consistent during this time period. Much smaller
rates were used prior to 1941 much larger rates were
applied shortly after 1966. After 1990 additional
aquifers above the lower Potomac were also pumped.
Secondly, water levels were declining past their past
maximum levels during this period indicating that
inelastic or permanent compaction was occurring
allowing for a more consistent evaluation of the
inelastic storage coefficient of the aquifer during this
time period. Fig. 15a and b represent the semi-log
plots of the time drawdown and time compaction
records, respectively. Although the log time scales
appear to be different, the portion of the slope
approximated with a straight line represents the
same period of record. The time axis values are
different because of the length of record differences as
shown in Fig. 14.
The Cooper Jacob straight-line method was used
to estimate the transmissivity from Fig. 15a assuming
a thickness of 175 m (from Fig. 13) for a confined
non-leaky aquifer system. A non-leaky system is
assumed because the thickness of the aquifer is much
greater than the overlying confining unit thickness and
the bottom of the system is bounded by crystalline
bedrock. The average pumping rate for the portion of
the record of concern is 1.54 105 m3/d. The earlytime data represent the earlier portion of the record
where lower pumping rates were used. In addition, the
data are responding to the fact that the entire record
from the start of pumping was not used. The estimated
transmissivity using Eq. (2) is 430 m2/d, which is
considered to be at the low end of the range of
estimated values. Although this estimate is in the
range of reported values, it is near the low end of
values in the literature. The reason is likely due to the
fact that most reported values are from modeling
estimates that do not take into account the significant
leakage from confining units. Thus, in order to
calibrate the model to measured head values, larger
transmissivities of the aquifer units are required
because of the source of water not accounted for
from these fine-grained units. Consequently,

18

T.J. Burbey / Journal of Hydrology 281 (2003) 322

Fig. 14. Water level and measured and simulated compaction records for the Franklin site. Unit numbers correspond to model unit numbers in
Fig. 12 (from Pope, 2002).

T.J. Burbey / Journal of Hydrology 281 (2003) 322

19

Fig. 15. Semi-log time drawdown and timecompaction plots for the Franklin data. Straight-line approximations are made through the data
from 1941 to 1966.

the model-estimated transmissivities are inflated. The


lower transmissivity estimated in this analysis is
likely closer to the actual value of the aquifer units.
This lower estimate of transmissivity was used to
estimate the storage using Eq. (6). The straight-line
estimate (Fig. 15b) was made for the period of record

as described above. Late time data do not fit because


the aquifer began to respond elastically at these later
times as heads began to equilibrate. The early-time
data do not fit the curve (Fig. 15b) for the same reason
the transmissivity data did not-much lower pumping
rates than the average rate used for the period of

20

T.J. Burbey / Journal of Hydrology 281 (2003) 322

record and because the semi-log data was not initiated


with the start of pumping. The amount of compaction
occurring over one log cycle of time through the
period of record was measured to be 0.173 m. The
estimated inelastic storage from the slope method (Eq.
6) is calculated to be 1.5 1025, which is the
identical value estimated by Pope (2002) by means
of one-dimensional compaction modeling that incorporated the extensometer data, past pumping history,
previous levelings, and water-level data.
As radar techniques progress allowing for real-time
compaction data to be obtained during pumping, the
graphical technique presented here will provide a
quick and accurate approach to obtaining a legitimate
storage value for the aquifer or confining unit of
question. Accurate storage estimates are necessary in
order for water managers to accurately predict the
quantity of potable water supplies available to them.

4. Summary and conclusions


Traditional semi-log straight-line methods for
estimating storage require that the time at s 0 be
evaluated. Not only can this estimate lead to
substantial errors, but it also requires the estimation
of the exact time since the start of pumping. A new
graphical technique based on the Cooper Jacob
method uses a highly diagnostic plot of compaction
or subsidence as a function of time to estimate the
storage coefficient (skeletal storage is nearly equal to
the storage in highly compressible aquifer systems).
This method does not require that one explicitly know
the time since the start of pumping. Time subsidence
data can be obtained from extensometer data, or from
newer radar techniques such as GPS or InSAR
imagery. The vertical deformation signatures provided from these techniques provide valuable information about the aquifer system compressibility, or
more precisely about the most highly compressible
units within the aquifer system during pumping. The
clay-rich layers that typically are one or two orders of
magnitude more compressible than the aquifer
deposits represent the units contributing most of the
subsidence in an aquifer system. Plotted on semi-log
paper, the slope of the time subsidence plot
occurring over one log cycle of time can be used to
accurately calculate the specific storage of the aquifer

system, or specifically of the semi-confining units


within the aquifer system.
Previously developed analytical solutions by
Hantush (1960) and Neuman and Witherspoon
(1969a c) for leaky aquifer systems depend exclusively on hydraulic head data to characterize the
aquifer properties. Without additional head or pressure data in the unpumped units, including the semiconfining units, only limited information can be
drawn from the hydrograph data regarding the semiconfining units. The method presented here does not
require additional sets of head data and the estimated
specific storage and vertical hydraulic conductivity
values obtained with this technique are consistently
close to the actual values of the system during
numerical tests.
In leaky aquifer systems, the time subsidence plot
is highly diagnostic with the semi-log plot representing three distinct periods of deformation. During the
early stages of the aquifer test compaction is almost
entirely occurring within the pumped aquifer and the
slope of the time drawdown curve is very small.
During the second stage of deformation the curve
quickly steepens and forms a straight line representing
the period when compression is occurring almost
exclusively within the more highly compressible
semi-confining units. If the head in the unpumped
aquifer is constant, the third portion of the time
drawdown curve will asymptotically approach a final
subsidence value. However, if the head within the
unpumped aquifer is variable, the third segment of the
curve will change noticeably to a smaller slope
indicating that drawdown within the unpumped
aquifer has occurred and a smaller pressure gradient
is developing across the confining unit.
By measuring the amount of subsidence occurring within one log cycle of time during the
straight-line portion of the time compaction plot,
a highly accurate estimate of specific storage of the
compacting unit can be readily obtained. The
leakage factor for the aquifer and adjacent semiconfining units can then be used to estimate the
vertical hydraulic conductivity of the semi-confining
units. Estimates of vertical hydraulic conductivity
using this technique are consistently close to the
numerical values used. This technique has been
successfully applied to systems with multiple
confining units as well.

T.J. Burbey / Journal of Hydrology 281 (2003) 322

The power of the method is not merely the


accurate values of specific storage and hydraulic
conductivity obtained, but also the minimum
amount of field data that needs to be obtained.
Extensive and expensive piezometer networks that
include observation points within each aquifer and
each semi-confining unit are not required. Only the
time drawdown data of the pumped aquifer and
the time subsidence data obtained from radar
techniques at the land surface are needed to use
this technique.
The method was applied to the Coastal Plain near
Franklin in southeastern Virginia where long-term
pumping associated with the milling of paper has
created an extensive cone of depression. Furthermore,
compaction has been monitored at the site and recent
compaction modeling data have provided a data set
that allows the application of the method presented
herein. Earlier estimates of storage for this aquifer
system varied by nearly four orders of magnitude. The
application of the graphical method described herein
produced a storage value that matched the storage
obtained from extensive numerical modeling using the
hydraulic head and compaction records from the site.

References
Amelung, F., Galloway, D.L., Bell, J.W., Zebker, H.A., Laczniak,
R.J., 1999. Sensing the ups and downs of Las Vegas: InSAR
reveals structural control of land subsidence and aquifer-system
deformation. Geology 27 (6), 483 486.
Anderson, M.P., Woessner, W.W., 1992. Applied Groundwater
Modeling, Academic Press, San Diego, CA, 381 p.
Burbey, T.J., 2001. Stress strain analyses for aquifer-system
characterization. Ground Water 39 (1), 128136.
Cooper, H.H., Jacob, C.E., 1946. A generalized graphical method
for evaluation formation constants and summarizing well-field
history. Transactions of the American Geophysical Union 27,
526534.
Davies, P., Blewitt, G., 2000. Methodology for global geodetic time
series estimation; a new tool for geodynamics. Journal of
Geophysical Research B, Solid Earth and Planets 105 (5),
11083 11100.
Fielding, E.J., Blom, R.G., Goldstein, R.M., 1998. Rapid subsidence
over oil fields mearured by SAR interferometry. Geophysical
Research Letters 25 (17), 32153218.
Galloway, D.L., Hednut, K.W., Ingebritsen, S.E., Phillips, S.P.,
Peltzer, G., Rogez, F., Rosen, P.A., 1998. InSAR detection of
system compaction and land subsidence, Antelope Valley,

21

Mohave Desert, California. Water Resources Research 34,


25732585.
Hamilton, P.A., Larson, J.D., 1988. Hydrogeology and analysis of
the ground-water flow system in the Coastal Plain of Southeastern Virginia. US Geological Survey Water-Resources
Investigation Report 87-4240, 175.
Hantush, M.S., 1956. Analysis of data from pumping tests in leaky
aquifers. Transactions of the American Geophysical Union 37,
702 714.
Hantush, M.S., 1959. Analysis of data from pumping wells near a
river. Journal of Geophysical Research 64 (11), 19211932.
Hantush, M.S., 1960. Modification of the theory of leaky aquifers.
Journal of Geophysical Research 65 (11), 37133725.
Hantush, M.S., 1967. Flow to wells in aquifers separated by a
semipervious layer. Journal of Geophysical Research 72 (6),
17091720.
Harmon, E.J., 2002. Extensometer data as an alternative means
of determining aquifer specific storage, San Luis Valley,
Colorado. Geological Society of America, Annual Meeting,
Denver, CO, October 26 31 34 (6) Abstract with programs,
paper 99-10.
Harsh, J.F., Lacnziak, R.J., 1990. Conceptualization and analysis of
the ground-water flow system in the Coastal Plain of Virginia
and adjacent parts of Maryland and North Carolina. US
Geological Survey Professional Paper 1404-F, 100.
Helm, D.C., 1975. One-dimensional simulation of aquifer-system
compaction near Pixley, California, 1. Constant parameters.
Water Resources Research 11 (3), 465478.
Hoffmann, J., Zebker, H.A., Galloway, D.L., Amelung, F., 2001.
Seasonal subsidence and rebound in Las Vegas Valley, Nevada,
observed by synthetic aperture radar interferometry. Water
Resources Research 37, 15511566.
Holdahl, S.R., Morrison, N.L., 1974. Regional investigations of
vertical crustal movements in the US, using precise relevelings
and mareograph data. Tectonophysics 23, 373 390.
Ikehara, M.E., 1994. Global positioning system surveying to
monitor land subsidence in Sacramento Valley, California,
USA. Hydrological Sciences 39 (5), 417429.
Jacob, C.E., 1940. On the flow of water in elastic artesian aquifers.
Transactions of the American Geophysical Union 21, 574586.
Jorgensen, D.G., 1980. Relationships between basic soils-engineering equations and basic ground-water flow equations. US
Geological Survey Water Supply Paper 2064, 40.
Leake, S.A., Prudic, D.E., 1991. Documentation of a computer
program to simulate aquifer-system compaction using the
modular finite-difference ground-water flow model. Tech of
Water Resources Invest.US Geol Surv, Book 6, A2.
McDonald, M.G., Harbaugh, A.W., 1988. A modular threedimensional finite-difference ground-water flow model. Techniques of Water Resources Investigation US Geological Survey,
Book 6, Ch A1.
Meinzer, O.E., 1928. Compressibility and elasticity of artesian
aquifers. Economic Geology 23, 263 291.
Meng, A.A., Harsh, J.F., 1988. Hydrogeologic framework of the
Virginia Coastal Plain. US Geological Survey Professional
Paper 1404-C, 82.

22

T.J. Burbey / Journal of Hydrology 281 (2003) 322

Moench, A.F., 1985. Transient flow to a large-diameter well in a


confined aquifer with storativity in semiconfining layers. Water
Resources Research 21, 11211131.
Neuman, S.P., 1979. Perspective on delayed yield. Water
Resources Research 15, 899908.
Neuman, S.P., Witherspoon, P.A., 1968. Theory of flow in aquicludes
adjacent to slightly leaky aquifers. Water Resources Research 4,
103112.
Neuman, S.P., Witherspoon, P.A., 1969. Transient flow of ground
water to wells in multiple aquifer systems. Geotechnical Engineering Report 69-1, University of California Berkeley, CA, 182 p.
Neuman, S.P., Witherspoon, P.A., 1969b. Theory of flow in a confined
two aquifer system. Water Resources Research 5, 803816.
Neuman, S.P., Witherspoon, P.A., 1969c. Applicability of current
theories of flow in leaky aquifers. Water Resources Research 5,
817829.
Neuman, S.P., Witherspoon, P.A., 1972. Field determination of the
hydraulic properties of leaky multiple aquifer systems. Water
Resources Research 8, 12841298.
Poland, J.F., Davis, G.H., 1969, Land subsidence due to
withdrawal of fluid. In: D.J. Varnes, G. Kiersch (Eds.),
Reviews in Engineering Geology, v.2, Boulder, Colorado,
Geological Society of America pp.187-269
Pope, J.P., 2002, Characterization and modeling of land
subsidence due to groundwater withdrawals from the confined
aquifers of the Virginia Coastal Plain. Unpublished MS Thesis,
Virginia Tech, 149 p.

Pope, J.P., Burbey, T.J., 2003 (in press). Multi-aquifer characterization from single extensometer records Ground Water.
Riley, F.S., 1969. Analysis of borehole extensometer data from
central California. International Association of Scientific Hydrology Publication 89, 423431.
Sridharan, K., Ramaswamy, R., Lakshmana Rao, N.S., 1987.
Identification of parameters in semiconfined aquifers. Journal of
Hydrology 93, 163173.
Theis, C.V., 1935. The relation between the lowering of the
piezometric surface and the rate and duration of discharge of a
well using ground-water storage. Transactions of the American
Geophysical Union 16, 519 524.
Thom, C., Bock, O., Kasser, M., Fourmaintraux, D., 1995.
Development of a new airborne laser subsidence measurement
system, aiming at mm-accuracy. In: F.B.J., Barends, F.J.J.,
Brouwer, F.H., Schroder (Eds.), Land Subsidence, Balkema,
Rotterdam, pp. 113 122.
Walton, W.C., 1970. Groundwater Resources Evaluation, McGrawHill, New York, 664 p.
Zebker, H.A., Rosen, P.A., Goldstein, R.M., Gabriel, A., Werner,
C.L., 1994. On the derivation of coseismic displacement fields
using differential radar interferometry: the Landers earthquake.
Journal of Geophysical Research 99 (B10), 1961719634.
Zebker, H.A., Rosen, P.A., Hensley, S., 1997. Atmospheric effects in
interferometric synthetic aperture radar surface deformation and
topographic maps. Journal of Geophysical Research 12 (B4),
75477563.

You might also like